JP7745911B2 - Computer program, game system using same, and control method - Google Patents

Description

The present invention relates to a computer program that is applied to a computer incorporated in a game system that provides element information, which is information relating to at least one of a group of elements for each user, in a game played by multiple users using each group of elements for each user.

There are game systems that provide element information, which is information about at least one of the element groups for each user, in a game played by multiple users using each user's own element group. For example, a system is known that uses information about a real-world shogi game, each move in that shogi game, and the "win rate" for each move as the game, element group, and element information, respectively (see, for example, Non-Patent Document 1).

Ryosuke Mine, "The win rate display is interesting! How to enjoy the 'ABEMA Shogi Program' where AI analyzes win rates," [online], December 26, 2020, Shogakukan, [Retrieved January 12, 2024], Internet, <URL: https://dime.jp/genre/1032083/>

Non-Patent Document 1 discloses technology that adds information such as "winning rate," "potential moves," and "predicted moves" simulated by AI (artificial intelligence) for each move to a live screen that captures a shot of the shogi board from above during a game. Games like shogi are known as complete information games, in which all information about the pieces used by each player is disclosed. On the other hand, there are also games known as incomplete information games. In these types of games, the content of at least one element of a group of elements (e.g., options) used by other users (including computers) is not shared with a single user. For this reason, in incomplete information games, information about elements that are not shared (concealed) with users can be important.

The present invention therefore aims to provide a computer program or the like that can help infer hidden elements in games with incomplete information.

The computer program of the present invention is configured to cause a computer incorporated in a game system that provides element information relating to at least one of a group of elements for each user in a game played by multiple users using each group of elements. The computer is configured to cause the computer to function as: information acquisition means for acquiring inferred information using a predetermined process set up to obtain inferred information for inferring the content of at least one element in an incomplete information game provided as the game, in which the content of at least one element of other users is not shared with a single user; and information provision means for providing the inferred information as the element information.

On the other hand, the game system of the present invention is a game system incorporating a computer that provides element information as information relating to at least one of the element groups for each user in a game played by multiple users using the element groups for each user, and the computer functions as an information acquisition means that acquires the inferred information using a predetermined process set up to obtain inferred information that infers the content of at least one element in the element groups of other users provided as the game, in an incomplete information game in which the content of at least one element of the element groups of the other users is not shared with a single user, and an information provision means that provides the inferred information as the element information.

The control method of the present invention also involves causing a computer incorporated in a game system that provides element information relating to at least one of a group of elements for each user in a game played by multiple users using the group of elements for each user to execute the following steps in an imperfect information game in which the content of at least one of the group of elements of other users provided as the game is not shared with a single user: acquiring the speculated information using a predetermined process configured to obtain speculated information for inferring the content of the at least one element; and providing the speculated information as the element information.

1 is a diagram showing a schematic configuration of a network system to which a game system according to an embodiment of the present invention is applied; FIG. 2 is a functional block diagram showing the main parts of a control system of the network system. FIG. 10 is a diagram schematically illustrating an example of a battle screen for playing a card game. FIG. 10 is a diagram showing an example of a procedure for playing a card game. FIG. 10 is a diagram schematically illustrating an example of a battle screen when an analysis service is applied. FIG. 10 is an explanatory diagram illustrating a list of options (actions) that a player can take in a certain situation on the battle screen. FIG. 10 is an explanatory diagram illustrating an example of a method in which an artificial intelligence model calculates a probability value. FIG. 10 is a diagram schematically illustrating an example of a battle screen when a guessing service is applied. FIG. 10 is an explanatory diagram illustrating an example of a method in which an artificial intelligence model calculates inferred information. FIG. 10 is an explanatory diagram for explaining an example of the types of inferred information output by a plurality of types of artificial intelligence models. FIG. 10 is an explanatory diagram for explaining a learning dataset for realizing estimated information for each period and each rank. 10 is a flowchart showing an example of a procedure for option analysis processing. 10 is a flowchart showing an example of a procedure for a display change process. 10 is a flowchart showing an example of a procedure for a back-side inference process. FIG. 10 is an explanatory diagram illustrating an example of a modified example regarding the display mode of superiority/inferiority information.

(Overall structure)
A control method according to one embodiment of the present invention, and a game system in which a computer program is implemented (a game system according to one embodiment of the present invention) will be described below with reference to the drawings. First, the overall configuration of a network system to which a game system according to one embodiment of the present invention is applied will be described with reference to Figure 1. As shown in Figure 1, the network system 1 is configured as a client-server system including a plurality of user devices 3 as clients and a game server 2 connected to each user device 3 via a network NT.

The user device 3 is a device used daily by a user, and is a computer device (information and communication terminal device) equipped with information and communication functions via the network NT. The user device 3 is equipped with a computer program according to one embodiment of the present invention, and functions as a game system according to one embodiment of the present invention in the network system 1. As an example, a smartphone or tablet terminal equipped with communication and call functions may be used as the user device 3. The user device 3 may be a PC (short for personal computer), or may be a personal or home game machine provided as a so-called consumer game machine. Furthermore, an arcade game machine may also be used as the user device 3.

By implementing specific software (applications), the user device 3 functions as a game machine and provides games. The user device 3 provides games that include selection opportunities in which the player must select one of multiple options. Such games may be configured as any suitable game, such as a role-playing game, simulation game, or action game. As an example, the game is configured as a competitive video game in which the user of the user device 3 competes against opponents (other users, including computers) according to specific rules via a game screen containing multiple objects. In competitive games, the user and opponents may function as members of a team, and the competitive game may be played in a one-on-many format (including cases in which multiple users other than the user are independent opponents, such as in mahjong) or many-on-many format; however, the following describes a one-on-one format as an example.

Furthermore, competitive games may be provided as perfect information games such as shogi or chess (both of which use pieces as objects), but as an example, they are provided as imperfect information games. In contrast to perfect information games in which all of the various elements necessary for selection in a selection opportunity are shared with the user, imperfect information games are games in which the contents of at least one element of the group of elements used by other users (including computers) (such as the contents of cards in a card game) are not shared with the user. Imperfect information games include various games such as mahjong, but the following explanation will focus on the case in which a card game is provided as an example.

The game server 2 may be configured by appropriately combining multiple server units (server devices), or may be configured by a single server unit. The game server 2 may also be configured as a cloud server using cloud computing technology. The game server 2 provides various game-related services to the user devices 3. These services include an analysis service and a guessing service. Both the analysis service and the guessing service are services to assist with selections during selection opportunities in the game. Specifically, the analysis service is configured to provide the user with analysis results regarding the relative merits and demerits of the influence of each option included in the selection opportunity for at least one option before selection. On the other hand, the guessing service is configured to provide guessed information regarding information that is not shared with the user in the card game. Details of the analysis service and guessing service will be described below.

In addition, the game server 2 may also provide other services to the user device 3, such as a distribution service that distributes computer programs and various data necessary for playing the game on the user device 3, a matching service that matches users who cooperate or compete in the game, and a relay service that relays game information that should be shared between user devices 3.

The network NT may be configured as appropriate, as long as it is capable of connecting the user device 3 to the game server 2. As an example, the network NT is configured to realize network communication using the TCP/IP protocol. Typically, the network NT is configured by combining the Internet as a WAN and an intranet as a LAN. In the example of Figure 1, the game server 2 is connected to the network NT via a router NTr, and the user device 3 is connected to the network NT via an access point PP.

(Network system control system)
Next, the main parts of the control system of the network system 1 will be described with reference to Figure 2. First, the game server 2 is provided with a control unit 21 and a storage unit 22 as storage means. The control unit 21 is configured as a computer that combines a processor unit that executes various arithmetic processes and operational control according to a predetermined computer program with internal memory and other peripheral devices required for the operation. The processor unit may include units such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and an NPU (Neural Network Processing Unit) as appropriate (including cases where each processor unit is integrated as appropriate, such as when a GPU is incorporated into a CPU).

The memory unit 22 is an external storage device realized by a storage unit including a non-volatile storage medium (computer-readable storage medium) such as a hard disk array. The memory unit 22 may be configured to store all data on a single storage unit, or may be configured to store data in a distributed manner across multiple storage units. The memory unit 22 stores a server program PG1 and server data SD. The server program PG1 is a computer program that causes the control unit 21 to execute the processes required to provide various services to the user device 3. The server program PG1 may include various programs appropriate to the processes to be executed by the control unit 21, but in the example of Figure 2, it includes an analysis program AP and an inference program FP.

The analysis program AP is a computer program that causes the control unit 21 to execute various processes to realize the analysis service. The analysis program AP may be configured as appropriate, for example, to cause the control unit 21 to execute a process that outputs analysis results based on predetermined logic (including predetermined rules such as rules), but in the example of Figure 2, it is configured to cause the control unit 21 (e.g., a GPU) to function as an artificial intelligence model that outputs analysis results. The analysis program AP may be configured as appropriate, but in the example of Figure 2, it is configured as an inference program that incorporates analytical trained data (parameters) AD that are output by having a training program (for analysis) learn a training dataset (for analysis).

The analytical trained data AD may be of a single type, but the example of Figure 2 includes multiple types of trained data AD. The multiple types of analytical trained data AD may be generated by applying the same training dataset to training programs with different algorithms (learning methods), but as an example, they are generated by applying different training datasets to the same training program. The analysis program AP incorporates and executes each of the multiple types of analytical trained data AD, causing the control unit 21 to function as an artificial intelligence model with different ways of thinking. In the example of Figure 2, the analysis programs AP that incorporate each of the multiple types of analytical trained data AD are distinguished as a first analysis program AP1, a second analysis program AP2, and so on. For this reason, the analysis program AP is provided with multiple analysis programs AP, such as the first analysis program AP1 and the second analysis program AP2.

The inference program FP is a computer program that causes the control unit 21 to execute various processes to realize the inference service. The inference program FP may be configured as appropriate, for example, to cause the control unit 21 to execute a process that outputs an inference result based on predetermined logic, but in the example of Figure 2, it is configured to cause the control unit 21 (e.g., a GPU) to function as an artificial intelligence model, similar to the analysis program AP. The inference program FP may also be configured as appropriate, but in the example of Figure 2, it is configured as an inference program that incorporates inference-use learned data (parameters) FD that is output by having a learning program (for inference) learn a learning dataset (for inference).

The trained data for inference FD may be of a single type, but in the example of Figure 2, multiple types of trained data for inference FD (which do not have to match the number of trained data for analysis AD) are included, similar to the trained data for analysis AD. Multiple types of trained data for inference FD may be generated by applying the same training dataset to training programs with different algorithms (learning methods), but as an example, they are generated by applying different training datasets to the same training program. The inference program FP incorporates and executes each of the multiple types of trained data for inference FD, causing the control unit 21 to function as an artificial intelligence model with different ways of thinking. In the example of Figure 2, the inference programs FP, which each incorporate multiple types of trained data for inference FD, are distinguished as a first inference program FP1, a second inference program FP2, and so on. For this reason, multiple inference programs FP, such as the first inference program FP1 and the second inference program FP2, are provided.

Server data SD is data referenced by the server program PG1 to provide various services. The server data may include appropriate data related to the various services, and the example in Figure 2 shows play data PD, analytical learned data AD, and inference learned data FD as examples. Play data PD contains information about each user's past playing history. Play data PD is not limited to playing history, and may also include other information necessary for managing each user, such as various personal information including attributes such as each user's gender or address. Each user's possessions in the card game may be managed as appropriate, but are managed in play data PD as an example. Specifically, the card game is configured to grant card objects to users who meet certain granting conditions (such as purchase, loan, lottery, or grant resulting from game progress). The card objects granted to each user (hereinafter sometimes referred to as "owned cards") are then managed in the play data PD so that they correspond to each user.

Various logical devices can be provided in the control unit 21 by combining the control unit 21's hardware resources with the server program PG1 as a software resource, but in the example of Figure 2, an analytical model unit 23 and an inference model unit 24 are provided. Both the analytical model unit 23 and the inference model unit 24 are logical devices that function as artificial intelligence models. Specifically, the analytical model unit 23 functions as an artificial intelligence model that outputs analysis results based on the analysis program AP, and the inference model unit 24 functions as an artificial intelligence model that outputs inference results based on the inference program FP.

The analytical model unit 23 (artificial intelligence model) is realized, for example, by a combination of analytical trained data AD and analytical program AP (analysis program AP executed incorporating analytical trained data AD), and executes various processes to provide analytical services. This type of artificial intelligence model can form different thoughts depending on various factors. For example, it tends to have different thoughts depending on at least one difference in the content of the training dataset, the learning program algorithm, and the inference program algorithm. The analytical model unit 23 can be provided with an appropriate type of analytical model unit 23 depending on these differences. The example in Figure 2 shows two types of analytical model unit 23: a first analytical model unit 23A and a second analytical model unit 23B.

The first analytical model unit 23A is the analytical model unit 23 corresponding to the first analytical program AP1. Similarly, the second analytical model unit 23B is the analytical model unit 23 corresponding to the second analytical program AP2. The first analytical model unit 23A and the second analytical model unit 23B execute various processes in accordance with the first analytical program AP1 or the second analytical program AP2. For example, this process includes option analysis processing. The procedure for option analysis processing will be described in detail below.

Similarly, the inference model unit 24 (artificial intelligence model) is realized, for example, by a combination of learned data for inference FD and an inference program FP (the inference program FP is executed by incorporating learned data for inference FD), and performs various processes to realize the inference service. The inference model unit 24 may also be provided with an appropriate type (including one type) of inference model unit 24. In the example of Figure 2, two types of inference model unit 24 are shown: a first inference model unit 24A and a second inference model unit 24B.

The first estimation model unit 24A is an estimation model unit 24 corresponding to the first estimation program FP1. Similarly, the second estimation model unit 24B is an estimation model unit 24 corresponding to the second estimation program FP2. The first estimation model unit 24A and the second estimation model unit 24B execute various processes in accordance with the first estimation program FP1 or the second estimation program FP2. For example, this processing includes back-side estimation processing. The procedure for back-side estimation processing will be described in detail below.

In addition, the control unit 21 may also be provided with a logical device such as a web service management unit for implementing processing related to various services, such as the distribution service, matching service, or relay service mentioned above. Similarly, input devices such as a keyboard and output devices such as a monitor may be connected to the control unit 21 as needed. However, these are not shown in the illustration.

On the other hand, the user device 3 is provided with a control unit 31 and a storage unit 32 as storage means. The control unit 31 is configured as a computer that combines a processor unit that executes various arithmetic processes and operational control in accordance with a predetermined computer program, with internal memory and other peripheral devices required for that operation. The processor unit may, for example, appropriately include units such as a CPU, GPU, and NPU, similar to the game server 2 (including cases where each processor unit is appropriately integrated, such as when a GPU is incorporated into the CPU).

The memory unit 32 is an external storage device realized by a storage unit including a non-volatile storage medium (computer-readable storage medium) such as a hard disk or semiconductor storage device. The memory unit 32 stores the game program PG2 and game data GD. The game program PG2 is a computer program that causes the control unit 21 to execute the processing required to make the user device 3 function as a game device. The game data GD is data referenced by the game program PG2 to provide the game. The user device 3 may be provided with an analysis program AP or an inference program FP from the game server 2 as appropriate, in which case the game data GD may include the analysis program AP or the inference program FP.

The game data GD may include various data (including various tables) necessary for playing the game, such as image data for displaying various images for the game, audio data for playing various sounds (including background music such as music), and card data defining each card object, as appropriate, but the example in Figure 2 shows play data PD. The play data PD is provided and saved by the game server 2 as needed. Furthermore, if the analysis program AP or the estimation program FP is provided to the user device 3 by the game server 2, the game data GD may include analytical learned data AD and estimation learned data FD.

The control unit 31 is provided with a progress control unit 33, a data management unit 34, and a model processing unit 35 as logical devices realized by combining the control unit 31's hardware resources and the game program PG2 as software resources.

The progress control unit 33 executes various processes necessary for the progress of the card game. This processing also includes processes necessary for enjoying the services provided by the game server 2. For example, the progress control unit 33 executes processes related to opponent matching in cooperation with the web service management unit, and executes processes such as reflecting the opponent's play in one's own progress during a match, and vice versa, reflecting one's own play in the opponent's. The progress control unit 33 also executes processes to provide selection opportunities in each turn and phase, as described below, and to switch between turns and phases.

The data management unit 34 performs various processes related to the management of the game data GD. For example, when a card object is granted to a user, the data management unit 34 performs a process to reflect the card object as an owned card in the play data PD.

The model processing unit 35 executes various processes necessary to provide analysis services and inference services to users of the user devices 3. The processes executed by the model processing unit 35 include processes realized in cooperation with the analytical model unit 23 or inference model unit 24 of the game server 2. For example, the model processing unit 35 executes processes in cooperation with the analytical model unit 23 to analyze the impact that at least one of multiple options selectable in a selection opportunity has on the progress of the card game, and to provide the analysis results. Similarly, the model processing unit 35 executes processes in cooperation with the inference model unit 24 to provide inference information regarding information that is not shared with users in the card game.

For example, the model processing unit 35 executes option analysis processing in cooperation with the analysis model unit 23 of the game server 2 as processing for realizing the analysis service. The model processing unit 35 also executes display change processing. Similarly, the model processing unit 35 executes behind-the-scenes inference processing in cooperation with the inference model unit 24 of the game server 2 as processing for realizing the inference service. The details of the procedure for the display change processing will be described later.

The model processing unit 35 also generates a situation log SL required for analysis by the analytical model unit 23 or for inference by the inference model unit 24, stores this in the control unit 31's internal storage device, and performs processing to appropriately update the situation log SL as the situation changes. The situation log SL may contain various information related to the game situation as appropriate. The situation log SL (data) contains various information (which may be limited to an appropriate portion of the information) used to determine the status of the card game, such as information on all card objects currently in use, information on each card object placed in the various placement locations described below, attributes of each card object (including the status of each card object, such as whether it has been summoned or left the field area, as described below), life points, phases, and information on options selectable in each situation.

The user device 3 is provided with appropriate output and input devices, and the example in Figure 2 shows a monitor MO, speaker SK, and touch sensor TS as examples of output devices. All of these are general-purpose hardware that is provided in information and communication terminals such as smartphones. For example, the touch sensor TS is an input device that inputs signals corresponding to the user's touch operation (operation with a finger) to the control unit 31. The speaker SK is an output device that plays various sounds in response to signals from the control unit 31. The monitor MO is an output device (display device) that presents a game screen, etc. in response to signals from the control unit 31. The user device 3 may also be provided with various other devices as appropriate, such as a gyro sensor, an acceleration sensor, a location information (e.g., GPS information) receiving device, etc.

(Game Overview)
Next, an overview of the card game will be described with reference to Figures 3 to 11. The card game may be configured as any suitable game, but as an example, it is configured as a video game played via a game screen including a plurality of card objects. More specifically, the card game is configured to be played using a deck card, which is a pile of card objects (a group of a predetermined number of card objects) selected by the user for play from a group of owned cards.

Figure 3 is a diagram schematically illustrating an example of a battle screen for playing a card game. The battle screen 50 is a game screen displayed when a user and an opponent battle using each other's deck cards. Various card objects may be arranged (displayed) on the battle screen 50, but in the example of Figure 3, for ease of explanation, card objects CO facing face-up (the orientation in which the contents of the card object are visible) are shown with a dotted pattern, regardless of type or content, and card objects facing back (the orientation in which the contents of the card object are not visible and are concealed) are shown with a right-slashed pattern. In this example, a predetermined number of card objects CO (which may be constant or variable, and may not match the number of deck cards in the opponent's deck) included in the user's deck card function as the multiple objects and card objects of the present invention. Furthermore, the owned card group and deck cards function as the denominator element group and two or more elements of the present invention, respectively.

As shown in Figure 3, the battle screen 50 includes a user area 51, a common area 52, and an opponent area 53. The user area 51 is an area exclusively for the user. The user area 51 can be configured as appropriate, but in the example of Figure 3, it includes a field area 54, a main deck placement area 55, an extra deck placement area 56, a hand placement area 57, a graveyard area 58, and a point display area LR.

The field area 54 is an area where multiple card spaces CP (only some of the symbols are shown) are formed. Each card space CP is a location (area) where a card object CO from the deck of cards is placed to progress the game. A card game can have any number of card objects CO with various roles (purposes) and effects. Examples of card objects CO available include card objects CO representing monsters (hereinafter sometimes referred to as "monster cards CO1"), card objects CO representing magic (predetermined effects) (hereinafter sometimes referred to as "magic cards CO2"), card objects CO representing traps (effects different from magic) (hereinafter sometimes referred to as "trap cards CO3"), and card objects CO representing special effects (hereinafter sometimes referred to as "special cards CO4"). When a card object CO is placed in a card space CP, that card object CO actually performs its role in the match.

For example, a monster card CO1 is assigned a role such as attacking or defending against the opponent or the opponent's monster card CO1. Monster cards CO1 can be classified into various types, including two types: a normal monster card CO1 and an extra monster card CO1. A normal monster card CO1 is a monster card CO1 that can be placed from the hand placement area 57 to the field area 54. An extra monster card CO1 is a monster card CO1 that is summoned (called) in exchange for a monster card CO1 leaving the field area 54. The placement (summoning) of an extra monster card CO1 is normally restricted to the common area 52, but placement in the field area 54 may be permitted if certain field conditions are met. For this reason, a monster card CO1 may have the role of summoning (calling) other monster cards CO1 to the field area 54, etc. Monster cards CO1 placed in the card placement area CP are used to actually assign roles such as attack, defense, or summoning to the game's progress. Similarly, magic cards CO2, trap cards CO3, and special cards CO4 are card objects CO that can be placed from the hand placement area 57 to the field area 54, and magic cards CO2, trap cards CO3, and special cards CO4 placed in the card area CP are used to activate the effects indicated by each card object CO.

Multiple card placement areas CP may be arranged appropriately in the field area 54, but in the example of Figure 3, they are arranged to form two rows: a front row 54A located closer to the opponent's area 53, and a back row 54B located behind it. The front row 54A and back row 54B are formed by six card placement areas CP and five card placement areas CP, respectively, arranged side by side. Any card object CO may be placed in each card placement area CP without any restrictions, but as an example, restrictions are placed on the types of card objects CO that can be placed. The front row 54A and back row 54B may be used appropriately, but as an example, they are used according to the types of card objects CO that can be placed.

Specifically, the front row 54A is used as card spaces CP where monster cards CO1, one of the card object CO types, should be placed. However, of the six card spaces CP that make up the front row 54A, the leftmost card space CP (which can be formed as desired, but in the example of Figure 3, it is formed slightly smaller than the other card spaces CP) is the only exception, and is used as a card space CP where special cards should be placed. In other words, the placement of card objects CO such as magic cards CO2 or trap cards CO3 in the front row 54A is restricted, the placement of special cards CO4 in any card space CP other than the leftmost card space CP in the front row 54A is restricted, and the placement of monster cards CO1 is further restricted in the leftmost card space CP.

On the other hand, the back row 54B is used as card spaces CP where magic cards CO2 and trap cards CO3, two types of card objects CO, should be placed. In other words, the placement of card objects CO such as monster cards CO1 or special cards CO4 in the back row 54B is restricted. Note that monster cards CO1 with the properties of magic cards CO2 (monster cards CO1 corresponding to monsters with magical effects) may also be prepared, and such monster cards CO1 with magical attributes may be allowed to be placed in specified card spaces CP in the back row 54B (for example, the card spaces CP at the left and right ends). In other words, card objects CO with multiple attributes may be included, and the placement of such card objects CO may be allowed in both the front row 54A and the back row 54B depending on the multiple attributes.

Card objects CO can be placed appropriately in the field area 54 according to the user's instructions, subject to the restrictions of each card placement area CP. In the example of Figure 3, one monster card CO1 is placed in the third card placement area CP from the right in the front row 54A. Furthermore, each card object CO may be placed in a variety of directions, such as vertically or horizontally, in the card placement area CP. The placement direction, such as vertically, can be used appropriately, but as an example, a monster card CO1 is placed horizontally when used for defense and vertically when used for attack.

The hand placement area 57 is an area where card objects CO (hereinafter sometimes referred to as hand CO5) from the deck of cards that are virtually in hand should be placed (displayed). Hand CO5s are candidate card objects CO to be placed in the field area 54 (each card placement area CP). Any number of hand CO5s can be placed in the hand placement area 57, but in the example of Figure 3, five hand CO5s are placed. In this example, the hand placement area 57 functions as the hand area of the present invention.

The main deck placement area 55 and the extra deck placement area 56 are both areas where card objects CO representing the remaining deck cards (the original deck cards minus the card objects CO in the field area 54 and hand placement area 57) should be placed, but they have different uses. Specifically, the main deck placement area 55 is an area where card objects CO (hereinafter sometimes referred to as the deck CO6) that are candidates for addition to the hand CO5 should be placed from among the remaining deck cards. The card objects CO in the deck CO6 are added to the hand CO5 through a draw in the draw phase, described below. On the other hand, the extra deck placement area 56 is an area where a group of card objects (hereinafter sometimes referred to as the summon deck cards CO7) representing a stack of extra monster cards CO1 should be placed. When an extra monster card CO1 is summoned via a normal monster card CO1, the extra monster card CO1 to be placed in the field area 54 is drawn from the summon deck cards CO7. In other words, the remaining deck cards are divided into stock cards CO6 and summon deck cards CO7, and placed in the main deck placement area 55 and extra deck placement area 56, respectively. In this example, the main deck placement area 55 functions as the stock card area of the present invention.

In principle, the graveyard area 58 is an area for storing card objects CO that have left the field area 54. For example, a monster card CO1 will leave the field area 54 when a certain condition (a condition for leaving the field area 54, etc.) is met, such as an attack by the opponent's monster card CO1 or the summoning of an extra monster card CO1. Similarly, card objects CO that exert a certain effect, such as a magic card CO2 or a trap card CO3, will leave the field area 54 after activating their effect in accordance with the card's effect. The graveyard area 58 is provided as a place (destination) for storing (placing) card objects CO that have left the field area 54. The graveyard area 58 may be configured to display card objects CO that have left the field area 54, but in the example of Figure 3, the display of card objects CO is omitted.

The point display area LR is an area for displaying the user's life points. In a card game, the user and opponent can each set an appropriate life point, but in the example of Figure 3, both life points are set to 8000 points ("8000"). The victory or defeat condition for determining victory or defeat in a battle can be set as appropriate, but is met, for example, when the opponent's life points are reduced to zero. In other words, life points function as a parameter for determining victory or defeat. Specifically, the user can reduce the opponent's life points by using monster cards CO1 placed in the field area 54 and common area 52 in attacks. If the opponent's life points are reduced to zero, the user is deemed to have won, and the battle ends. Conversely, if the user's life points are reduced to zero by an attack from the opponent's monster card CO1, the user is deemed to have lost, and the battle ends.

The common area 52 is an area shared by the user and the opponent. The common area 52 can be configured as appropriate, but in the example of Figure 3, two card placement areas CP are provided. The user's card objects CO and the opponent's card objects CO are placed in each card placement area CP as appropriate. Any type of card object CO may be placed in each card placement area CP in the common area 52, but as an example, it is limited to extra monster cards CO1. In other words, an extra monster card CO1 summoned from a summon deck card CO7 is first placed in each card placement area CP in the common area 52, and is only allowed to be placed in the field area 54 if certain field conditions are met. In this example, the user area 51 and each card placement area CP in the common area 52 (which may include the extra deck placement area 56) function as the multiple card placement areas of the present invention.

The opponent area 53 is an area reserved for the opponent. The opponent area 53 plays the same role as the user area 51 for the opponent. For this reason, the opponent area 53 also has a field area 54, a main deck placement area 55, an extra deck placement area 56, a hand placement area 57, a graveyard area 58, and a point display area LR, but because their roles are the same as those of the user area 51, their explanation will be omitted.

Figure 4 is a diagram showing an example of the procedure for playing the card game in the example of Figure 3. As shown in Figure 4, the card game includes a user turn and an opponent turn (hereinafter, sometimes referred to as the "player" when there is no distinction between the two). The game proceeds in a so-called turn-based format, with the user and opponent taking turns alternately throughout these turns. Specifically, as preparation for the game, for example, the players' deck cards are shuffled and placed in the main deck placement area 55, and a predetermined number of card objects CO are drawn from each player's deck cards (the deck CO6) and displayed (placed) as their hand CO5 in the hand placement area 57. Once preparation is complete, the game begins with the turn of the player going first (for example, the user going first). One turn is divided into multiple phases. Phases are a concept for dividing the procedures to be performed in one turn into multiple stages depending on their content and nature. In the example of Figure 4, one turn is divided into six stages, from the draw phase to the end phase, but this is merely an example.

In each phase, the player given a turn can select an appropriate action within the range set for that phase. An example is as follows: In the draw phase, a card object CO6 is drawn from the deck CO6, and in the standby phase, the effect of a card object CO6 designated for effect processing in that phase can be activated. In the first main phase, various actions are permitted using card objects CO6 in the field area 54 as appropriate, such as summoning various objects such as monsters used in battle, setting card objects CO6 with unique effects such as spells and traps, or activating their effects. In the battle phase, a battle is conducted using card objects CO6. For example, a battle is conducted by selecting a monster card CO6 to be used in the user's attack during their turn and a monster card CO6 to be the target of the opponent's attack (a direct attack on Life Points if no monster card CO6 is present in the field area 54). The outcome of the battle is determined by parameters such as the attributes and strength of the monster card CO6. In the second main phase, the same actions as in the first main phase are permitted. The end of the turn is indicated (declared) during the end phase.

Fighting during the Battle Phase can be avoided by the player who has been given a turn. In that case, the Battle Phase and the Second Main Phase are skipped. Similarly, the player can avoid the Second Main Phase during the Battle Phase by choosing to do so. The end of a phase is indicated by a specified ending operation. When a turn ends, the turn passes to the opposing player. As turns are taken alternately, the game ends when a specified victory or defeat condition is met. As an example of a victory or defeat condition, as described above, the victory or defeat condition is met when the life points set for each player P decrease to a specified value (for example, 0) due to combat.

(Analysis services)
Next, the analysis service will be described in detail with reference to FIG. 5. As described above, the analysis service is a service that provides superiority/inferiority information as an analysis result for at least one option in a selection opportunity. The analysis service can be applied to any situation including a selection opportunity, but is applied, for example, to a selection opportunity included in a battle screen 50. FIG. 5 schematically shows an example of a battle screen 50 when the analysis service is applied. Note that in the example of FIG. 5, components that are the same as those in the battle screen 50 of FIG. 3 are assigned the same reference numerals, and their description will be omitted.

As shown in FIG. 4, a battle on the battle screen 50 is played by players repeatedly taking turns including multiple phases. For this reason, options on the battle screen 50 include a phase end command to end each phase (for example, entering the next battle phase in the first main phase), or a turn end command to end each turn. Furthermore, on the battle screen 50, deck cards (a group of card objects) are used throughout each phase as described above, and are placed in the field area 54, etc., via the hand CO5. Therefore, the user is required to select which hand CO5 to place in which card area CP and how. For magic cards CO2 and trap cards CO3, the user must also select when and how to activate their effects. Furthermore, for monster cards CO1, the user must also select whether to attack in the battle phase, and, if so, which monster card CO1 to attack. When the analysis service is applied to the battle screen 50, information on the relative merits of these options is provided.

Specifically, as shown in Figure 5, when an analysis service is applied, an advice display field 60 is additionally displayed compared to the example in Figure 3. The advice display field 60 is a field for displaying information on the relative merits of each option (analysis results) provided by the analysis service. Whether or not the analysis service is used, in other words, whether or not the advice display field 60 is displayed, can be set as appropriate, and may be displayed by a specified operation that requests its display, but as an example, it is automatically displayed at each selection opportunity without the need for a request from the user.

The advice display field 60 may display any number of pieces of superiority/inferiority information; for example, superiority/inferiority information for all options may be displayed. In the example of Figure 5, five pieces of advice are displayed via five advice units 61, the first advice unit 61A to the fifth advice unit 61E. The analysis service also uses multiple artificial intelligence models provided on the game server 2, which output multiple pieces of superiority/inferiority information (analysis results) for each option. For this reason, the advice display field 60 displays multiple pieces of superiority/inferiority information output by the multiple artificial intelligence models for each option. The number (types) of artificial intelligence models may be any number, but in the example of Figure 5, two types of artificial intelligence models, "offensive AI" and "defensive AI", are used, and two types of superiority/inferiority information are displayed for each option.

"Offensive AI" and "Defensive AI" refer to an offensive artificial intelligence model (hereinafter sometimes referred to as offensive AI) and a defensive artificial intelligence model (hereinafter sometimes referred to as defensive AI), respectively. The five advice sections 61 in the advice display column 60 are categorized into advice sections 61 that display superiority/inferiority information based on offensive AI and advice sections 61 that display superiority/inferiority information based on defensive AI. Advice sections 61 corresponding to offensive AI are displayed in white in the example of Figure 5, and correspond to, for example, the first advice section 61A, the third advice section 61C, and the fourth advice section 61D. On the other hand, advice sections 61 corresponding to defensive AI are displayed in gray in the example of Figure 5, and correspond to, for example, the second advice section 61B and the fifth advice section 61E.

As mentioned above, each artificial intelligence model is realized by the contents of the analytical trained data AD, for example, and each analytical trained data AD is generated by the contents of the training dataset. The offensive AI corresponds to a trained artificial intelligence model based on the analytical trained data AD generated using the offensive training dataset, and the defensive AI corresponds to a trained artificial intelligence model based on the analytical trained data AD generated using the defensive training dataset. For example, of the multiple types of analytical model units 23, the first analytical unit 23A corresponds to the offensive AI, and the second analytical model unit 23B corresponds to the defensive AI. The offensive training dataset and the defensive training dataset can be configured as appropriate, but for example, they could be configured by a dataset including a record set of wins achieved with an offensive-focused strategy and a dataset including a record set of wins achieved with a defensive-focused strategy, respectively.

Each advice section 61 may be configured as appropriate, but in the example of Figure 5, it includes an advice information section 62 and an option information section 63. The advice information section 62 is a section that displays detailed information including superiority/inferiority information. The superiority/inferiority information may be appropriate information regarding the impact of each option, such as symbols indicating the degree of impact each option has on victory for A, B, C, etc., or text information that intuitively indicates the impact each option has, such as "likely to lead to victory." In the example of Figure 5, however, it is configured as numerical information that indicates the likelihood of progressing advantageously in the battle through numerical magnitude. For example, the first advice section 61A displays numerical information such as "42.0%" as superiority/inferiority information. The numerical information may be a number indicating various advantages, such as a number indicating the likelihood of progressing advantageously (less likely to be disadvantaged) within a specified range of turns or options being analyzed. In the example of Figure 5, however, it is displayed as a probability value indicating the possibility (probability) of ultimately winning the battle.

The probability value is information that indicates the winning rate compared to other options output by the same artificial intelligence model. In other words, for example, the probability value for an offensive AI model is set so that the sum of the probability values of all options output by the offensive AI model equals 100%. The numerical information may be independent values (the sum of all options does not necessarily equal 100%), such as the evaluation results for each option by the offensive AI model, but as an example, it is calculated as a probability value that sums to 100% as shown here. Furthermore, the detailed information may include appropriate information in addition to probability value information (superiority information), and in the example of Figure 5, it includes information indicating the artificial intelligence model that output the probability value, such as "Offensive AI Type A Evaluation." Note that in the example of Figure 5, only one type (e.g., "Type A") of both the offensive AI and defensive AI is used, but multiple artificial intelligence models of the same type (e.g., offensive) based on different performance sets, such as "Type B," may also be used to output probability values.

The option information section 63 is a section that displays information about the options that are the subject of superiority/inferiority information (probability value). For example, the option information section 63 of the first advice section 61A displays information about the option "Enter the battle phase" (the option corresponding to the start of the battle phase). In this case, the first advice section 61A indicates that the probability value (superiority/inferiority information) for the option "Enter the battle phase" is "42.0%". Note that in the example of Figure 5, the display of the advice information sections 62 and option information sections 63 of the third advice section 61C to the fifth advice section 61E is simplified, but details may be displayed in the same way as the first advice section 61A, for example, by touching each advice section 61.

In addition, in the example of Figure 5, the probability value for the option "Enter Battle Phase" is output only by the offensive AI, but probability values for similar options are also output by the defensive AI. Specifically, both the offensive AI and defensive AI output probability values for all options that can be selected in each situation on the battle screen 50. Therefore, for each selectable option, at least two types of probability values are output, one for the offensive AI and one for the defensive AI. The advice display column 60 may be configured to display all two types of probability value information for each option, but in the example of Figure 5, it is configured to display only the top five options (some of the options) in terms of probability value.

The ranking of probability values may be evaluated for each artificial intelligence model, such as offensive AI and defensive AI, with the top ranking for each displayed in the advice display column 60; however, in the example of Figure 5, the ranking is evaluated across offensive AI and defensive AI (different artificial intelligence models). For example, if "Enter Battle Phase" is evaluated as having the highest probability value for the offensive AI, but "Enter Battle Phase" is evaluated as having a low probability value for the defensive AI, it is possible that only the analysis results for the offensive AI will be displayed in the advice display column 60. In this way, only one of the offensive AI and defensive AI may be displayed in the advice display column 60 as probability values for the same option; however, as described above, both the offensive AI and defensive AI output probability values for all options, and both may be displayed as advice in the advice display column 60.

(Method of calculating probability value)
Next, with reference to FIGS. 6 and 7 , a method for calculating the probability value (superiority/inferiority information) for each option output by each artificial intelligence model, such as an offensive AI, will be described. FIG. 6 is an explanatory diagram illustrating a list of options (actions) that a player can take in a certain situation on a battle screen 50. In the example of FIG. 6 , the options (conceptualized by "○") that a player can take are displayed in a tree format, with each option branching to the next option. As shown in FIG. 6 , the options that a player can take on the battle screen 50 are classified into multiple layers of option groups, and the options in the deepest layer in each tree correspond to the options that can actually be selected. Then, all of the options that can actually be selected are input to multiple artificial intelligence models, such as an offensive AI and a defensive AI, and a probability value for each option is output by each of the multiple artificial intelligence models.

Specifically, the example in Figure 6 shows the options available in a situation where three monster cards CO1 have already been placed in the front row 54A of the field area 54 in the example in Figure 5, and three magic cards CO2 have already been placed in the back row 54B, and the player has two cards, an "A monster card" and a "B magic card," in their hand CO5. In this case, the first options available to the player are: use "A monster card," use "B magic card," "phase transition," and "end turn." Therefore, these four options form the first tier of options that can be selected initially.

Furthermore, when using "A Monster Cards" and "B Magic Cards," there are also options for where and how to use them, forming a second layer of options. For example, in the example of Figure 5, the front row 54A of the field area 54 is divided into card areas 1 through 5, excluding the leftmost card area CP (where Monster Card CO1 cannot be placed), and proceeding from the leftmost card area CP to the right, and three Monster Cards CO1 are placed in card areas 3 through 5. If three Monster Cards CO1 are placed in card areas CP, the only card areas CPs where an "A Monster Card" (Monster Card CO1) can be placed are the first card area CP and the second card area CP. Therefore, these two card areas CP correspond to the options for where to place the "A Monster Card." Furthermore, there are two ways to place a Monster Card CO1 in a card area CP: "Summon" (where Monster Card CO1 is placed face-up) and "Set" (where Monster Card CO1 is placed face-down). For this reason, for each of the two options for the first card space CP and the second card space CP, there are two further options: "Summon" and "Set." As a result, there are four options for using an "A Monster Card," corresponding to the two card space CPs and two placement methods. There may also be various other options for Monster Card CO1 depending on the situation. For example, in the case of Extra Monster Card CO1, it is necessary to specify a predetermined number (e.g., three) of Monster Card CO1s that should be removed from the field area 54 (to be moved to the Graveyard area 58) for the summon. In this case, options such as the predetermined number (e.g., three) x the type of Extra Monster Card CO1 x the card space CP may be derived.

On the other hand, for example, in the field area 54 in the example of Figure 5, if the back row 54B is divided into the 6th card place CP through the 10th card place CP from left to right, and three Magic Cards CO2 and the like are placed in the 8th card place CP through the 10th card place CP, then the two card place CPs in which a "B Magic Card" (Magic Card CO2) can be placed are the 6th card place CP and the 7th card place CP. Therefore, these two card place CPs correspond to the options corresponding to the placement destination of the "B Magic Card." Furthermore, while a "B Magic Card" can be configured as any Magic Card CO2, for example, if a "B Magic Card" is a Magic Card CO2 that has two placement methods for a card place CP, "activate" (the option to activate the effect) and "set" (the option to withhold the activation of the effect), then these two placement methods are available as options. As a result, there are also four options for using "B Magic Cards," corresponding to the two card placement locations (CP) and two placement methods. These four options, plus the four options for "A Monster Cards," make up a total of eight options that form the second tier of options.

If an "A Monster Card" has an effect set, further options will be generated. For example, if an "A Monster Card" has an effect that requires it to be placed in the card space CP by "Summoning," then there are two further options for the "Summon" placement method: "Activate effect" and "Do not activate effect." Therefore, four options corresponding to these two options x two card space CP will be generated from the second tier options for "A Monster Cards."

Similarly, options are derived when two types of effects are set for a "B Magic Card." For example, if a "B Magic Card" has two types of effects, Effect A and Effect B, there are two further placement options: "Activate Effect A" and "Activate Effect B." Therefore, four options corresponding to these two options x two card placement locations CP are generated from the second-tier options group for the "B Magic Card." These four options are then joined by four options derived from the second-tier options group for the "A Monster Card," resulting in a total of eight options forming the third-tier options group. Note that various other options may exist for the Magic Card CO2 depending on the situation. For example, if the destination of the effect can be specified, further options for selecting that destination may be generated. Although not shown in the example in Figure 6, the same applies to other card objects CO, such as the Trap Card CO3.

Furthermore, fourth-layer options are derived from the third-layer options. For example, if two of the three monster cards CO1 already placed in the card space CP (e.g., a "C Monster Card" and a "D Monster Card") can be targets for the effect of a "B Magic Card," these two targets are derived as options. Therefore, eight options corresponding to these two targets x four options (two types of effects that occur for each of the two card space CPs) form the fourth-layer options.

In the example of Figure 6, in the tree of options derived from "A Monster Card," part of the second-layer options (when placed in a "set" state in each card place CP) and the third-layer options correspond to the options in the deepest layer. Therefore, each option in these option groups is input into an artificial intelligence model such as an offensive AI, and a probability value (number in "○") is output for each option. Similarly, in the tree of options derived from "B Magic Card," part of the second-layer options (when placed in a "set" state in each card place CP) and the fourth-layer options correspond to the options in the deepest layer. Therefore, each option in these option groups is input into an artificial intelligence model such as an offensive AI, and a probability value is output for each option.

On the other hand, there are no options derived from "Phase Transition" (phase end instruction) and "Turn End" (turn end instruction), and for these, the first layer options group corresponds to the deepest layer options group. For this reason, the two options "Phase Transition" and "Turn End" are input into an artificial intelligence model such as an offensive AI, and a probability value is output for each option. Note that the example in Figure 6 shows probability value information output by one type of artificial intelligence model out of multiple artificial intelligence models as an example, but in reality, multiple probability value information is output for each option by each of the multiple artificial intelligence models.

The example in Figure 6 shows an example of a set of options in the battle phase, and in addition to phase end instructions, it shows the card object CO to be placed from the hand CO5 in each card place CP, and the options corresponding to the card place CP where that card object CO should be placed. However, options are prepared for each phase depending on the phase, such as adding (drawing) a card from the deck CO6 to the hand CO5 in the draw phase. For this reason, at least some of the options available for each phase (for example, the end of the phase or the end of the turn may be the same) are different. In other words, the options available in each phase are different. Then, probability value information for each option available for each phase is output. In this example, the set of options at the deepest level functions as an example of the multiple options of the present invention. Furthermore, the set of options for each phase functions as two or more options of the present invention.

Figure 7 is an explanatory diagram illustrating an example of a method by which an artificial intelligence model calculates probability values. Generally, artificial intelligence models are generated by machine learning a training dataset using a predetermined learning method (algorithm), and tend to have a way of thinking (algorithm) that corresponds to the learning method. Learning methods in machine learning include various methods, such as imitation learning and reinforcement learning (imitation learning is sometimes considered a type of reinforcement learning). Imitation learning is often classified as a method such as behavior cloning, dataset aggregation, or inverse reinforcement learning, but rewards are often not explicitly defined. On the other hand, reinforcement learning is often classified as a method such as dynamic programming (DP), Monte Carlo (MC), or temporal difference learning (TD), and is often a method that maximizes rewards.

The artificial intelligence model (analysis program AP) for the analysis service may be generated using various learning methods, but as an example, it is generated using the TD method of reinforcement learning. The TD method may also include techniques such as SARSA, but as an example, the Q-learning method is used to generate the artificial intelligence model for the analysis service.

Q-learning is a method for evaluating actions (options) by calculating an action-value function (Q-function). A Q-function is generally defined as a function that predicts the future reward (often called a Q-value) that will result from performing a certain action in a certain state (the specific function formula will not be explained). However, with Q-learning, when the Q-function table (input information) becomes large (huge in amount), the calculations tend to be unable to keep up, making it difficult to implement. On the other hand, for example, the input information (information in the situation log SL) on the battle screen 50 is thought to be enormous. For this reason, a trained artificial intelligence model for analysis services is generated using the DQN (Deep Q Network) method, which uses a neural network to obtain an approximation of the Q-value. In this case, the artificial intelligence model for analysis services is configured to calculate probability values (Q-values) using the DQN method.

The reward (a predetermined result obtained by an action, which may be included in the algorithm of the learning program) that is the subject of the Q value may be set as appropriate, and a specific way of winning, such as a narrow victory, or a tendency for winning, may be set. The AI model for the analysis service may think differently depending on the reward (predetermined result) in Q learning, but winning a match is set as an example of a reward, and the AI model for the analysis service is configured to calculate the probability value (Q value) for winning a match using the DQN method.

The example in Figure 7 shows an overview of the DQN method for calculating probability values (Q values). As shown in Figure 7, the DQN method uses a neural network (deep learning), forming an input layer, an intermediate layer, and an output layer, each of which cooperates to output calculation results for data. The input layer is responsible for collecting data. The intermediate layer is responsible for calculations to calculate probability values. Intermediate layers often include multiple layers (generally, the more layers there are, the higher the accuracy tends to be). While only two intermediate layers are shown in the example in Figure 7, any number of intermediate layers may be formed. The output layer is a layer for outputting the results of calculations performed in the intermediate layers. An artificial intelligence model trained using the DQN method is configured to output the probability value of winning a match in the output layer. Furthermore, connecting lines (often called synapses) are provided between the input, intermediate, and output layers. Each connection line is assigned a weight (often represented by the letter w) that indicates importance (strength of connection), and the importance of the information is determined by the size of the weight value. In the example of Figure 7, both input values (input information) and output values (sometimes commonly called nodes) are represented by "○".

Specifically, the input layer first receives information from the situation log SL, which indicates the current state of the battle screen 50. The situation log SL contains information on many dimensions (input values). It is desirable for the number of dimensions to be less than approximately 5,000. As mentioned above, the situation log SL can contain various information used to determine the status of the card game. For example, in addition to information on the placement status in the field area 54, such as the status of the first card place CP to the tenth card place CP, it also contains information on possible options such as ending the turn, summoning an A monster card, activating the effect of an A monster card, and activating a B magic card (see, for example, the example in Figure 6).

A predetermined function formula using appropriate weighting (weight value w) is applied to the input values in the input layer, and the output values of the intermediate layer (first layer) are calculated using this function formula. Any number of output values can be calculated for the first intermediate layer, and in the example of Figure 7, four output values are calculated. A predetermined function formula using appropriate weighting is also applied to the four output values, and the output values of the intermediate layer (second layer) are calculated using this function formula. Any number of output values can be calculated for the second intermediate layer, and in the example of Figure 7, four output values are also calculated. The weighting (weight value w) corresponding to each connecting line is calculated as a learning result and is managed, for example, in the learned data for analysis AD. The connecting lines (synapses) connecting the input layer and intermediate layer have been omitted from Figure 7 as appropriate.

The output value of the final intermediate layer (second layer) is calculated as a Q value using a predetermined function formula with appropriate weighting in the output layer. The Q value (output value of the output layer) is calculated for each predetermined action (option on the battle screen 50). For example, in each tree in the example of Figure 6, a Q value is calculated for each option (selectable option) in the option group at the deepest layer using the method shown in Figure 7. The calculated Q value is then converted to a probability value using a normalization function (e.g., the Softmax function) so that the sum of the values for all selectable options equals 100%. As an example, this method is used to calculate probability values for all selectable options for each situation. Similar calculations are also performed by multiple artificial intelligence models, such as offensive AI or defensive AI (e.g., the first analysis program AP1 and the second analysis program AP2). The top probability values calculated by the multiple artificial intelligence models (e.g., the five in the example of Figure 5) are then provided to the user via the advice display field 60. Note that applying the normalization function is not required. For example, if conditions such as learning methods are different, the application of the normalization function may be omitted as appropriate.

(guessing service)
Next, the details of the guessing service will be described with reference to FIG. 8 . The guessing service is a service that provides guess information for guessing the contents of an opponent's elements (e.g., card objects CO) that are not shared with the user, among a group of elements (e.g., card objects CO) prepared for each player in a card game. The guessing service can be applied to any situation involving elements that are kept secret from the user. For example, the guessing service is applied to guessing card objects CO, such as card objects CO, hand cards CO5, deck cards CO6, or summon deck cards CO7, placed face down in a hidden field area 54 on the battle screen 50. In other words, on the battle screen 50, guess information for guessing the contents of card objects CO, etc., placed face down is provided via the guessing service. FIG. 8 schematically illustrates an example of a battle screen 50 when the guessing service is applied. Note that in the example of FIG. 8 , components similar to those of the battle screen 50 in FIG. 3 are designated by the same reference numerals, and their description will be omitted.

As shown in FIG. 8, when the guessing service is applied, the battle screen 50 additionally displays a guess information section 70 compared to the example of FIG. 3. The guess information section 70 is a section for displaying guess information. The guess information section 70 may be configured as appropriate, but in the example of FIG. 8, it includes a target designation section 71, a candidate image 72, and a candidate arrangement line 73. The target designation section 71 is a section indicating the target (guessing target) for which guess information is displayed. The target designation section 71 may be configured as appropriate as long as it allows the user to identify the target, but in the example of FIG. 8, it is indicated by a dashed line surrounding the guessing target card object CO (a card object CO whose face-up side is hidden). Furthermore, the guess information section 70 may be displayed to target any appropriate card object CO as the guess target; for example, it may be displayed to target all face-down card objects CP or any appropriate portion of them. In the example of FIG. 8, it is displayed to target the face-down card object CP designated (selected) by the user. In this case, the target designation unit 71 has the function of distinguishing the card object CP designated by the user as the target for guessing from other face-down card objects CP.

Candidate images 72 are images that represent candidates for the card object CO to be guessed. More specifically, images of the front side (contents) of a card object CO that is guessed to correspond to the card object CO to be guessed are displayed as candidate images 72. While only the most likely candidate card object CO may be displayed as candidate image 72, the example in FIG. 8 displays two candidate images 72 ("A" and "B") corresponding to the top two most likely candidates (this is just an example, and any appropriate number may be used). A numerical value indicating the probability (hereinafter sometimes referred to as a match probability value) is displayed above each candidate image 72. For example, the value "56%" is displayed above the candidate image 72 corresponding to "A," and the value "35%" is displayed above the candidate image 72 corresponding to "B." These match probability values are output by the trained artificial intelligence model for the guessing service (guessing program FP) based on the situation log SL.

The candidate arrangement line 73 is a line that indicates the relationship between the target designation section 71 and each candidate image 72. Each candidate image 72 is arranged below and along the candidate arrangement line 73. Therefore, the candidate arrangement line 73 also functions as a reference for aligning each candidate image 72 horizontally.

(Method of calculating estimated information)
Next, with reference to FIG. 9 , a method for calculating inferred information (candidates and match probability values) output by a trained artificial intelligence model for the inference service will be described. FIG. 9 is an explanatory diagram illustrating an example of a method by which an artificial intelligence model calculates inferred information. The example of FIG. 9 illustrates a case in which inferred information is provided on a battle screen 50. While the artificial intelligence model for the inference service may also be trained using an appropriate learning method, the example of FIG. 9 illustrates a case in which the artificial intelligence model is generated using a neural network (deep learning) technique. As shown in FIG. 9 , in this case, the inferred information is calculated using an input layer, an intermediate layer, and an output layer, similar to the calculation of probability values. Specifically, in the input layer, information indicating the situation on the battle screen 50 is input as input information (input value) to the trained artificial intelligence model for the inference service. Information different from that used in the calculation of probability values may be used as input information; however, as an example, a similar situation log SL (only a portion of which is shown in FIG. 9 ) is used.

In the intermediate layer, output values are calculated from input values in the input layer using a predetermined function formula that utilizes appropriate weighting (weight values), but the weight values and function formula used there differ from those in the artificial intelligence model for the analysis service. The weight values are defined by the learned data for prediction FD (data generated by the above-mentioned learning), and the function formula is configured to predict the contents of the face-down card object CO. Any number of intermediate layers can be formed to calculate the prediction information, but in the example of Figure 9, two intermediate layers are formed. Similarly, any number of output values (nodes) can be provided in each intermediate layer, but in the example of Figure 9, four output values are provided in each intermediate layer.

In the output layer, a value indicating the likelihood that the card object CO corresponds to the card object CO being predicted is calculated for each card object CO from the output value of the final intermediate layer (second layer). The card objects CO for which the likelihood is calculated may be all types that can be used in a card game (the opponent's deck cards do not need to be taken into consideration), but as an example, they are limited to card objects CO that the opponent has incorporated into their deck cards. This limitation may be implemented as appropriate; for example, if a learning dataset containing only the opponent's deck cards as performance samples is prepared in advance and a trained artificial intelligence model (prediction model unit 24) that has trained this learning dataset is provided, the limitation may be implemented by using that artificial intelligence model, but as an example, after prediction candidates are calculated, only cards other than the opponent's deck cards are filtered out from the candidates.

Additionally, rules are reflected appropriately in card games. For example, rules may limit the number of card objects CO of a particular type that can be used in a deck, such as limiting the number of cards that can be used to three. In this case, if all three card objects CO of that particular type are already open (for example, placed face up in the field area 54, etc.), those card objects CO of that particular type will be excluded (filtered out) from the scope of speculation information.

In addition, rules may be revised from time to time (including regularly scheduled times, such as once a year, and irregular times, such as temporary times to respond to situations as they arise). For example, the number of cards that can be used may change, such as changing the number of a certain type of card object CO that can be included in a deck card from three to two. Similarly, regulations may be changed from time to time, such as limit regulations (restrictions that make certain card objects CO unusable). For example, if a card object CO that is too strong exists, that card object CO may always be included, which could lead to a bias in deck cards. Alternatively, this could lead to a sense of unfairness in the results of matches. For this reason, for example, if a certain card object CO is banned from use, that specific card object CO is also excluded (filtered out) from the targets used to calculate possibilities. Alternatively, if the number of cards used is changed to two, that card object CO is excluded from the target of speculation information once both cards are revealed (if the limit is increased, it will not be excluded from the target until the limit is reached).

In a card game, as mentioned above, various changes, such as changes in number or type, can occur in the usable card objects CO (elements) in response to rule changes, etc. These changes are then reflected in the targets of the speculated information. Specifically, the range for which the value indicating the possibility is calculated (in other words, card objects CO that are candidates for the hidden card object CO) is limited to card objects CO that are usable under the current rules and that the opponent has included in their deck cards.

Furthermore, the values indicating the possibilities in the output layer (the values of each candidate after being filtered to the available candidates) are normalized using a normalization function (e.g., the Softmax function) so that the sum of the values of all selectable options (e.g., all card objects CO included in the opponent's deck) equals 100%. The normalized value is then calculated as the guess information (match probability value). As an example, using this method, candidate card objects CO and match probability values are calculated for each guess target card object CO (card objects CO whose contents are concealed) using a trained artificial intelligence model for the guess service. Then, of the calculated candidate card objects CO, the ones with the highest match probability values (e.g., the two in the example of Figure 8) are provided to the user via the guess information unit 70. Furthermore, if the information input in the input layer includes executable actions (options), even if the situation is the same, consideration is given to how they were used to arrive at the current situation (board state). This enables the calculation of highly accurate guess information.

(Types of inferred information)
Next, with reference to FIGS. 10 and 11 , the types of inferred information that may be provided via the inference service will be described. While only one type of trained artificial intelligence model may be used for the inference service, as an example, multiple types of artificial intelligence models (e.g., a first inference program FP1 and a second inference program FP2) may be prepared. Each artificial intelligence model outputs inferred information based on different inference-use trained data FD. Because different artificial intelligence models may reach the same conclusion, i.e., the same inferred information, multiple types of artificial intelligence models do not necessarily output different inferred information. However, in many cases, multiple types of inferred information are output. FIG. 10 is an explanatory diagram illustrating an example of the types of inferred information output by multiple types of artificial intelligence models. Furthermore, each artificial intelligence model may output various types of inferred information as appropriate. In the example of FIG. 10 , (1) recommended information by period and (2) recommended information by rank are shown. Furthermore, while appropriate recommended information (including all) from among these multiple recommended information may be displayed, in the example of FIG. 10 , only a portion of the recommended information selected by the user is displayed in both (1) and (2).

Recommendation information for each period is recommended information output by a trained artificial intelligence model based on the trained data for inference FD for each period generated using the training data set for each period. In this case, multiple trained data for inference FD are generated based on multiple training data sets (each containing performance samples for multiple periods). The inference information for each period may be provided as appropriate, but as an example, it is switched based on a user instruction (selection). The user instruction may be executed as appropriate, but the example in Figure 10 shows a case where it is executed via the information switching unit 80.

On the other hand, the recommended information for each rank is recommended information output by a trained artificial intelligence model based on the trained data for prediction FD for each rank generated using the training data set for each rank. Card games are equipped with a function for ranking players (information indicating the player's skill) according to their performance, and the recommended information for each rank corresponds to the output result of an artificial intelligence model generated from a training data set containing performance samples of the same rank. In this case, multiple trained data for prediction FD corresponding to multiple training data sets (each containing performance samples of multiple ranks) are prepared. The predicted information for each rank may be provided as appropriate, but as an example, it is provided so that it can be switched by user instruction via the information switching unit 80, similar to the predicted information for each period. In the example of Figure 10, the predicted information unit 70 is enlarged and schematically shown when the battle screen 50 includes the information switching unit 80 in both (1) and (2).

As shown in (1) and (2) of FIG. 10, the information switching unit 80 includes an information type unit 81, a first switching unit 82, and a second switching unit 83. The information type unit 81 is a unit for displaying the type of inferred information currently being displayed. The first switching unit 82 and the second switching unit 83 are both units that indicate the position where an instruction (for example, a touch operation) for switching the type of inferred information being displayed should be executed. When a touch operation is executed on the first switching unit 82 or the second switching unit 83, the currently displayed object is switched to the previous or next displayed object, respectively.

Specifically, as shown in (1) of Figure 10, when inferred information for each period is provided, the type of inferred information is expressed, for example, by the information "23.1-23.2" indicating the period. The information "23.1-23.2" (indicating the one-month period from January to February 2023) indicates the period during which the performance samples included in the training dataset from which the information was generated were collected. On the other hand, when a touch operation is performed on the first switching unit 82, the display object is switched to a period prior to the currently displayed period (e.g., the one-month period from December to January 2022). In other words, the display object is switched to inferred information (candidate image 72 including a matching probability value) output by an artificial intelligence model generated using multiple training datasets, each of which includes performance samples from the previous period. Similarly, when a touch operation is performed on the second switching unit 83, the display object is switched to a period subsequent to the currently displayed period (e.g., the one-month period from February to March 2023). In the example of Figure 10, the range of display targets is divided into monthly intervals, but this is not limited to this and may be set as appropriate.

Furthermore, as shown in (2) of FIG. 10, when inferred information for each rank is provided, the type of inferred information is represented by "second rank" information indicating the rank. The "second rank" information indicates the rank of the player from whom the performance samples included in the training dataset from which the inferred information was generated were collected. In this case, this means that the inferred information was generated from multiple training datasets, each containing performance samples only of players of the second rank. On the other hand, when a touch operation is performed on the first switching unit 82, the display object is switched to a rank lower than the rank of the currently displayed object (e.g., first rank). In other words, the display object is switched to inferred information (candidate image 72 including a match probability value) output by an artificial intelligence model generated using multiple training datasets, each containing performance samples only of players of the first rank, which is one rank lower than the current second rank. Similarly, when a touch operation is performed on the second switching unit 83, the display object is switched to a rank higher than the rank of the currently displayed object (e.g., third rank).

In the example of Figure 10, the range of display targets is divided into ranks, but this is not limited to this and may be any appropriate range (for example, two or more ranks), in which case information indicating that range may be added. The initial display target (default display) may also be set as appropriate, for example, to the user's own rank or the opponent's rank. In this case, estimated information tailored to the user's own or opponent's rank is displayed by default, thereby improving usability. The initial display for the period may also be set as appropriate, for example, to the most recent period or the most frequently used period.

Each artificial intelligence model can output various types of inferred information based on differences in algorithms, etc., but as mentioned above, in the example of Figure 10, multiple types of inferred information are output based on differences in the inferred learned data FD. In card games, trends may occur in the types of card objects CO used (deck card combinations). When trends occur, an artificial intelligence model that reflects those trends is more likely to output more accurate inferred information. Trends often occur over time. Therefore, if inferred information for each time period is provided, trends can be reflected in the inferred information, thereby enabling more accurate inferred information. Similarly, with card objects, there are often differences in the card objects CO incorporated into the card deck and how they are used depending on their rank (player's skill). Therefore, if inferred information for each rank is provided, usage trends for each rank can be reflected in the inferred information, thereby enabling more accurate inferred information.

The multiple learning datasets are not limited to the above-mentioned learning datasets for each period or each rank, and any suitable learning datasets including results collected from various perspectives, such as rules (regulations) and the number of cards that can be used, may be used. For this reason, for example, a learning dataset including performance samples for all card objects CO included in a card game (all regardless of period or rank) may be used. In other words, the multiple learned data for inference FD may be data generated using various learning datasets that provide appropriate thinking to the inference program.

Figure 11 is an explanatory diagram illustrating learning datasets for realizing inferred information for each period and for each rank. The inferred information for each period and for each rank may be prepared separately as independent inferred information, and may be realized by separate learning datasets, such as multiple learning datasets for each period or multiple learning datasets for each rank. On the other hand, inferred information that takes both into consideration may also be prepared, such as inferred information for each period being prepared for each rank. The example in Figure 11 shows the types of learning datasets when inferred information that takes both into consideration is prepared.

As shown in FIG. 11, when inferred information that takes both into account is prepared, types of training datasets corresponding to the number of periods x the number of ranks are prepared. Specifically, a training dataset corresponding to the first rank (including performance samples of first-rank players) is prepared for each period, such as a "first training dataset" corresponding to the period "2023.1-2023.2" and a "second training dataset" corresponding to the period "2023.1-2023.2" (both training datasets include performance samples of players for the corresponding period). In this case, for example, an artificial intelligence model generated using the "first training dataset" will output inferred information that takes into account (reflects) both the trends of first-rank players and the trends of deck cards that were popular from January to February 2023.

The same applies to the second and third ranks, where training datasets are prepared for each period, such as the "third training dataset" through the "sixth training dataset." An artificial intelligence model based on these will then output inferred information that takes into account (reflects) the trends for each rank and each period. Note that while the example in Figure 11 shows first through third ranks, a card game may have any number of ranks, and training datasets corresponding to that number may be prepared. The same applies to periods.

(Network system processing)
Next, with reference to Figures 12 to 14, the choice analysis process, display change process, and behind-the-scenes inference process will be described as examples of processes performed by the network system 1. The choice analysis process is a process for providing multiple types of superiority/inferiority information for each choice in a selection opportunity using multiple types of artificial intelligence models. While the choice analysis process may be configured to provide multiple types of superiority/inferiority information by using multiple types of artificial intelligence models in parallel, the example of Figure 12 illustrates a case in which multiple types of artificial intelligence models are used in series to provide multiple types of superiority/inferiority information. Furthermore, the choice analysis process is realized by cooperation between the analytical model unit 23 of the game server 2 and the model processing unit 35 of the user device 3. However, in the example of Figure 12, the process primarily executed by the analytical model unit 23 is shown as the game server 2, and the process primarily executed by the model processing unit 35 is shown as the user device 3.

When a predetermined analysis time (e.g., a user's turn on the battle screen 50 or each phase) arrives, the model processing unit 35 begins the option analysis process of FIG. 12 and first requests the game server 2 to analyze each option at the selection opportunity (step S101). This request may be executed appropriately, for example, by issuing a request to the analytical model unit 23, and may be executed so that the analysis results of each analytical model unit 23, such as the first analytical model unit 23A and the second analytical model unit 23B, are output in a predetermined order in the analytical model unit 23. As an example, the request is executed in a predetermined order to one of the analytical model units 23, such as the first analytical model unit 23A and the second analytical model unit 23B. Specifically, in step S101, the model processing unit 35 first requests the first analytical model unit 23A to analyze each option. This request also includes a situation log SL indicating the situation at the time of the request.

When an analysis request is sent, the first analytical model unit 23A starts the choice analysis process of FIG. 12 and first acquires the request (step S201). The first analytical model unit 23A then performs an analysis of each choice based on the situation log SL included in the acquired request (step S202). This analysis may be performed on only some of the choices in the selection opportunity, but as an example, it is performed on all choices. Once the analysis is performed, the first analytical model unit 23A outputs the analysis results (step S203) and sends them to the model processing unit 35 (step S204). After sending the results, the first analytical model unit 23A then terminates the current choice analysis process.

On the other hand, when the analysis results are transmitted from the first analytical model unit 23A, the model processing unit 35 acquires the analysis results and determines whether the number of types of acquired analysis results (in other words, artificial intelligence models) has reached a set number (for example, two: offensive AI and defensive AI) (step S102). If the number of types of acquired analysis results has not reached the set number (step S102: No), the model processing unit 35 returns to step S101 and requests analysis from the next analytical model unit 23 (second analytical model unit 23B) in accordance with a predetermined order (step S101).

When an analysis request is sent, the second analytical model unit 23B starts the option analysis process of FIG. 12 and first receives the request (step S201). Then, similar to the first analytical model unit 23A, it performs an analysis of each option (step S202), outputs the analysis results (step S203), and sends them to the model processing unit 35 (step S204). After sending the results, the second analytical model unit 23B ends the current option analysis process.

On the other hand, if the number of types of analysis results obtained reaches the set number (step S102: Yes), the model processing unit 35 extracts the analysis results (superiority/inferiority information) to be displayed from all the obtained analysis results (including the analysis results of multiple types of analytical model units 23) (step S103). The analysis results to be displayed can be set as appropriate, but as an example, they are set to the analysis results showing the top five probability values across the analysis results of multiple types of analytical model units 23. Therefore, in step S103, the model processing unit 35 extracts the analysis results corresponding to the top five probability values from all the analysis results as the display targets. Next, the model processing unit 35 displays an advice display column 60 on the battle screen 50, which includes the five analysis results extracted in step S104 as advice fields 61. After this display, the model processing unit 35 terminates the current option analysis process. This allows the analysis service to be realized on the battle screen 50 via the advice display column 60. More specifically, a battle screen 50 is realized that includes an advice display section 60 that carefully selects and displays the top five probability values for each option, based on information output from multiple artificial intelligence models.

The display change process is a process for changing the display mode of each analysis result in the advice display field 60, such as the display target or sorting order. The display mode of the advice display field 60 may be fixed, for example, displaying the top five results in order of probability value, but is also variable and changes according to user specifications. When a user instructs the model processing unit 35 to change the display mode of the advice display field 60, the model processing unit 35 starts the display change process of Figure 13 and first determines the conditions specified by the user (step S301). The user may be allowed to specify any desired settings, but are limited to pre-prepared conditions, for example. The pre-prepared conditions may include appropriate conditions, such as filter conditions and sort conditions.

Filter conditions are conditions for limiting the analysis results to a portion of the display target. Filter conditions may be various conditions for limiting the display target to an appropriate portion, but include, for example, limiting the display target to the analysis results of a specific artificial intelligence model (if three or more types of analytical model units 23 are provided, this may be two or more types of analytical model units 23) based on user specification. Sorting conditions (rearrangement conditions) are conditions for rearranging the analysis results in a predetermined order specified by the user. For this reason, in step S301, the model processing unit 35 determines conditions such as filter conditions or sorting conditions as specified conditions. In the case of filter conditions, the model processing unit 35 also determines the specific artificial intelligence model specified as the filter target as a specified condition. On the other hand, in the case of sorting conditions, the model processing unit 35 also determines the predetermined order specified as the sorting condition as a specified condition. Both filter conditions and sorting conditions may be specified, in which case both are determined.

Next, the model processing unit 35 determines the analysis results (superiority information) of the target corresponding to the specified conditions determined in step S301 from all analysis results output by the analytical model unit 23 (all analysis results acquired in step S102 of the option analysis processing in FIG. 12) (step S302). Specifically, the model processing unit 35 determines the analysis results output by the specific artificial intelligence model specified as the filter condition (a specific type of analytical model unit 23, such as the first analytical model unit 23A or the second analytical model unit 23B). Alternatively, the model processing unit 35 determines each analysis result corresponding to a predetermined sort order. The predetermined sort order may be specified as appropriate, and may be specified as an order for each artificial intelligence model, such as the second analytical model unit 23B (defensive AI) or the first analytical model unit 23A (offensive AI), or in order of decreasing probability value, in order of the top of a predetermined range of probability values, or in order of the sum of probability values from multiple artificial intelligence models (which may be in order of highest or lowest, as appropriate).

Next, the model processing unit 35 changes the display of the advice display field 60 to display the analysis results of the targets determined in step S302 according to the specified conditions (step S303). For example, if the analysis results of the first analytical model unit 23A (e.g., an offensive AI) are specified as the filter conditions, the display of the advice display field 60 is changed to display only the analysis results of the first analytical model unit 23A. If the sorting condition specifies a sorting order for each AI model, the display of the advice display field 60 is changed to arrange the advice units 61 for each AI model. The same applies when a sorting order is specified, such as lowest probability value, highest probability value within a specified range, or total probability value order for multiple AI models. In these cases, the number of displayed targets may be changed as appropriate, but is set to five, as before the change, as an example. Therefore, the model processing unit 35 changes the display of the advice display field 60 to display advice units 61 corresponding to the top five probability values of the first analytical model unit 23A or the top five probability values in the specified sorting order. After this change, the model processing unit 35 terminates the current display change process. This changes the display of the advice display field 60 so that the analysis results specified by the user are displayed in the specified order, etc.

The back-side guessing process is a process that uses a trained artificial intelligence model to provide guess information about the opponent's back-side card object CO (information not shared with the user) on the battle screen 50. The back-side guessing process is realized through cooperation between the guessing model unit 24 of the game server 2 and the model processing unit 35 of the user device 3; however, in the example of Figure 14, the process executed primarily by the guessing model unit 24 is shown as the game server 2, and the process executed primarily by the model processing unit 35 is shown as the user device 3.

When a predetermined start condition is met (for example, when a face-down card object CO is selected by the user on the battle screen 50 by a touch operation, etc.), or when an instruction to switch the type of guessed information is given via the first switching unit 82 of the information switching unit 80, etc., the model processing unit 35 starts the back-side guessing process of FIG. 14, and first requests the game server 2 to guess the target face-down card object CO on the battle screen 50 (the card object CO selected by a touch operation, or the card object CO selected before the instruction to switch) (step S401). This request includes a situation log SL that indicates the situation at the time of the request. Furthermore, when an instruction to switch the type of guessed information is given via the first switching unit 82 of the information switching unit 80, etc., the request also includes information on the type of guessed information specified in that instruction.

When a request for inference is sent, the inference model unit 24 initiates the back-side inference process shown in FIG. 14 and first acquires the request (step S501). The inference model unit 24 then infers the contents of the face-down card object CO based on the situation log SL included in the acquired request (step S502). If an instruction to switch the type of inference information is issued via the first switching unit 82 of the information switching unit 80, the inference is performed by the inference model unit 24 of the type corresponding to the instructed type, such as the first inference model unit 24A or the second inference model unit 24B. The inference model unit 24 then outputs the inference results of the designated artificial intelligence model, such as the first inference model unit 24A (step S503). In this case, the inference model unit 24 may output only the number of inference results to be displayed as candidate images 72 (e.g., the top two), but as an example, the output includes information on all candidates. Furthermore, the estimation model unit 24 may output estimation information regarding all face-down card objects CP included in the battle screen 50 as estimation results (in this case, if a target other than the currently specified target is subsequently specified as a new target for estimation, the user device 3 may extract and display the estimation information for the newly specified target from the estimation information that has already been output), but as an example, only estimation information regarding the currently specified face-down card object CP is output.

Next, the inference model unit 24 adjusts the inference result (step S504). The adjustment can be implemented as appropriate, but as an example, it is performed to reflect rules such as limit regulations and the opponent's deck cards. For example, the inference model unit 24 adjusts the inference result to exclude from the output inference result any card objects CO that have become prohibited for use and any card objects CO that the opponent has not included in his deck cards. This exclusion may also be performed by simply excluding prohibited card objects CO from the candidates output as the inference result, but as an example, it is performed by recalculating (adjusting) the match probability value after the exclusion. The inference model unit 24 then transmits the adjusted inference result to the model processing unit 35 as inference information (step S505). After this transmission, the inference model unit 24 terminates this back-side inference process. Note that step S504 may be omitted as appropriate if adjustment is not required (including if adjustment is not performed at all), such as when processing is performed by an artificial intelligence model that has trained on a training dataset limited to the adjusted target (when similar results can be obtained without adjustment), as described above.

On the other hand, when the inferred information is transmitted from the inferred model unit 24, the model processing unit 35 acquires the inferred information (step S402). The model processing unit 35 then identifies the inferred information to be displayed from the acquired inferred information (step S403). For example, if the number of candidate images 72 to be displayed in the inferred information unit 70 is the top two in terms of match probability, the model processing unit 35 identifies the inferred information with the top two in terms of match probability from the acquired inferred information as the display target. The model processing unit 35 then displays the inferred information to be displayed identified in step S404 on the battle screen 50 (step S404). Specifically, the inferred information unit 70 including the inferred information identified in step S404 is displayed on the battle screen 50. After this display, the model processing unit 35 terminates the current back-side inference process.

The procedure in Figure 14 realizes a guessing service on the battle screen 50 via the guess information unit 70. More specifically, a battle screen 50 is realized that includes a guess information unit 70 that carefully selects and provides the top two guess information for each face-down card object CO (e.g., information including the contents of candidate card objects CO and the probability of a match) output from an artificial intelligence model as information for guessing the contents of face-down card objects CO. Note that the display targets or order of the guess information unit 70 may be changed as appropriate. In this case, the display targets, etc. may be changed using a procedure similar to the display change process in the example of Figure 13 (except that the guess information and guess information unit 70 are used instead of the analysis results and advice display column 60).

As explained above, according to this embodiment, multiple probability values (superiority information) indicating the probability of winning the battle for each option on the battle screen 50 are output by multiple types of trained artificial intelligence models (first analytical model unit 23A, second analytical model unit 23B) via the analysis service (which may be information indicating various possibilities for favorable progression within a specified range of the analysis target), and information on the top five probability values among these is provided before an option is selected. Therefore, it is possible to provide probability value information (superiority information) from various perspectives for each selection option, such as which card object CO should be placed in which card place CP in order to win the battle, or whether to end a phase, in other words, numerical information indicating the magnitude of the impact on winning the game.

In addition, in card games, players are given alternating turns each containing multiple phases, and a choice opportunity is provided for each phase. However, the end of a phase and the end of a turn are also included in the set of options for each choice opportunity. When the change of turn, etc., functions as one of the options, the impact of choices made at choice opportunities on progress tends to become more complex. This is likely to become even more complex when each turn includes multiple choice opportunities across multiple phases, and each of these choice opportunities ends selectively. Since each choice opportunity requires a complex choice involving many options (including options derived from each option), even if turn instructions, etc., are not selective, this does not change the fact that a complex choice is required. Furthermore, multiple probability value information can be provided from various perspectives regarding such complex choices. Furthermore, the multiple probability value information can be limited to a portion or appropriately sorted using filter conditions, etc., before being provided. This improves the convenience of accessing specific probability value information (option information) through filter conditions, etc.

On the other hand, via the guessing service, information about the opponent's card object CO (element) that is not shared with the user, for example, the contents of a card object CO that is concealed from the user by being placed face down, is guessed by a trained artificial intelligence model (guessing model unit 24), and the guessed results are provided as guessed information. Therefore, the guessed information can help guess the concealed card object CO in an incomplete information game.

Furthermore, in a card game, a user uses a deck card selected by the user from among all card objects CO owned. For this reason, the card game is configured as a competitive game in which there is a difference in assets between players. In this case, it is likely that it will be more difficult to guess the hidden card object CO than when all element groups are used. Similarly, in a card game, the available card objects CO may change depending on the time of year, etc., making it more likely that it will be more difficult to guess the hidden card object CO than when they are fixed (unchanging). For this reason, while an experienced player may be able to guess the contents of a hidden card object CO by looking at it face down, for example, based on experience, such guessing is generally difficult, and is often particularly difficult for beginners. Even when it is difficult to guess the hidden card object CO, guessing can be aided by guessing information. Therefore, guessing information can help close the gap with experts (eliminate the gap). Furthermore, when changes in the available card objects CO are reflected in the guess information, the accuracy of the guess information can be improved compared to when similar changes are not reflected. Therefore, guessing information can further help eliminate the gap with experts.

Furthermore, multiple pieces of inferred information are provided by multiple trained AI models corresponding to time and rank. In card games, trends in the types of card objects (element groups) used can occur. When trends exist, AI models that reflect those trends are likely to output more accurate inferred information. Trends often occur over time. Similarly, the types of card objects CO incorporated into the card object groups used and their usage methods often vary depending on the rank (user skill). For this reason, by training a learning dataset containing performance data for a specified period, trends over that period can be reflected in the inferred information. Alternatively, by training a learning dataset containing performance data for each rank, usage trends for each rank can be reflected in the inferred information. This allows for more accurate inferred information. Furthermore, the type of inferred information (target period and rank) can be switched at the user's discretion. This allows for selection of inferred information according to the opponent (whether the opponent is a beginner, or which period's popularity the opponent's deck cards fall into, etc.) (in other words, the user's inferences about the opponent can be reflected in the inferred information), further improving the accuracy of the inferred information.

In the above embodiment, the first analytical model unit 23A and the second analytical model unit 23B of the game server 2 (or the processes of FIG. 12 executed by them, respectively) function as multiple types of processes of the present invention. Meanwhile, the model processing unit 35 of the user device 3 functions as the superiority/inferiority acquisition means and superiority/inferiority providing means of the present invention by executing the procedure of FIG. 12. Specifically, the model processing unit 35 functions as the superiority/inferiority acquisition means by executing step S102 of FIG. 12, and as the superiority/inferiority providing means by executing step S104. Furthermore, the model processing unit 35 of the user device 3 functions as the filter means or sort means of the present invention by executing the procedure of step S303 of FIG. 13.

Furthermore, the estimation model unit 24 of the game server 2 (or the process of FIG. 14 executed by it) functions as a predetermined process of the present invention. Meanwhile, the model processing unit 35 of the user device 3 functions as an information acquisition means and information provision means of the present invention by executing the procedure of FIG. 14. Specifically, the model processing unit 35 functions as an information acquisition means by executing step S402 of FIG. 14, and as an information provision means by executing step S404.

The present invention is not limited to the above-described embodiments and may be embodied in embodiments with appropriate modifications or alterations. Furthermore, the present invention may be embodied in embodiments obtained by appropriately combining various technical means included in the above-described embodiments and the following modified embodiments. In the above-described embodiments, superiority/inferiority information and speculation information are provided to players in a card game provided as a video game. However, the present invention is not limited to such embodiments. For example, superiority/inferiority information and speculation information may be provided to players in a game in real space, or to viewers watching the game. Specifically, for example, superiority/inferiority information or speculation information in a game in real space may be provided by adding it to a video (viewed video) of the game, or it may be provided via audio in the real space where the game is being played (which may include spectators as well as players).

In the above-described embodiment, each analytical model unit 23 (trained artificial intelligence model for analysis service), such as the first analytical model unit 23A, and each inference model unit 24 (trained artificial intelligence model for inference service), such as the first inference model unit 24A, are given different thoughts due to differences in learned data, such as the trained data for analysis AD or the trained data for inference FD. However, the present invention is not limited to this embodiment. Each analytical model unit 23 and each inference model unit 24 may also be given different thoughts due to differences in, for example, the algorithm of the learning program (including the reward, i.e., the predetermined result, in reinforcement learning such as Q-learning) and the algorithm of the inference program.

In the above-described embodiment, superiority/inferiority information is provided via the advice display section 60. However, the present invention is not limited to this embodiment. Superiority/inferiority information may be provided as appropriate. Figure 15 is an explanatory diagram illustrating an example of a modified form of the display mode of superiority/inferiority information. In the example of Figure 15, two modified forms are shown: (1) the first modified form, and (2) the second modified form. As shown in Figure 15 (1), in the first modified form, the probability values of each analytical model section 23 are provided via the target card image 90, the first analysis result section 91, and the second analysis result section 92.

The target card image 90 is an image showing a candidate card object CO as an option. The target card image 90 may be the card object CO itself, or a thumbnail-like image that replaces it. The first analysis result section 91 and the second analysis result section 92 are sections that indicate the type of analytical model section 23 (type of analysis result), such as the first analytical model section 23A (e.g., offensive AI) or the second analytical model section 23B (e.g., defensive AI), respectively. The first analysis result section 91 and the second analysis result section 92 can be formed as appropriate, but in the example of Figure 15, both are formed as rectangles surrounding the target card image 90, and are indicated by dashed lines and thick solid lines, respectively.

The first analysis result section 91 and the second analysis result section 92 each include a numerical information section 93 indicating the probability value (analysis result) of each analytical model section 23, such as "46.2." A summation information section 94 may also be provided. For example, in the first modified example of FIG. 15, three candidates are shown, with the middle candidate being provided with a summation information section 94. The summation information section 94 indicates the sum of the probability values of multiple artificial intelligence models for one candidate (option). While target card images 90 may be displayed for all candidates, in the example of FIG. 15, only a portion, such as the top three with the highest probability values, is displayed. Because probability values are output for all three candidates by multiple analytical model sections 23, the summation information section 94 may be displayed for all of them, but in the example of FIG. 15, only a portion (the middle target card image 90) is displayed. The display target for the summation information section 94 can be set as appropriate; as an example, it is set to the candidate whose sum exceeds the highest probability value. For example, the middle target card image 90 has numerical information sections 93 showing "25.2" and "32.0" in the first analysis result section 91 and second analysis result section 92, respectively, and the total information section 94 shows the sum of these values, "57.2." This total value is higher than the probability values of any of the other target card images 90. In such cases, the total information section 94 is provided.

Furthermore, as shown in (2) of Figure 15, in the second modified example, the probability value of each analytical model section 23 is provided via a target card image 90, a first gauge section 95, and a second gauge section 96. The target card image 90 is the same as in the first modified example. The first gauge section 95 and the second gauge section 96 are sections that respectively indicate each analytical model section 23 (type of analysis result) such as the first analytical model section 23A, and are configured to visually indicate the magnitude of the probability value. The first gauge section 95 and the second gauge section 96 can be formed as appropriate, but in the example of Figure 15, both are formed in the shape of a bar extending upward from a predetermined position depending on the magnitude of the probability value, and are shown in white and black, respectively.

In addition, a detailed information section 97 is provided in both the first gauge section 95 and the second gauge section 96. The detailed information section 97 may include various types of information, but in the example of FIG. 15, it includes information on the type of analytical model section 23 ("offensive AI" or "defensive AI"), such as the first analytical model section 23A (e.g., offensive AI) or the second analytical model section 23B (e.g., defensive AI), as well as information on the probability value ("46.1" or "32.0"). Although omitted in the example of FIG. 15, a total information section 94 may be provided as appropriate, and total value information may be displayed. Superiority/inferiority information may be provided as appropriate, and is not limited to the example of FIG. 5, and may be provided, for example, as in the modified example of FIG. 15.

In the above-described embodiment, the network system 1 includes a game server 2. However, the present invention is not limited to this embodiment. For example, in cases where an offline game that can be played without connecting to the network NT is provided, the game server 2 may be omitted, and the user device 3 may function alone as the game system of the present invention. In other words, various artificial intelligence models such as the analytical model units 23 may be provided in the user device 3. Therefore, for example, in the example of FIG. 12 or the example of FIG. 14, the processing (role) of the game server 2 may be performed by an artificial intelligence model (a logical device similar to the analytical model unit 23, etc.) provided in the user device 3 and called up by the model processing unit 35.

Alternatively, conversely, the game server 2 may perform all or part of the functions (various processes, etc.) of the user device 3. For example, learned data such as analytical learned data AD or inference learned data FD may be stored on the game server 2, and programs for artificial intelligence models such as the analysis program AP or the inference program FP may be provided on the user device 3. Furthermore, data necessary for implementing the present invention may be appropriately distributed and recorded on the game server 2 and the user device 3, such as portions of various learned data or portions of programs for artificial intelligence models. The game system of the present invention may then be realized through cooperation between the game server 2 and the user device 3. The same applies to the game program PG2 and the game data GD. In these cases, a combination of the user device 3 and the game server 2 (e.g., the network system 1), or the game server 2 alone (including when composed of multiple server devices), may function as the game system of the present invention. Furthermore, programs and control methods implemented in devices such as the network system 1, the user device 3, or the game server 2 may function as the game program and control method of the present invention.

Various aspects of the present invention derived from the above-described embodiments and variations are described below. In the following description, corresponding components shown in the accompanying drawings are noted in parentheses to facilitate understanding of each aspect of the present invention, but this does not mean that the present invention is limited to the illustrated forms.

The computer program of the present invention is configured to cause a computer (31) incorporated in a game system (3) that provides element information as information relating to at least one of a group of elements (CO) for each user in a game played by multiple users using the group of elements for each user to function as information acquisition means (35) that acquires the inferred information using a predetermined process (24) set to obtain inferred information for inferring the content of at least one element of the group of elements of other users provided as the game, in an incomplete information game of a type in which the content of at least one element of the group of elements provided as the game is not shared with the one user, and information provision means (35) that provides the inferred information as the element information.

According to the present invention, elements that are not shared with a single user, i.e., elements that are kept secret from a single user, are inferred through a predetermined process, and the inferred results are provided as inferred information. This makes it possible to aid in the inference of hidden elements in an incomplete information game using inferred information. Note that the game of the present invention may be a game in real space and does not necessarily have to be provided by a game system. Similarly, the predetermined process does not necessarily have to be provided by a game system. For example, the predetermined process may be executed by a system (operated by a different system) different from the game system.

Various elements included in imperfect information games may be used as element groups as appropriate. For example, when an imperfect information game such as mahjong or a card game is provided, elements such as other users' tiles or hand (cards in hand) may be used as element groups. Furthermore, like mahjong tiles in mahjong, each user's element group may be appropriately allocated from a fixed element group (mahjong tiles) that is common to all users. The same applies to card games such as Sevens or Poker. Alternatively, an imperfect information game may be played using a portion of an element group arbitrarily designated by the user from an element group associated with each user. For example, in one aspect of the computer program of the present invention, each user's element group may be formed from two or more elements selected by the user as the denominator of the element group from a denominator element group associated with each user.

An imperfect information game may be configured as appropriate. For example, in an imperfect information game, the available elements may be fixed or may vary depending on appropriate conditions such as time. Furthermore, the available elements may change as appropriate; for example, the number or type of available elements may change. That is, unavailable elements may arise or new elements may be added depending on a predetermined time period. When the available elements change, the change may or may not be reflected in the speculation information. Specifically, for example, in one aspect of the computer program of the present invention, the imperfect information game may be provided so that the available elements change at a predetermined time period. Furthermore, in this aspect, the change in the available elements may include at least one of a change in number and a change in type. Similarly, in an aspect in which the available elements change, the predetermined process may be configured to reflect the change in the available elements in the speculation information. When the available elements change depending on time, etc., it is likely that the hidden elements will be more difficult to guess than when the available elements are fixed. Even when guessing such hidden elements is difficult, the speculation information can assist in the guessing process. Furthermore, when changes in available elements are reflected in the inferred information, the accuracy of the inferred information can be improved compared to when similar changes are not reflected.

An imperfect information game may be configured as an appropriate game. For example, it may be configured as a cooperative game (including games played in teams) or a competitive game (including one-on-one, one-on-many (including cases where multiple users other than the player are independent opponents, such as in mahjong), and many-on-many). Similarly, it may be configured as a game played simply by one person. Furthermore, an imperfect information game may be a game played in real space or an electronic game. Therefore, the inferred information may be provided as appropriate. For example, in a game played in real space, the inferred information may be provided by adding it to a video of the game (such as displayed on a monitor or a wearable device), or it may be provided via audio in the real space where the game is played. Similarly, in an electronic game, the inferred information may be displayed on the game screen for that game, or it may be provided via audio. Furthermore, an electronic game may be configured as appropriate, for example, it may be configured to be provided via an appropriate object on a display device that displays a game screen including various objects. Specifically, in one aspect of the computer program of the present invention, the incomplete information game may be provided as a card game in which the one user and the other user play against each other according to predetermined rules via a plurality of card objects (CO) on a display device (MO) that displays the plurality of card objects (CO) that function as element groups for each user. Also, in this aspect, the display device may display a game screen (50) including the plurality of card objects, and the game system may provide the card game as a video game via the game screen.

Furthermore, in cases where a card game is provided as an incomplete information game, changes in usable elements (e.g., usability due to a rule change) may or may not be reflected in the guess information as described above, but if they are reflected in the guess information, such reflection may be achieved appropriately. For example, the unusable element may be reflected in the guess information by simply excluding the unusable element from the guess result after the guess result is obtained. Alternatively, the unusable element may be reflected in the guess information by adjusting the guess result again after the unusable element has been excluded (e.g., by reflecting the impact of the unusable element on other guesses). In either case, the unusable element may be reflected by excluding the unusable card object from the guess information. Specifically, for example, in a case where a card game is provided as an incomplete information game, changes to usable card objects in the card game include a change that makes a specific card object unusable due to the rule change, and the predetermined process may be set to reflect the unusable card object in the guess information by excluding the specific card object.

A card game may be played according to appropriate rules. For example, the card game may be played using fixed cards. Alternatively, the card game may be configured so that appropriate card objects are granted to users who meet certain granting conditions, and some of the card objects are owned by each user, and the game may be played using the card objects owned by each user. All of the card objects owned by each user may be used in play, or only an appropriate portion, such as card objects selected by each user, may be used in play. Specifically, for example, in a mode in which the card game is provided as an incomplete information game, the card game may be configured to grant card objects (CO) to users who meet certain granting conditions, and the denominator element group may be made up of the card objects (CO) granted to each user. In this case, compared to when the available elements are fixed, the candidates for the hidden elements change with each play, making it more difficult to guess. Even when guessing such hidden elements is difficult, guess information can help with the guess.

When a card game is provided as an incomplete information game, the card game may be configured to appropriately use multiple card objects (element groups). An appropriate portion of the multiple card objects may be the subject of speculation information. For example, in a mode in which the card game is provided as an incomplete information game, the card game is played using multiple card placement areas (CP) into which card objects to be used in a match with the other user should be placed, a hand area (57) into which hand cards (CO5) serving as candidate card objects to be placed in each card placement area should be placed, and a deck area (55) into which deck cards (CO6) serving as candidate card objects to be added to the hand should be placed, and the predetermined processing may be configured to obtain the content of at least one card object in the multiple card placement areas, the hand, and the deck from the multiple card objects of the other user as the speculation information.

The predetermined process may be configured as appropriate as long as it can obtain inferred information. For example, each process may be configured to output inferred information according to predetermined logic (including predetermined rules, calculation formulas, etc.). Alternatively, the predetermined process may be configured as an artificial intelligence model (so-called AI) generated to obtain inferred information. Furthermore, the predetermined process may be configured as a single process that obtains one type of inferred information, or as multiple processes that obtain multiple pieces of inferred information for one secret element. For example, in one aspect of the computer program of the present invention, the predetermined process may be configured as a trained artificial intelligence model generated to obtain the inferred information by machine learning a predetermined training dataset. Furthermore, in this aspect, the trained artificial intelligence model may include multiple trained artificial intelligence models (24A, 24B), each generated by machine learning multiple training datasets (FD) with different contents. In card games, trends may occur in the types of card objects (groups of elements) used. When trends occur, artificial intelligence models that reflect those trends are more likely to output more accurate inferred information. Trends often occur periodically. Similarly, there are often differences in the types of card objects incorporated into the group of card objects used and the way they are used depending on the rank (user skill). For this reason, by training a learning dataset containing performance data for a specified period, it is possible to reflect trends for each period in the estimated information. Alternatively, by training a learning dataset containing performance data for each rank, it is possible to reflect usage trends for each rank in the estimated information. This allows for more accurate estimations.

When the trained AI model includes multiple trained AI models, multiple pieces of inferred information obtained by the multiple trained AI models may be provided, or a portion of the multiple pieces of inferred information may be provided. When a portion of the multiple pieces of inferred information is provided, the portion may be identified appropriately, for example, based on the playing situation, or based on evaluation over a certain period of time (such as accuracy rate or user evaluation). Alternatively, the portion may be identified based on a user specification. For example, when the trained AI model includes multiple trained AI models, the inferred information may include multiple pieces of inferred information obtained using the multiple trained AI models, and the information providing means may provide the single user with inferred information corresponding to the trained AI model selected by the single user from the multiple pieces of inferred information.

On the other hand, the game system of the present invention is a game system incorporating a computer that provides element information as information relating to at least one of the element groups (CO) for each user in a game played by multiple users using the element groups for each user, and the computer functions as information acquisition means (35) that acquires the inferred information using a predetermined process (24) set to obtain inferred information for inferring the content of at least one element in an incomplete information game provided as the game, in which the content of at least one of the element groups of other users is not shared with a single user, and as information provision means (35) that provides the inferred information as the element information.

The control method of the present invention also involves causing a computer (31) incorporated in a game system (3) that provides element information as information relating to at least one of a group of elements (CO) for each user in a game played by multiple users using the group of elements for each user to execute the following steps in an imperfect information game in which the content of at least one of the group of elements of other users provided as the game is not shared with a single user: acquiring the speculated information using a predetermined process (24) configured to obtain speculated information for inferring the content of the at least one element; and providing the speculated information as the element information.

(Reference example)
Non-Patent Document 1 discloses a technology that adds information such as "winning rate,""potentialmoves," and "reading lines" that are simulated for each move by AI (artificial intelligence) to a live broadcast screen that captures a shot of the shogi board from above during a game. However, this information (simulation results) is merely the result of a simulation based on one type of AI.

In general, AI tends to have a way of thinking (or perhaps even a personality) that is influenced by the training datasets that form the basis of machine learning. For example, differences in the content of the training dataset or the machine learning learning methods (algorithms) often result in AI developing its own unique way of thinking. For this reason, if the same move in the shogi game described in Non-Patent Document 1 were simulated by multiple types of AI, the simulation results from one AI may differ from those from other AIs. If only simulation results based on one type of AI are provided, the simulation results may be biased, or the simulation may be performed for a purpose other than the move's intention. As a result, the reliability of the simulation results may decrease, and even if this is not the case, situations may arise where AIs with different ways of thinking, in other words, simulation results from various perspectives, are widely required. Simulation results may be calculated using various logic other than AI, and similar situations may arise in such cases.

The present invention aims to provide a computer program or the like that can provide information on the relative merits of each option from various perspectives in a game that includes a selection opportunity in which some of multiple options must be selected.

The computer program of the reference example is configured to cause a computer (31) incorporated in a game system (3) that provides a user with superiority/inferiority information regarding the relative merits of the influence of each option in a game including a selection opportunity to select some of a plurality of options before the user selects one of the options, to function as superiority/inferiority acquisition means (35) that acquires multiple pieces of superiority/inferiority information for each of the at least one option using multiple types of processing (23A, 23B) that outputs the superiority/inferiority information for each of the at least one option, and superiority/inferiority provision means (35) that provides the multiple pieces of superiority/inferiority information for each option before the user selects the some of the options.

In the reference example, multiple pieces of superiority/inferiority information are output for each option using multiple types of processing, and this information is provided as superiority/inferiority information before some of the options are selected. This makes it possible to provide superiority/inferiority information for each option from various perspectives in a game that includes a selection opportunity where some of the options must be selected. Note that the game in the reference example may be a game that takes place in real space, and does not necessarily have to be provided by a game system.

Superiority information may be provided in any game that includes a choice opportunity. For example, superiority information may be provided in cooperative games (including those played in teams) that include a choice opportunity, or competitive games (including one-on-one, one-on-many (including cases such as mahjong where multiple users other than the player are independent opponents), and many-on-many). Similarly, superiority information may be provided in games that are simply played alone. Furthermore, various influences of each option in each game may be output as superiority information. For example, in a game that has a concept of victory, the influence on victory may be output as superiority information. Alternatively, the influence on a part of the progress, such as a specific development provided within the game (including specific missions and the acquisition of bonus points, etc.), may be output as superiority information. Specifically, for example, in one embodiment of the computer program of the reference example, a competitive game in which the user competes against an opponent is provided as the game, and the multiple types of processing may be configured to output information regarding the superiority or inferiority of the influence of each option on victory in the competitive game as the multiple pieces of superiority information. In this case, information regarding the influence on victory in the competitive game can be provided for each option from various perspectives.

The superiority/inferiority information may be any appropriate information regarding superiority/inferiority. For example, it may be symbolic information such as A, B, or C that distinguishes the degree of superiority/inferiority, or information that intuitively expresses superiority/inferiority, such as "looks good" or "looks bad." Alternatively, the superiority/inferiority information may be information that quantifies superiority/inferiority. When superiority/inferiority is quantified, the numerical value may indicate the likelihood of progressing advantageously (least likely to be at a disadvantage) within a predetermined range (including any appropriate range, such as a predetermined number of selection opportunities, a predetermined time, or a predetermined development), or may of course indicate the likelihood of ultimately winning the battle. For example, in a mode in which a competitive game is provided, the multiple types of processing may be configured to output, as the multiple pieces of superiority/inferiority information, any numerical information that indicates, by the magnitude of the numerical value, the likelihood of progressing advantageously in the competitive game.

Competitive games may be configured as appropriate. For example, a competitive game may be configured so that the user and opponent make choices simultaneously (in parallel), or so that turns are given alternately, each including a choice opportunity. Even when turns are given alternately, the choice of the player outside of their turn (which may function as a reference choice opportunity) may not be completely eliminated in each turn, for example, by requiring the opponent to make a choice in response to a choice made during the user's turn. Furthermore, when turns are given alternately, the turn change may be automatically performed based on time, the number of instructions, etc., or may be performed as one of the choices in each turn. When a turn changes due to a choice, that choice may function as a factor in superiority/inferiority information, or may be excluded from the factor in superiority/inferiority information. Furthermore, each turn may be configured as appropriate, for example, it may include only one choice opportunity, or it may include multiple choice opportunities. The multiple choice opportunities may be distinguished as appropriate through phases, etc., and the choices in each choice opportunity may all be the same, or at least some (including all) of the choices may differ. Furthermore, the end of each phase may be automatic or optional. If the end of each phase is optional, that choice may or may not be subject to superiority information.

For example, in a mode in which a competitive game is provided, the competitive game may be configured so that turns including the selection opportunities are alternately provided between the user and the opponent, and the plurality of options may include an option corresponding to a turn end instruction to end the user's turn. Furthermore, in this mode, each turn includes multiple phases each providing multiple selection opportunities, and the selection opportunities in each phase are configured to provide two or more options as the plurality of options, each including different options between the selection opportunities in each phase, and the two or more options in each selection opportunity may each include an option corresponding to a phase end instruction to end the phase. When a turn change functions as one of the options, the impact of a choice made at a selection opportunity on progress tends to be more complex. This is likely to become even more complex when each turn includes multiple selection opportunities and each of these selection opportunities is selectively terminated. When options corresponding to a turn end instruction or the like are the subject of superiority/inferiority information, multiple pieces of superiority/inferiority information can be provided from various perspectives regarding complex choices.

Games that include selection opportunities may be games played in real space or electronic games. Therefore, superiority/inferiority information may be provided as appropriate. For example, in games played in real space, superiority/inferiority information may be provided by adding it to a video of the game (such as displayed on a monitor or a wearable device), or may be provided via audio in the real space where the game is played. Similarly, in electronic games, superiority/inferiority information may be displayed on the game screen for that game, or may be provided via audio. Furthermore, electronic games may be configured as appropriate, and may be configured to be provided via appropriate objects on a display device that displays a game screen including various objects. Specifically, in a mode in which a competitive game is provided, the competitive game may be provided as a game in which a player competes against an opponent according to predetermined rules via a plurality of objects (CO) on a display device (MO) that displays the plurality of objects (CO).

When a competitive game is provided via multiple objects, the multiple objects may be used as appropriate. For example, when an electronic shogi or chess game is provided, electronic pieces may be used as the multiple objects. Or, when an electronic mahjong or card game is provided, electronic mahjong tiles or cards may be used as the multiple objects. Also, in a game that includes various characters (including not only human-like beings but also various objects such as cars and animals), the multiple characters may function as the multiple objects. These objects may be used as appropriate in accordance with the rules of the game. Similarly, various elements set in accordance with the rules of the game may function as multiple options. For example, each object (e.g., a character) may or may not function as an option.

Specifically, for example, in a mode in which a competitive game is provided, the competitive game is provided as a card game in which a plurality of card objects (CO) are used as the plurality of objects, and the card game is played using a plurality of card placement areas (CP) in which the card objects to be used in the match against the opponent should be placed, a hand area (57) in which hand cards (CO5) as candidate card objects to be placed in each card placement area should be placed, and a deck area (55) in which deck cards (CO6) as candidate card objects to be added to the hand should be placed, and the plurality of options may include options corresponding to at least one of the plurality of card placement areas, the hand cards to be placed in each card placement area, and adding cards from the deck to the hand. In this case, it is possible to provide multiple pieces of superiority/inferiority information from various perspectives in a game that requires complex choices including many options (including further options derived from each option).

The multiple types of processes may be configured as appropriate as long as they are able to output multiple pieces of superiority/inferiority information. For example, each process may be configured to output superiority/inferiority information according to a predetermined logic (including predetermined rules, calculation formulas, etc.). Alternatively, at least one of the multiple types of processes may be configured as an artificial intelligence model (so-called AI) generated to output superiority/inferiority information. Specifically, in one aspect of the computer program of the reference example, at least one of the multiple types of processes may be configured as a trained artificial intelligence model generated by machine learning a predetermined training dataset into a pre-trained model to output superiority/inferiority information regarding a predetermined result in the game as the superiority/inferiority information. Furthermore, in this aspect, the multiple types of processes may all be configured as the trained artificial intelligence model, and may be configured to output the multiple pieces of superiority/inferiority information based on differences in at least one of the predetermined result, the training dataset, and the algorithm of the pre-trained model used to train the training dataset.

Multiple pieces of superiority/inferiority information may be provided as appropriate. For example, the superiority/inferiority information may be provided in a predetermined order for each option, or if quantified, may be provided in descending order, or other order according to numerical magnitude. The order in which each piece of superiority/inferiority information is provided may be fixed or variable. For example, multiple pieces of superiority/inferiority information may be provided in an order specified by a user. Alternatively, multiple pieces of superiority/inferiority information may be provided after being limited to a portion thereof based on predetermined filter conditions. The filter conditions may be fixed or variable. For example, the filter conditions may be specified by a user. Specifically, in one aspect of the computer program of the reference example, the superiority/inferiority providing means may be provided with at least one of filter means (35) that limits the multiple pieces of superiority/inferiority information to a portion that satisfies predetermined filter conditions, and sort means (35) that rearranges the multiple pieces of superiority/inferiority information according to predetermined sort conditions. In this case, the multiple pieces of superiority/inferiority information are provided after being limited to a portion thereof based on filter conditions, etc., or rearranged as appropriate. This improves the convenience of accessing specific superiority/inferiority information through filter conditions, etc.

On the other hand, the game system of the reference example is a game system (3) incorporating a computer (31) that provides a user with superiority/inferiority information regarding the relative merits of the influence of each option for at least one of a plurality of options in a game including a selection opportunity in which some of the options must be selected, before the user selects one of the options; the computer functions as superiority/inferiority acquisition means (35) that acquires multiple pieces of superiority/inferiority information for each of the at least one option using multiple types of processing that outputs the superiority/inferiority information for each of the at least one option, and as superiority/inferiority provision means (35) that provides the multiple pieces of superiority/inferiority information for each option before the user selects one of the options.

The control method of the reference example also includes causing a computer (31) incorporated in a game system (3) that provides a user with superiority/inferiority information regarding the relative merits of the influence of each option for at least one of a plurality of options in a game including a selection opportunity in which some of the options must be selected, before the user selects one of the options, to execute the following steps: acquiring multiple pieces of superiority/inferiority information for each option of the at least one option using multiple types of processing that outputs the superiority/inferiority information for each of the at least one option; and providing the multiple pieces of superiority/inferiority information for each option before the user selects one of the options.

3. User Device (Game System)
31 Control unit (computer)
35 Model processing unit (superiority acquisition means, superiority provision means, filter means, sorting means, information acquisition means, information provision means)
50 Battle screen (game screen)
55 Main Deck Placement Area (Deck Area)
57 Hand placement area (hand area)
CO Card Object (Object)
CP card storage area (card placement area)
MO Display device CO5 Hand CO6 Deck PG2 Game program (computer program)

Claims (15)

A computer incorporated in a game system that provides element information as information regarding at least one element of a group of elements for each user in a game played by a plurality of users by placing each element in one of a plurality of placement locations on a game field using a group of elements for each user,
In an incomplete information game in which the content of at least one of a group of elements of other users is not shared with a single user, the game is provided as the game, and the information acquisition means acquires inferred information using a predetermined process set to obtain inferred information for inferring the content of the at least one element based on information on the game field including information on individual placement locations of each element as information on the placement locations of each element; and information provision means provides the inferred information as the element information.
A computer program configured to function as The computer program of claim 1, wherein each user's element group is formed by two or more elements selected by each user from a denominator element group associated with each user as the denominator of the element group. The computer program described in claim 2, wherein the imperfect information game is provided so that available elements change over a given period of time. The computer program of claim 3, wherein the changes to the available elements include at least one of a change in number and a change in type. The computer program of claim 3, wherein the predetermined processing is configured to reflect changes in the available elements in the estimated information. The computer program of claim 5, wherein the incomplete information game is provided as a card game in which the one user and the other user play against each other according to predetermined rules via a plurality of card objects on a display device that displays the plurality of card objects functioning as element groups for each user. The change in the card objects that can be used in the card game includes a change that makes a specific card object unusable in accordance with the change in the rules,
7. The computer program according to claim 6, wherein the predetermined process is set so that the unusability of the specific card object is reflected in the guess information by excluding the specific card object. the card game is configured to award a card object to a user who satisfies a predetermined awarding condition;
The computer program according to claim 6 , wherein the denominator element group is constituted by card objects assigned to each user. the card game is played using a plurality of card placement areas in which card objects to be used in a match against the other user are to be placed, a hand area in which a hand of cards as candidate card objects to be placed in each card placement area is to be placed, and a deck area in which a deck of cards as candidate card objects to be added to the hand is to be placed;
9. The computer program according to claim 8, wherein the predetermined processing is set so that the content of at least one card object of the other user's card placement areas, hand, and deck is obtained as the speculated information. the display device displays a game screen including the plurality of card objects;
10. The computer program according to claim 6, wherein the game system provides the card game as a video game via the game screen. The computer program described in any one of claims 1 to 9, wherein the predetermined processing is configured as a trained artificial intelligence model generated to obtain the inferred information by machine learning a predetermined training dataset. The computer program of claim 11, wherein the trained artificial intelligence model includes multiple trained artificial intelligence models each generated by machine learning multiple training datasets with different contents. the inferred information includes a plurality of pieces of inferred information obtained by using the plurality of trained artificial intelligence models,
The computer program according to claim 12 , wherein the information providing means provides the one user with inferred information corresponding to a trained artificial intelligence model selected by the one user from the plurality of inferred information. 1. A game system incorporating a computer that provides element information as information relating to at least one of a group of elements for each user in a game played by a plurality of users by placing each element in one of a plurality of placement locations on a game field, the game system comprising:
The computer
In an incomplete information game in which the content of at least one of a group of elements of other users is not shared with a single user, the game is provided as the game, and the information acquisition means acquires inferred information using a predetermined process set to obtain inferred information for inferring the content of the at least one element based on information on the game field including information on individual placement locations of each element as information on the placement locations of each element; and information provision means provides the inferred information as the element information.
It functions as a game system. A computer incorporated in a game system that provides element information as information regarding at least one of a group of elements for each user in a game played by a plurality of users by placing each element at one of a plurality of placement locations on a game field using a group of elements for each user,
In an incomplete information game provided as the game, in which the content of at least one of a group of elements of other users is not shared with a single user, a step of acquiring inferred information using a predetermined process set to obtain inferred information for inferring the content of the at least one element based on information on the game field including information on individual placement locations of each element as information on placement locations of each element;
providing the inferred information as the element information;
A control method for executing the above.