JP7801282B2 - Information processing device, information processing method, and information processing program - Google Patents

Description

The present invention relates to an information processing device, an information processing method, and an information processing program.

Technology has traditionally existed for presenting answers to user questions. Technology also exists for delivering advertisements related to user questions.

Patent No. 6849982 Japanese Patent Application Laid-Open No. 2023-061663

In conventional technology, it has sometimes been impossible to deliver appropriate advertisements to users who are presented with answers to their input. For example, in conventional technology, advertisements are delivered to the user who asked the question along with the answer to the user's question, but the delivery of advertisements to users who view the answer posted by the user on a social networking service (SNS) or the like is not sometimes taken into consideration.

To solve the above-mentioned problems and achieve the objective, the information processing device is characterized by having an estimation unit that estimates user images of other users who will provide answers to inputs made by the user based on information about the user; a selection unit that selects advertisements based on the user images estimated by the estimation unit; a generation unit that generates sentences for a generative model so that the advertisements selected by the selection unit are included in the answers; and an output unit that outputs the sentences generated by the generation unit to the generative model.

This invention makes it possible to deliver advertisements appropriate to the user to whom the answer to the user's input is presented.

FIG. 1 is a diagram illustrating an outline of a process performed by an information processing apparatus according to an embodiment. FIG. 2 is a block diagram illustrating an example of the configuration of the information processing apparatus according to the embodiment. FIG. 3 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 4 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 5 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 6 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 7 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 8 is a diagram illustrating an example of processing performed by the information processing apparatus according to the embodiment. FIG. 9 is a flowchart illustrating an example of processing by the information processing apparatus according to the embodiment. FIG. 10 is a diagram illustrating an example of a computer that executes an information processing program.

Embodiments of an information processing device, information processing method, and information processing program according to the present application will be described in detail below with reference to the drawings. Note that the present invention is not limited to these embodiments. Furthermore, in the drawings, identical parts are denoted by the same reference numerals, and duplicate explanations will be omitted.

[Introduction]
First, an overview of the information processing device 100 will be described using FIG. 1. FIG. 1 is a diagram illustrating an overview of the processing performed by the information processing device 100. Conventionally, generation models such as large-scale language models and image generation models have been used to generate answers to inputs such as questions and requests from users. However, with conventional technology, it has sometimes been impossible to deliver appropriate advertisements to users who are the recipients of answers to inputs made by the users.

For example, when an influencer or creator creates an image, they may make it publicly available to their followers on social media. In such a situation, conventional technology delivers an advertisement to the user who asked the question along with the answer to the user's question, but does not sometimes take into consideration the delivery of advertisements to other users who view the answer posted by the user on social media, etc.

The information processing device 100 according to this embodiment therefore performs the following processing. First, the information processing device 100 estimates the user image of other users to whom an answer from a user who inputs a question, request, etc. via an application (FIG. 1 (1)) will be presented, based on information about the user who made the input. Next, the information processing device 100 selects an advertisement to be delivered (FIG. 1 (2)) based on the estimated user image. Next, as shown in FIG. 1 (3), the information processing device 100 generates a sentence for the generative model: "The user image of other users to whom the user who made the input will present an answer is ____. Please reply to the user who made the input, including the advertisement. The input content is as follows: _____"" so that the selected advertisement will be included in the answer from the generative model.

The information processing device 100 then outputs the generated text to the generative model. This allows the information processing device 100 to generate and output text to the generative model that includes an advertisement corresponding to the user's persona, to which the answer to the user's input will be presented, thereby enabling the delivery of appropriate advertisements tailored to the user, to which the answer to the user's input will be presented.

In other words, the information processing device 100 generates and outputs text for a generative model so that the answer to a user's input includes an advertisement that matches the persona of the other user to whom the user's answer is presented. This makes it possible to provide the information sought by the user who provided the input, such as an influencer or creator, as an answer, and to deliver advertisements that are sought by the other users to whom the answer to the user's input is presented, or that will pose little psychological resistance to them.

[Configuration of information processing device]
Next, the configuration of the information processing device 100 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the configuration of the information processing device 100 according to the embodiment. As shown in Fig. 2, the information processing device 100 includes a communication unit 110, a control unit 120, and a storage unit 130. Note that these units may be held in a distributed manner by multiple devices. The processing of these units will be described below.

The communication unit 110 is implemented using a NIC (Network Interface Card) or the like, and enables communication between the control unit 120 and external devices via telecommunications lines such as a LAN (Local Area Network) or the Internet. For example, the communication unit 110 enables communication between the external devices and the control unit 120.

The memory unit 130 is realized by semiconductor memory elements such as RAM (Random Access Memory) or flash memory, or storage devices such as hard disks or optical disks. Information stored in the memory unit 130 includes, for example, information about users, user profiles, information about advertisements, algorithms, learning data for machine learning, and trained models. Here, information about users includes information such as user ID, user attributes (age, gender, birthplace, place of residence, occupation, characteristic search keywords, etc.), personality (characteristics, values, concerns, etc.), lifestyle (daily habits, consumption patterns, friendships, etc.), search history, browsing history, location information, conversational tendencies, conversation content, sentences recognized by voice recognition, social media posts, context, and follower information.

Here, a user profile includes user attributes and is information indicating the user's characteristics inferred from information about the user; it is a so-called persona. A user profile may also include summaries of conversations involving the target user over a specified period of time in the past. For example, the user profile for User A is "Gender: Male, Age: 20, Eats out with friends 3-5 times a month, Dissatisfied with the hassle of finding restaurants and making reservations, Currently looking for a place to eat with User B." Here, the user profile may be stored in the form of a vector. Furthermore, a user profile stored in the form of a vector is updated as needed based on information about the user (e.g., the content of the conversation, etc.).

Characteristic search keywords are search keywords of a specific user that are narrowed down based on information such as the number of searches, frequency of searches, and date and time of searches, such as the number of searches, frequency of searches, and recent search dates and times, from the user's search history. Furthermore, user attributes, personality, and lifestyle may be registered in advance, or may be inferred from text entered on the application in the past.

Conversation content includes information entered on applications with communication functions. For example, user A's conversation content includes information entered in a chat room in which user A and users who follow user A on SNS "Y" are members. SNS posts include images (videos and still images), audio, text, date and time, location information, etc. posted by the user to the SNS.

Context is information that is inferred from the context of a conversation, such as the flow of a conversation. As an example of context, if User A contacts a chat application asking, "Have you decided on a restaurant for the weekend celebration?" and the next day User B contacts the chat application saying, "We're having yakiniku for the weekend," it can be determined from the flow of the conversation that "that" refers to the restaurant for the celebration. Note that the subject of context inference is not limited to the content of the conversation, but can include all information about the user, such as search history, browsing history, and location information. Furthermore, sentences and context such as the content of the conversation may be information that has already been analyzed using natural language processing. Furthermore, information related to advertising includes information such as the ad ID, ad type, budget, period, and targeting information (SNS destination, user attributes, keywords, conversation trends, context, etc.).

Follower information is information about other users who follow a certain user. For example, follower information is attribute information about other users who follow a certain user. Here, follower information may be information obtained by aggregating and analyzing follower attribute information. For example, follower information for user A may be "Gender: 40% Male, 60% Female, Age: 60% Teens, 30% Twenties." In the above example, follower information includes gender and age, but follower information also includes information necessary to estimate the follower's user profile, such as the follower's hobbies and preferences and information on the date and time of access to SNS. Note that the information stored in the memory unit 130 is not limited to the above example.

The control unit 120 is implemented using a CPU (Central Processing Unit), NP (Network Processor), FPGA (Field Programmable Gate Array), etc., and executes processing programs stored in memory. As shown in FIG. 2, the control unit 120 has an estimation unit 121, a selection unit 122, a generation unit 123, and an output unit 124. Each unit of the control unit 120 will be described below.

The estimation unit 121 estimates the image of other users who will provide answers to inputs made by the user, based on information about the user. For example, the estimation unit 121 estimates the user image using one or more of the following information about the user: user attributes, search history, browsing history, location information, conversation content, conversation trends, SNS posts, context, and follower information.

More specifically, the estimation unit 121 estimates the user image of other users to whom user A submits answers as "interested in music" based on user A's attributes of "gender: male, age: 20s, occupation: singer" as information about user A. Note that the user image estimated by the estimation unit 121 is updated as needed in accordance with information about the user.

Furthermore, since the information about user A includes many posts on SNS made by user A that are related to outdoor activities and camping, the estimation unit 121 estimates that the user profile of other users to whom user A provides replies is "interested in outdoor activities and camping."

Furthermore, the estimation unit 121 estimates that the user profile of other users to whom user A provides answers is "interested in interior design and cooking" based on the fact that conversations held in a chat room to which user A belongs in an application with communication functions contain many topics related to interior design and cooking as information about user A.

Furthermore, the estimation unit 121 estimates, as information about user A, that the user profile of other users to whom user A will provide answers is "gender: male, age: 40s, hobby: cooking," based on the fact that, in an application with communication functions, many of the users participating in the chat room to which user A belongs have the attributes "gender: male, age: 40s, hobby: cooking."

The estimation unit 121 can also analyze the aggregated attribute information of user A's followers and estimate, for example, attribute information with a high percentage as the follower's user profile. For example, based on the follower attributes of "gender: 'male: 30%', 'female: 70%', age: 'teens: 30%', 'twenties: 40%', 'twenties: 20%'" as information about user A, the estimation unit 121 estimates the user profile of other users to whom user A will submit answers as "gender: female, age: 20s."

The estimation unit 121 also estimates a user image from information about the user using a model that has learned the relationship between information about the user and the user image. For example, the estimation unit 121 inputs information about the user into a learning model that takes information about the user as input and outputs user images of other users to whom the user submits answers, and estimates the output result as the user image of other users to whom the user submits answers. For example, this model is located in a server that processes information and was created independently by a business operator (Yahoo). It is desirable to keep input information, such as personal information, confidential by conducting learning in a way that prevents the input information from being used as a new answer.

The selection unit 122 selects advertisements based on the user image estimated by the estimation unit 121. For example, the selection unit 122 selects advertisements that target the user image estimated by the estimation unit 121 for ad delivery. For example, the selection unit 122 compares the user image estimated by the estimation unit 121 with the targeting information of the advertisement and selects advertisements with a high degree of match between the user image and the targeting information of the advertisement. At this time, the selection unit 122 may select the advertisement with the highest degree of match, or may select a predetermined number of advertisements with the highest degree of match.

More specifically, the selection unit 122 compares the user profile information of the other user to whom user A submits an answer, estimated by the estimation unit 121, "Gender: Male, Age: 20s, Context: Likes Store A," with the targeting information of each advertisement, and selects Store A with the highest match rate (targeting information "Male, 20s, Store A") as the advertisement.

The generation unit 123 generates sentences for the generative model so that the advertisement selected by the selection unit 122 is included in the answer. For example, the generation unit 123 generates sentences for the generative model that include a sentence by the user, a user image, and an advertisement.

More specifically, the generation unit 123 generates a sentence for the generative model such as, "The user image of other users to whom User A will provide an answer is '____'. Please include the selected Ad A in your answer to User A. The input content is as follows: '____'."

As another example, the generation unit 123 generates a sentence for the generation model such as, "Please answer the following question based on the user image of the other user to whom user A will provide the answer. Please also include the following advertisement text in your answer. The input from user A is as follows: "~~~~~~" The user image of the other user to whom user A will provide the answer is as follows: "~~~~~~" The advertisement is as follows: "~~~~~~"."

The output unit 124 outputs the sentences generated by the generation unit 123 to a generative large-scale language model. For example, the output unit 124 outputs the sentences generated by the generation unit 123 to a trained large-scale language model that has been independently created by a business (Yahoo). It is desirable to keep the input information confidential, such as personal information, by training it so that it will not be used as a new answer.

[Processing performed by information processing device]
Next, an example of processing performed by the information processing device 100 according to the embodiment will be described with reference to Figures 3 to 8. Figures 3 to 8 are diagrams illustrating an example of processing performed by the information processing device 100 according to the embodiment.

First, as shown in FIG. 3, the information processing device 100 receives input from user A on an application, such as "Show me a map of the area around Tokyo Station." Note that while this example shows an example in which input from the user is received on an application used for communication, the information processing device 100 can receive input from the user in any format. For example, the information processing device 100 may receive input from the user in the form of voice recognition.

The information processing device 100 then estimates the user profile of other users to whom User A, who has received the input, will submit answers. Figure 4 (1) shows posts made by User A to an SNS for sharing images, many of which are related to fashion and camping. From the content of the SNS posts by User A shown in Figure 4 (1) as user information about User A, the estimation unit 121 estimates the user profile of other users to whom User A will submit answers as "outdoorsy, hobbies: fashion, camping," as shown in Figure 4 (2).

Here, the estimation unit 121 can estimate the user image of other users to whom a user submits posts for each SNS. For example, Figure 5 (1) shows posts made by user A on a text-sharing SNS, many of which are related to movies and novels. In this example, the estimation unit 121 estimates the user image of other users to whom user A submits replies as "indoor type, hobbies: movies, reading" as shown in Figure 5 (2) from the content of the SNS posts by user A shown in Figure 5 (1) as user information about user A.

As another example, the estimation unit 121 estimates the user profile of other users who participate in a talk group to which a certain user belongs in an application with a communication function. Figure 6 (1) shows the content of a conversation that took place in a talk room to which users A, B, and C belong in an application with a communication function, with conversations such as "User B: My hobby is cooking. Do you have any favorite foods, A?", "User C: I'm curious", and "User A: I've been hooked on chicken cutlet lately. It's delicious, so please try it if you have the chance."

In this example, the estimation unit 121 estimates the user profile of the other user to whom user A provides an answer as "Hobby: Cooking" as shown in Figure 6 (2) from the content of the conversation in the chat room between user A, user B, and user C shown in Figure 6 (1) as information about user A.

The selection unit 122 then selects advertisements that target the user profile estimated by the estimation unit 121 for ad delivery. More specifically, the selection unit 122 compares the information about the user profile of user A estimated by the estimation unit 121, "outdoorsy, hobbies: fashion, camping," with the targeting information of each advertisement, and selects an advertisement with a high degree of match between the user profile and the targeting information of the advertisement (for example, outdoor brand store A, "targeting information: fashion, camping").

Next, as shown in FIG. 7, the generation unit 123 generates a sentence including the question "Show me a map of the area around Tokyo Station" from user A whose input has been accepted, information about user A's user profile "outdoorsy, hobbies: fashion, camping" estimated by the estimation unit 121, and targeting information for advertisement "Store A" selected by the selection unit 122, such as "The user profile of other users to whom user A will provide an answer is 'outdoorsy, hobbies: fashion, camping'. Please answer to user A including an advertisement for Store A. The question content is as follows: 'Show me a map of the area around Tokyo Station'" as shown in FIG. 7 (2).

The output unit 124 then outputs the sentence generated by the generation unit 123 to the generative model. For example, the output unit 124 outputs the sentence to the large-scale language model. As a result, the large-scale language model responds by including an advertisement for the selected store A, such as "The map of the area around Tokyo Station is as follows. [Image of map showing store A]", as shown in FIG. 8. Note that if multiple user images are estimated, the information processing device 100 can perform the above-described processing for each user image and output multiple answers from a generative model such as a large-scale language model.

In this way, the information processing device 100 generates and outputs text for the generative model so that the user's response includes an advertisement that corresponds to the persona of the other user who submits the answer, thereby enabling the delivery of appropriate advertisements tailored to the user to whom the answer to the user's input is presented. In other words, it is possible to provide the information the user is looking for as an answer, while also delivering advertisements that the other users who submit the answer are looking for or that the user finds less psychologically resistant to.

〔flowchart〕
Next, the flow of processing by the information processing device 100 will be described with reference to Fig. 9. Fig. 9 is a flowchart showing an example of the flow of processing according to this embodiment.

First, the information processing device 100 receives advertising information (step S101). For example, the information processing device 100 receives information such as an advertising ID, advertising type, budget, period, and targeting information as advertising information, and stores the information in the storage unit 130.

Next, the information processing device 100 accepts input such as a question or request from the user (step S102). For example, the information processing device 100 accepts a question entered by the user on an application and stores it in the storage unit 130.

Next, the estimation unit 121 estimates a user image of a user to whom a response to an input made by the user will be presented, based on the information about the user (step S103). For example, the estimation unit 121 estimates a user image of a user to whom a response to an input made by the user will be presented, using one or more pieces of information about the user stored in the storage unit 130 (user attributes, search history, browsing history, location information, conversation content, conversation trends, SNS posts, context, and follower information).

Next, the selection unit 122 selects an advertisement based on the estimated user image (step S14). For example, the selection unit 122 selects an advertisement that targets the estimated user image for ad delivery.

Next, the generation unit 123 generates a sentence for the generative model so that the advertisement selected by the selection unit 122 is included in the answer (step S105). For example, the generation unit 123 generates a sentence for the generative model that includes a sentence by the user, a user image, and an advertisement.

Then, the output unit 124 outputs the sentences generated by the generation unit 123 to the generative model (step S106). For example, the output unit 124 outputs the sentences generated by the generation unit 123 to a trained large-scale language model independently created by a business operator (Yahoo).

〔effect〕
As described above, the information processing device 100 according to the embodiment includes an estimation unit 121 that estimates the user image of other users who will provide answers to inputs made by the user based on information about the user, a selection unit 122 that selects advertisements based on the user image estimated by the estimation unit 121, a generation unit 123 that generates sentences for a generation model so that the advertisements selected by the selection unit 122 are included in the answers, and an output unit 124 that outputs the sentences generated by the generation unit 123 to the generation model.

As a result, the information processing device 100 estimates the personas of other users to whom the user submits answers from the attributes of the user who accepted the input and from SNS posts, etc., selects advertisements corresponding to the estimated personas, and generates and outputs text for the generative model so that the advertisements are included in the input answers, enabling the user who accepted the input to deliver appropriate advertisements to other users to whom the user submits answers.

In addition, the information processing device 100 generates and outputs text for a generative model so that answers to user input include advertisements that correspond to the personas of other users who submit answers from the user. This makes it possible to provide information that is desired by the user who provided the input, such as an influencer or creator, as an answer, while also delivering advertisements that are desired by other users who submit answers from the user who provided the input, or that will pose little psychological resistance to the user.

The estimation unit 121 of the information processing device 100 according to the embodiment estimates a user image using one or more of the following information about the user: user attributes, search history, browsing history, location information, conversation content, conversation trends, SNS posts, context, and follower information.

As a result, the information processing device 100 uses information about the user who accepted the input, such as user attributes, search history, browsing history, location information, conversation content, conversation trends, SNS posts, context, and follower information, to estimate the personas of other users to whom the user who made the input will submit answers, select advertisements corresponding to the personas, and generate and output text for the generative model so that the advertisements are included in the answers to the input, enabling the user who accepted the input to deliver appropriate advertisements to other users to whom answers will be submitted.

The estimation unit 121 of the information processing device 100 according to the embodiment estimates a user image from information about the user using a model that has learned the relationship between information about the user and the user image. As a result, the information processing device 100 uses the model to estimate the persona of the other user to whom the answer will be presented based on the attributes of the user who accepted the question, selects an advertisement corresponding to the persona, and generates and outputs text for the generative model so that the advertisement is included in the answer to the input, enabling the user who accepted the input to deliver appropriate advertisements to other users who will submit answers.

The selection unit 122 of the information processing device 100 according to the embodiment selects advertisements that target the user image estimated by the estimation unit 121 for ad delivery. As a result, the information processing device 100 selects advertisements that target the persona of the other user to whom the answer is to be presented, estimated from the attributes of the user who accepted the input, and generates and outputs text for the generative model so that the answer to the input includes the advertisement, enabling the user who accepted the input to deliver appropriate advertisements to other users who will submit answers.

The generation unit 123 of the information processing device 100 according to the embodiment generates a sentence containing a user's sentence, a user image, and an advertisement as a sentence for the generative model. As a result, the information processing device 100 generates and outputs a sentence containing a user's sentence, a user image, and an advertisement for the generative model so that the answer to the question includes an advertisement, enabling the user who accepted the input to deliver an appropriate advertisement according to other users who submit answers.

〔program〕
It is also possible to create a program written in a computer-executable language that executes the processes executed by the information processing device 100 described in the above embodiment. In this case, the same effects as those of the above embodiment can be achieved by having a computer execute the program. Furthermore, such a program may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read and executed by a computer to achieve the same processes as those of the above embodiment.

Figure 10 is a diagram showing an example of a computer that executes an information processing program. As shown in Figure 10, the controller 200, or computer 1000, has, for example, memory 1010, a CPU 1020, a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM 1012. The ROM 1011 stores a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1041. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1041. The serial port interface 1050 is connected to, for example, a mouse 1110 and a keyboard 1120. The video adapter 1060 is connected to, for example, a display 1130.

Here, as shown in FIG. 10, the hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. Each table described in the above embodiment is stored, for example, in the hard disk drive 1090 or memory 1010.

The control program is stored on the hard disk drive 1090, for example, as a program module that describes commands to be executed by the computer 1000. Specifically, a program module 1093 that describes each process executed by the controller 2000 described in the above embodiment is stored on the hard disk drive 1090.

In addition, data used for information processing by the control program is stored as program data, for example, on the hard disk drive 1090. The CPU 1020 then reads the program module 1093 and program data 1094 stored on the hard disk drive 1090 into the RAM 1012 as needed, and executes each of the procedures described above.

The program module 1093 and program data 1094 related to the control program do not necessarily have to be stored on the hard disk drive 1090, but may be stored on a removable storage medium, for example, and read by the CPU 1020 via the disk drive 1041 or the like. Alternatively, the program module 1093 and program data 1094 related to the control program may be stored in another computer connected via a network such as a LAN (Local Area Network) or WAN (Wide Area Network), and read by the CPU 1020 via the network interface 1070.

〔others〕
Although various embodiments have been described in detail herein with reference to the drawings, these embodiments are merely examples and are not intended to limit the present invention. In other words, the present invention makes it possible to learn, build, and update various multi-class classification models and estimate desired probabilities by changing the features used as input. The features described herein can be realized in various ways, including various modifications and improvements based on the knowledge of those skilled in the art.

Furthermore, the above-mentioned "module (-er suffix, -or suffix)" can be interpreted as unit, means, circuit, etc. For example, a communication module, a control module, and a storage module can be interpreted as a communication unit, a control unit, and a storage unit, respectively.

REFERENCE SIGNS LIST 100 Information processing device 110 Communication unit 120 Control unit 121 Estimation unit 122 Selection unit 123 Generation unit 124 Output unit 130 Storage unit

Claims (7)

an estimation unit that estimates a user image of another user who will provide an answer to an input made by the user based on information about the user;
a selection unit that selects an advertisement based on the user image estimated by the estimation unit;
a generation unit that generates a sentence for a generative model so that the advertisement selected by the selection unit is included in the answer;
an output unit that outputs the sentence generated by the generation unit to the generative model. The information processing device described in claim 1, characterized in that the estimation unit estimates the user image using one or more of the following information about the user: user attributes, search history, browsing history, location information, conversation content, conversation trends, SNS posts, context, and follower information. The information processing device described in claim 1, characterized in that the estimation unit estimates the user image from information about the user using a model that has learned the relationship between information about the user and the user image. The information processing device described in claim 1, characterized in that the selection unit selects advertisements that target the user image estimated by the estimation unit for advertisement delivery. The information processing device described in claim 1, characterized in that the generation unit generates a sentence containing the user's sentence, the user image, and the advertisement as a sentence for the generative model. An information processing method executed by an information processing device,
an estimation step of estimating a user image of another user who will provide an answer to an input made by the user based on information about the user;
a selection step of selecting an advertisement based on the user image estimated by the estimation step;
a generation step of generating a sentence for a generative model so that the advertisement selected by the selection step is included in an answer;
an output step of outputting the sentence generated in the generation step to the generative model. an estimation step of estimating a user image of another user who will provide an answer to an input made by the user based on information about the user;
a selection step of selecting an advertisement based on the user image estimated by the estimation step;
a generation step of generating a sentence for a generative model so that the advertisement selected by the selection step is included in an answer;
an output step of outputting the sentence generated in the generation step to the generative model.