JP7779759B2 - 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.

Various technologies have been proposed for providing information to users. For example, Patent Document 1 proposes a technology for delivering advertisements according to the quadrant to which a user's category of interest belongs, out of multiple quadrants that are classified based on factors related to commercial transactions.

Japanese Patent Application Laid-Open No. 2021-099631

However, while the above-mentioned conventional technology can classify objects such as elements related to commercial transactions into categories that interest users, it does not actively analyze changes in a user's interests in those objects. If it were possible to accurately analyze changes in a user's interests in those objects, it would be possible to provide information to users more appropriately, for example, and there is room for improvement in this regard.

The present application has been made in light of the above, and aims to provide an information processing device, information processing method, and information processing program that can accurately analyze changes in a user's interests in a subject.

The information processing device according to the present application includes an identification unit and an estimation unit. The identification unit identifies a second object whose transition pattern is similar to that of the first object in an interest space in which each of multiple objects, including a first object, is represented by a vector according to the relevance to the interests of multiple users, and the vector is updated every predetermined period. The estimation unit estimates the reason for the transition of the second object based on the commonality between the first object and the second object.

One aspect of this embodiment has the effect of enabling accurate analysis of changes in a user's interests in a subject.

FIG. 1 is a diagram illustrating an example of information processing according to an embodiment. FIG. 2 is a diagram illustrating an example of the configuration of an information processing system including the information processing apparatus according to the embodiment. FIG. 3 is a diagram illustrating an example of a user information table stored in the user information storage unit of the information processing apparatus according to the embodiment. FIG. 4 is a diagram illustrating an example of a content table stored in the content storage unit of the information processing device according to the embodiment. FIG. 5 is a diagram illustrating an example of interest space information stored in the interest space information storage unit of the information processing device according to the embodiment. FIG. 6 is a diagram illustrating an example of a first target identified by a first identification unit of the information processing apparatus according to the embodiment. FIG. 7 is a diagram illustrating an example of a first target identified by a first identification unit of the information processing apparatus according to the embodiment. FIG. 8 is a diagram illustrating an example of a second target identified by the second identification unit of the information processing apparatus according to the embodiment. FIG. 9 is a diagram illustrating another example of a second target identified by the second identification unit of the information processing device according to the embodiment. FIG. 10 is a flowchart showing a processing procedure performed by the processing unit of the information processing apparatus according to the embodiment. FIG. 11 is a flowchart showing an information providing procedure performed by the processing unit of the information processing apparatus according to the embodiment. FIG. 12 is a hardware configuration diagram illustrating an example of a computer that realizes the functions of the information processing apparatus according to the embodiment.

Below, detailed explanations will be given of the information processing device, information processing method, and information processing program (hereinafter referred to as "embodiments") according to the present application, with reference to the drawings. Note that the information processing device, information processing method, and information processing program according to the present application are not limited to these embodiments. Furthermore, the embodiments can be combined as appropriate to the extent that the processing content is not contradictory. Furthermore, the same components in the following embodiments will be assigned the same reference numerals, and redundant explanations will be omitted.

[1. An example of information processing]
First, an example of information processing according to an embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram illustrating an example of information processing according to an embodiment. The information processing according to the embodiment is processing executed by an information processing device 1, and includes a search process, a model generation process, and an information provision process.

First, the search process and model generation process will be described. The information processing device 1 shown in Fig. 1 provides search services to users _U1 to _Un . For example, the information processing device 1 has a search target database, which is a database in which search targets are indexed and stored, and executes search processing using information from the search target database. For example, the information from the search target database is stored in a storage unit 11 (see Fig. 2).

1, users U ₁ to U _n operate terminal devices 2 ₁ to 2 _n to cause terminal devices 2 ₁ to 2 _n to execute a process of transmitting a search query from terminal devices 2 ₁ to 2 _n to information processing device 1 (steps S1 ₁ to S1 _n ), where n is an integer of 2 or greater.

For example, in step _S11 , user _U1 operates terminal device ₂₁ to cause terminal device 21 to execute a process of transmitting a search query from terminal device ₂₁ to information processing device 1. Also, in step _S1n , user _Un operates terminal device _2n to cause terminal device _2n to execute a process of transmitting a search query from terminal device _2n to information processing device _1. Hereinafter, when referring to each of users _U1 to _Un without distinguishing them individually, they may be referred to as user U, and when referring to each of terminal devices ₂₁ to _2n without distinguishing them individually, they may be referred to as terminal device 2.

A search query includes one or more search terms (search keywords) entered into the terminal device 2 by the user U. For example, if the user U enters "sneakers" into the terminal device 2 as one or more search terms, the search query includes "sneakers." Also, if the user U enters "women's sneakers" into the terminal device 2 as one or more search terms, the search query includes "women's sneakers." "Women's sneakers" includes the two search terms "sneakers" and "women's," separated by a space.

The information processing device 1 receives search queries transmitted from the terminal devices 2 ₁ to 2 _n (steps S2 ₁ to S2 _n ). For example, the information processing device 1 receives a search query from the terminal device 2 ₁ in step S2 ₁ , and receives a search query from the terminal device 2 _n in step S2 _n .

Next, the information processing device 1 executes a search process based on the search query received in steps _S21 to _S2n (steps _S31 to _S3n ). For example, in step _S31 , the information processing device 1 executes a search process to search the search target database for objects corresponding to one or more search terms included in the search query transmitted from terminal device ₂₁ and received in step _S21 . In addition, in step _S3n , the information processing device 1 executes a search process to search the search target database for objects corresponding to one or more search terms included in the search query transmitted from terminal device _2n and received in step _S2n .

Next, the information processing device 1 transmits the search results resulting from the search processing in steps _S31 to _S3n to the terminal devices ₂₁ to _2n (steps _S41 to _S4n ). For example, in step S41, the information processing device ₁ transmits the search results resulting from the search processing in step _S31 to the terminal device _21. In addition, in step _S4n , the information processing device 1 transmits the search results resulting from the search processing in step _S3n to the terminal device _2n .

Next, the information processing device 1 generates an interest model based on the search query received in steps S2 ₁ to S2 _n (step S5). The interest model generated in step S5 is a model that receives information indicating each of a plurality of targets as input and outputs an M-dimensional vector. M is, for example, an integer in the range of 500 to 2000, but is not limited to this example. Furthermore, the M-dimensional vector may be represented, for example, by a distributed representation or by a representation other than the distributed representation. Hereinafter, the M-dimensional vector will be simply referred to as a vector.

In step S5, the information processing device 1 generates an interest model, which is a trained model, by learning the characteristics of each of the multiple search terms, regarding two or more search terms that satisfy predetermined conditions as having similar characteristics. The two or more search terms that satisfy the predetermined conditions are multiple search terms included in the same search query, or search terms included in multiple search queries sent from the terminal device 2 by the same user U within a predetermined time period.

The information processing device 1 uses two or more search terms that satisfy predetermined conditions as training data and performs training so that the vectors of the two or more search terms are similar to each other. The information processing device 1 can also treat search terms included in search queries sent from the terminal device 2 by the same user U as two or more search terms that satisfy predetermined conditions, regardless of when the search queries were sent.

Note that a search term consists of one search keyword, but may also consist of two or more search keywords. For example, if the search query contains the string "ladies' sneakers," the information processing device 1 will treat "sneakers" and "ladies" as different search terms, but it can also treat the pair of "sneakers" and "ladies" as a single search term.

The information processing device 1 generates an interest model that outputs vectors (e.g., distributed representations) from information indicating a target, such as a search term, using, for example, Deep Structured Semantic Model (DSSM) technology, which uses Long Short-Term Memory (LSTM), a type of Recurrent Neural Network (RNN), also known as a recursive neural network, for vector generation (e.g., distributed representation generation). Note that the method for generating an interest model that outputs vectors from information indicating a target is not limited to the above-mentioned example.

The information processing device 1 generates an interest model for each predetermined period TA. For example, if the predetermined period TA is one month, the information processing device 1 generates an interest model for each month of January, February, March, etc. in 2022.

Next, the information provision process will be described. The information processing device 1 determines the transition pattern of the first object in the space of interest, which is updated every predetermined period TA (step S6).

In the interest space, each of multiple targets is represented by a vector according to their relevance to the interests of multiple users U. In the interest space, targets represented by vectors are not limited to targets represented by search terms, and may be targets other than those represented by search terms. The information processing device 1 generates an interest space including vectors for each target by inputting information representing the target into the interest model generated in step S5 and performing a process for each target to obtain the target vector output from the interest model. The information processing device 1 generates an interest space that is updated for each period TA using the interest model generated for each period TA in step S5.

The first target is, for example, a target among multiple targets that satisfies a predetermined condition. The predetermined condition is, for example, that the target has a predetermined transition pattern based on an interest vector, which is a vector indicating the interests in the interest space of the user U to be determined among multiple users U. The predetermined transition pattern is, for example, a transition pattern in which the target approaches a reference vector in the interest space and enters an interest range, which is a range similar to the reference vector, or a transition pattern in which the target moves away from the reference vector in the interest space and moves from within the interest range to outside the interest range.

Here, the interest vector and interest range of the user U to be judged will be explained. The information processing device 1 inputs search terms included in multiple search queries sent from the terminal device 2 by the same user U to be judged into the interest model generated in step S5, and performs a process of obtaining the vector of the search term output from the interest model for each search term. The information processing device 1 calculates an average vector by averaging the vectors of the search terms included in multiple search queries sent from the terminal device 2 by the same user U to be judged, and judges the calculated average vector to be the interest vector. Note that the information processing device 1 can also calculate the average vector by, for example, weighting the vector of a search term more heavily for search terms in search queries that are received more recently.

In addition, the information processing device 1 can also identify, as the first object, an object among multiple objects that has a specific change pattern in the interest space that is updated every period TA. The information processing device 1 can also treat an object specified by the user U of the information processing device 1 as the first object.

Next, the information processing device 1 identifies a second object in the interest space whose transition pattern is similar to that of the first object (step S7). For example, the information processing device 1 identifies as the second object an object whose transition pattern is similar to that of the first object in the latest interest space from the interest space up to K periods ago and approaching the interest vector of user U. K is, for example, an integer greater than or equal to 2.

Next, the information processing device 1 estimates the reason for the change in the second object based on the commonality between the first object and the second object (step S8). The commonality between the first object and the second object is, for example, the commonality of the object's attributes. If the object is a trading object, the object's attributes include various attributes corresponding to the object, such as the category, usage scenario, specifications (such as the specifications or functions of the trading object), the object's appearance date (such as the product's release date), issues that can be solved by the object, and people's concerns that can be solved by the object.

For example, if the first object and the second object are in the same category, the information processing device 1 estimates that the transition pattern of the second object is similar to that of the first object due to the common category. For example, suppose an object that approaches the interest vector of user U in the interest space and enters the interest range, which is the range similar to the interest vector of user U, is identified as the first object. In this case, the information processing device 1 estimates that the reason for the transition of the second object is that user U's interest in the category common to the first and second objects is increasing.

Also, for example, suppose an object that has moved away from user U's interest vector in the interest space and from within the interest range to outside the interest range is identified as the first object. In this case, the information processing device 1 infers that the reason for the change in the second object is that user U's interest in a category common to the first object and the second object has decreased.

Next, the information processing device 1 provides the user UA of the information processing device 1 with information indicating the reason for the change in the second target estimated in step S8 (step S9). The information processing device 1 provides the user UA of the information processing device 1 with information indicating the reason for the change in the second target by transmitting the information indicating the reason for the change in the second target estimated in step S8 to the terminal device 3 of the user UA of the information processing device 1. This allows the user UA of the information processing device 1 to know the reason for the change in the second target and to understand the changes in the user U's interest in the second target.

In this way, the information processing device 1 identifies a second object whose transition pattern is similar to that of a first object in the interest space, which is updated every predetermined period, and estimates the reason for the transition of the second object based on the commonality between the first object and the second object. This allows the information processing device 1 to accurately analyze the transition of user U's interests in objects.

2. Configuration of the information processing system
Fig. 2 is a diagram showing an example of the configuration of an information processing system including an information processing device 1 according to an embodiment. As shown in Fig. 2, the information processing system 100 includes the information processing device 1, a plurality of terminal devices 2 ₁ to 2 _n , and a terminal device 3. The information processing device 1, the plurality of terminal devices 2 ₁ to 2 _n , and the terminal device 3 are connected via a network N so as to be able to communicate with each other via wired or wireless communication.

The information processing device 1 according to the embodiment is an information processing device capable of communicating with various devices via a predetermined network N such as the Internet, and is realized, for example, by a server device or a cloud system. For example, the information processing device 1 is connected to various other devices via the network N so that it can communicate with them.

The information processing device 1 also provides online services such as web services to the terminal device 2 of each user U. For example, in addition to the search service and information provision service described above, the information processing device 1 also provides other online services such as SNS (Social Networking Service), electronic commerce (EC), posting sites, electronic payments, online games, online banking, online trading, accommodation and ticket reservations, video and music distribution, news, maps, route searches, route guidance, line information, operation information, and weather forecasts. The information processing device 1 can also link with various servers that provide the online services described above and act as an intermediary for the online services.

The terminal device 3 is, for example, a terminal device operated by a user UA of the information processing device 1. The user UA can operate the terminal device 3 to obtain, from the information processing device 1, information such as information indicating a change in the user U's interests that is provided by the information processing device 1. The information indicating a change in the user U's interests includes, for example, information indicating the reason for the change in the second target described above.

3. Configuration of information processing device 1
The following describes an example of the functional configuration of the information processing device 1. As shown in FIG.

3.1. Communication Unit 10
The communication unit 10 is realized by, for example, a network interface card (NIC). The communication unit 10 is connected to a network N by wire or wirelessly, and transmits and receives information to and from various other devices. For example, the communication unit 10 transmits and receives information to and from terminal devices 2 ₁ to 2 _n via the network N.

[3.2. Storage unit 11]
The storage unit 11 is realized by, for example, a semiconductor memory element such as a random access memory (RAM) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 11 also has a search information storage unit 20, a user information storage unit 21, a content storage unit 22, and an interest space information storage unit 23.

3.2.1. Search Information Storage Unit 20
The search information storage unit 20 stores information on a plurality of search targets provided by the information processing device 1 in the search service. For example, the search information storage unit 20 stores a search target database, which is a database in which each of a plurality of search targets is indexed and stored.

The information to be searched is, for example, information on various content such as web pages collected by a crawler. The information to be searched stored in the search information storage unit 20 includes, but is not limited to, the URL (Uniform Resource Locator) and summary of the content.

3.2.2. User Information Storage Unit 21
The user information storage unit 21 stores user information including information on users U ₁ to U _n . Fig. 3 is a diagram showing an example of a user information table stored in the user information storage unit 21 of the information processing device 1 according to the embodiment.

As shown in Figure 3, the user information table stored in the user information storage unit 21 includes information such as "User ID (Identifier)," "User Name," "Attributes," and "Search History" for each user U. "User ID" is identification information unique to each user U. "User Name" is information indicating the name of the user U.

"Attributes" is information indicating the attributes of user U. User U's attributes include user U's demographic attributes or psychographic attributes. Demographic attributes are demographic attributes of user U. Psychographic attributes are attributes that indicate user U's values, lifestyle, personality, interests, etc.

In the example shown in Figure 3, user U's demographic attributes include information such as "gender" and "age." "Gender" is information indicating user U's gender, and "age" is information indicating user U's age. User U's demographic attributes further include, for example, user U's job title, job responsibilities, annual income, address, commuting route, training history, family composition, etc. User U's preferences include, for example, user U's level of interest in each of items such as clothes, travel, cars, motorcycles, computers, and lunch.

"Search history" is information about the search history of online services provided by information processing device 1 or online services mediated by information processing device 1 by user U. For each search query, such "search history" includes, for example, information indicating the date and time when the search query was received by information processing device 1, and information about one or more search terms (search keywords) included in the search query.

3.2.3. Content storage unit 22
The content storage unit 22 stores content provided by the information processing device 1 through online services other than search services. Fig. 4 is a diagram showing an example of a content table stored in the content storage unit 22 of the information processing device 1 according to the embodiment. In the example shown in Fig. 4, the content storage unit 22 includes a "content ID,""content," and the like for each piece of content.

A "content ID" is identification information unique to each piece of content. A "content" is information related to the content associated with a "content ID." Specifically, the content may indicate information related to the content's content. For example, the content is content provided by an online service. For example, the content may be content related to a portal site, news site, auction site, weather forecast site, shopping site, or finance (stock price) site. The content may also be content related to a route search site, map provider site, travel site, restaurant introduction site, blog site, posting site, music distribution site, video distribution site, or SNS site.

For example, in Figure 4, the content with content ID "C1" is "CO1." Note that in the example shown in Figure 4, the content is expressed using abstract codes such as "CO1," but the content may also be in the form of a file containing specific numerical values, specific character strings, and various types of information. Note that the content storage unit 22 is not limited to the above example, and may store various types of information depending on the purpose.

3.2.4. Interest Space Information Storage Unit 23
The interest space information storage unit 23 stores interest space information for each predetermined period TA. The interest space information includes information such as vectors of multiple objects placed in the interest space. The interest space is an M-dimensional space. The period TA is, for example, one month, but is not limited to this example and may be, for example, one week, two weeks, or three months. Furthermore, the period TA may be a period during which the number of search queries or new search terms newly received by the information processing device 1 exceeds a predetermined threshold.

Figure 5 is a diagram showing an example of interest space information stored in the interest space information storage unit 23 of the information processing device 1 according to the embodiment. In the example shown in Figure 5, the interest space information stored in the interest space information storage unit 23 includes a "target ID," a "target," a "vector," and the like for each target.

The "target ID" is identification information unique to each target. A "target" is a target that may be of interest to user U and is placed in the interest space, and may belong to various categories, such as shopping, travel, news, sports, entertainment, finance, games, movies, or music.

For example, in Figure 5, the object with object ID "Q1" is "O1" and the vector is "V1". Note that in the example shown in Figure 5, the object is represented by an abstract code such as "O1", but the object may be represented by a specific string of characters or by an image, etc. Also, in the example shown in Figure 5, the vector is represented by an abstract code such as "V1", but the vector is an M-dimensional vector and is represented, for example, by the values of the vector components of each dimension.

3.3. Processing Unit 12
The processing unit 12 is a controller, and is realized by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit) using RAM as a work area to execute various programs (examples of information processing programs) stored in a storage device inside the information processing device 1. The processing unit 12 is also a controller, and is realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

As shown in FIG. 2, the processing unit 12 has a reception unit 30, a search unit 31, a learning unit 32, an identification unit 33, an estimation unit 34, and a provision unit 35, and realizes or executes the information processing functions and actions described below. Note that the internal configuration of the processing unit 12 is not limited to the configuration shown in FIG. 2, and may be any other configuration that performs the information processing described below.

3.3.1. Reception unit 30
The reception unit 30 receives various requests. The reception unit 30 receives various requests from external information processing devices. For example, the reception unit 30 receives requests from each terminal device 2.

The reception unit 30 receives a search query containing one or more search terms (search keywords) entered by the user U from the terminal device 2 via the network N and the communication unit 10. The reception unit 30 also receives a content transmission request from the user U from the terminal device 2 via the network N and the communication unit 10. A content transmission request is a request that specifies content.

The reception unit 30 also receives requests from the terminal device 3. For example, the reception unit 30 receives a target-specifying query including information about a target specified by the user U as a first target from the terminal device 3 via the network N and the communication unit 10. The reception unit 30 also receives a user-specifying query including information indicating the user U to be judged from the terminal device 3 via the network N and the communication unit 10.

3.3.2. Search unit 31
The search unit 31 searches for information to be searched for that corresponds to one or more search terms included in the search query received by the reception unit 30 from among the plurality of pieces of information to be searched for that are stored in the search information storage unit 20. The search unit 31 transmits the searched information to the terminal device 2 that sent the search query as a search result via the network N and the communication unit 10.

The search unit 31 can also search for information to be searched for that corresponds to one or more search terms included in the search query received by the reception unit 30 from among multiple pieces of content stored in the content storage unit 22. The search unit 31 transmits the searched content as a search result to the terminal device 2 that sent the search query via the network N and the communication unit 10.

3.3.3. Learning Unit 32
The learning unit 32 generates an interest model for each period TA, which is a trained model that has learned the characteristics of each of multiple search terms, assuming that two or more search terms that satisfy predetermined conditions have similar characteristics.

Two or more search terms that satisfy a predetermined condition are multiple search terms included in the same search query, or search terms included in multiple search queries sent from the terminal device 2 by the same user U within a predetermined time period. Note that the learning unit 32 can also treat search terms included in search queries sent from the terminal device 2 by the same user U as two or more search terms that satisfy a predetermined condition, regardless of when the search queries were sent.

The learning unit 32 generates training data using search terms included in search queries with search dates and times within the time period TA. For example, the learning unit 32 acquires two or more search terms that satisfy predetermined conditions for each time period TA from the user information storage unit 21, and generates an interest model using the acquired two or more search terms as training data. For example, the learning unit 32 performs training using two or more search terms that satisfy predetermined conditions as training data, so that the vectors of the two or more search terms are similar to each other.

The learning unit 32 generates an interest model that outputs vectors from information indicating targets, such as search terms, using, for example, DSSM technology, which uses LSTM, a type of RNN also known as a recursive neural network, for vector generation. Note that the method of generating an interest model that outputs vectors from information indicating targets, such as search terms, is not limited to the example described above; any known technology can be used as long as learning can be performed so that vectors indicating multiple similar targets are similar to each other.

The learning unit 32 acquires vectors for each of the multiple targets using the generated interest model for each period TA, and stores information including the acquired vectors of the multiple targets as interest space information in the interest space information storage unit 23 in the storage unit 11 for each period TA.

[3.3.4. Specification part 33]
The identification unit 33 identifies a second object whose transition manner is similar to that of the first object in an interest space in which each of a plurality of objects is represented by a vector according to relevance to the interests of a plurality of users U and the vector is updated for each period TA. The identification unit 33 includes a first identification unit 40 and a second identification unit 41.

[3.3.4.1. First specific part 40]
The first identification unit 40 identifies an object among the plurality of objects that satisfies a predetermined condition as the first object. The object that satisfies the predetermined condition may be, for example, an object that has a predetermined transition pattern based on the interest vector of the user U to be determined, an object among the plurality of objects that has a specific change pattern, or an object designated by the user UA of the information processing device 1. The interest vector of the user U to be determined is a vector in the latest interest space, but may also be a vector in the interest space of each period TA.

The target specified by the user UA of the information processing device 1 is, for example, the target indicated in the target specification query received by the reception unit 30. In this case, the first identification unit 40 can determine the user U to be determined based on the relationship with the target specified by the user UA. The user U to be determined may also be, for example, the user UA indicated in the user specification query received by the reception unit 30. The information processing device 1 can also identify, as the first target, a target among multiple targets in the interest space updated every period TA, whose transition pattern is a specific transition pattern.

The transition pattern predetermined based on user U's interest vector is, for example, a transition pattern in which the user approaches user U's interest vector in the interest space and enters an interest range that is a range similar to user U's interest vector, or a transition pattern in which the user moves away from user U's interest vector in the interest space and moves from within the interest range to outside the interest range.

The interest space is updated for each period TA. For example, if there is an interest space for each month from October 2021 to February 2022, the latest interest space is the interest space for February 2022. The interest space for each month from October 2021 to February 2022 is obtained using the interest model for each month from October 2021 to February 2022. The interest space for October 2021 is obtained by inputting information indicating each of multiple targets into the interest model for October 2021, and the interest space for February 2022 is obtained by inputting information indicating each of multiple targets into the interest model for February 2022. In addition, the interest space will be updated sequentially to the state of the interest space in October 2021, the state of the interest space in November 2021, the state of the interest space in December 2021, the state of the interest space in January 2022, and the state of the interest space in February 2022.

The first identification unit 40, for example, uses an interest model for each period TA to identify the range of interests of each user U for each period TA. For example, the first identification unit 40 acquires, for each user U, from the user information storage unit 21, search terms included in all search queries sent from the terminal device 2 by the same user U within each period TA. The first identification unit 40 then averages the vectors of the acquired search terms to calculate an average vector, and performs a process for each period TA in which the calculated average vector is determined as the user interest vector for each user U. Note that the first identification unit 40 can also calculate the average vector by, for example, weighting the vector of the search term more heavily the more recent the date and time it was received.

The first identification unit 40 identifies, for each user U, a range similar to the user interest vector of the user U as the user U's interest range. The range of similarity to the user interest vector is, for example, a range of cosine similarity determined in advance, but is not limited to this example. Note that the first identification unit 40 can also identify the user U's interest range, for example, each time the number of new search queries or new search terms for the same user U exceeds a predetermined threshold.

After identifying the interest vector of user U, the first identification unit 40 identifies, for example, an object having a predetermined transition pattern based on the interest vector of user U to be determined as the first object. The interest vector of user U is a vector in the latest interest space, but it may also be a vector in the interest space of each period TA.

6 is a diagram showing an example of a first object identified by the first identification unit 40 of the information processing device 1 according to the embodiment. Fig. 6 shows a portion of an interest space visualized by reducing the dimension of an M-dimensional interest space. The example shown in Fig. 6 shows the position P ₁ (t -3 ) of the object O1 in the interest space three periods ago, the position P 1 (t _-2 ) of the object O1 in the interest space two periods ago, the position P ₁ (t _-1 ) of the object O1 in the interest space _one period ago, and the position P ₁ ( _{t 0 )} of O1 in the latest interest space.

As shown in FIG. 6 , target O1 is not within user _U1 's range of interests in the interest space from three periods ago to one period ago, but is within user _U1 's range of interests in the most recent interest space. Furthermore, the more recent the interest space in the interest space from three periods ago to one period ago, the closer it is to user _U1 's range of interests. In this case, the first identification unit 40 determines that target O1 has a transition pattern in which it approaches user _U1 's range of interests in the interest space and enters user _U1 's range of interests, and identifies target O1 as the first target. Note that the first identification unit 40 can also identify as the first target a target that has transitioned into user _U1 's range of interests even if it has moved away from user _U1 's range of interests at some point.

Fig. 7 is a diagram showing an example of a first object identified by the first identification unit 40 of the information processing device 1 according to the embodiment. Similar to Fig. 6, Fig. 7 shows a part of an interest space visualized by reducing the dimension of an M-dimensional interest space. In the example shown in Fig. 7, similar to Fig. 6, the positions P ₁ (t _-3 ), P ₁ (t _-2 ), P ₁ (t _-1 ), and P ₁ (t ₀ ) of an object O1 from three periods ago to the latest interest space are shown.

In the example shown in Figure 7, the target O1 was within the range of user _U1 's interests in the interest space three periods ago, but was outside the range of user _U1 's interests in the interest space two periods ago. The closer the target O1 is to the most recent interest space, the further it is from the range of user _U1 's interests. In this case, the first identification unit 40 determines that the target O1 has a transition pattern in the interest space that moves from within the range of user U's interests to outside the range of interests, and identifies the target O1 as the first target. Note that the first identification unit 40 can also identify as the first target a target that has transitioned further away from the range of user _U1 's interests, even if it has approached the range of user _U1 's interests at some point.

The first identification unit 40 can also identify as the first target an object among multiple objects whose transition pattern is a specific transition pattern, or an object designated by a user U of the first identification unit 40. The first identification unit 40 can also identify as the first target an object whose movement speed or movement amount during a predetermined period is equal to or greater than a threshold value.

[3.3.4.2. Second specific part 41]
The second identification unit 41 identifies a second object whose transition manner is similar to that of the first object identified by the first identification unit 40. For example, the second identification unit 41 determines, in an interest space updated for each period TA, a change in the difference between the vectors of the object between two consecutive periods of interest spaces as the transition manner of the object.

For example, let the vectors of the first object from three periods before to the latest space of interest be _V13 , _V12 , _V11 , and _V10 . Also, let the differences of the vectors of the first object between two spaces of interest in consecutive periods be _ΔV132 , _ΔV121 , and _ΔV110 . _ΔV132 is _ΔV132 = _V13 - _V12 , _ΔV121 is _V12 - _V11 , and _ΔV110 is _ΔV110 = _V11 - _V10 .

Also, let _VX3 , _VX2 , _VX1 , and _VX0 be the vectors of a target _OX different from the first target from three periods before to the latest space of interest. Let _ΔVX32 , ΔVX21, and _ΔVX10 be the differences in the vectors of target _OX between two spaces of interest with consecutive periods. _ΔVX32 is _ΔVX32 = _VX3 - _VX2 , _ΔVX21 is _ΔVX21 = _VX2 - _VX1 , and _ΔVX10 is _ΔVX10 = _{VX1 - VX0 .}

In this case, the second identification unit 41 identifies the object O X as a _{second object} whose transition manner is similar to that of the first object when, for example, the result of weighting and adding the similarity between ΔV ₁₃₂ and ΔV _{X 32} , the similarity between ΔV ₁₂₁ and ΔV _{X 21} , and the similarity between ΔV ₁₁₀ and ΔV X 10 is less than a threshold value. The similarity is, for example, cosine similarity, but is not limited to this example.

Furthermore, the second identification unit 41 may identify the target _{O_X as the second target, for example, when the transition pattern of the first target from the space of interest K periods ago to the latest space of interest is similar to the transition pattern of the vector of the target O_X in} the space of interest for a certain period. K is, for example, an integer equal to or greater than 5. The certain period may be the same period for the first target and the second target, or may be different periods for the first target and the second target.

The same period is, for example, the period from P periods ago to the most recent period, where P is an integer greater than 1 and less than M. For example, different periods are, for the first target, the period from P-N periods ago to the most recent period, and for the second target, the period from P periods ago to N periods ago. N is, for example, an integer greater than or equal to 1.

Fig. 8 is a diagram showing an example of a second object identified by the second identification unit 41 of the information processing device 1 according to the embodiment. Similar to Fig. 6 and Fig. 7, Fig. 8 shows a part of an interest space visualized by reducing the dimension of the M-dimensional interest space. In the example shown in Fig. 8, in addition to the positions P ₁ (t _-3 ), P 1 (t _-2 ), P ₁ (t _-1 ), and _{P 1 (} t ₀ ) of the object O1 shown in Fig. 6, positions P _X ( _{t -3} ), P _X (t _-2 ), P _X (t _-1 ), and P _X (t ₀ ) of the object O X are shown.

Position P _X (t _-3 ) is the position of object O _X in the space of interests three periods ago, position P _X (t _-2 ) is the position of object O _X in the space of interests two periods ago, position P _X (t _-1 ) is the position of object O _X in the space of interests one period ago, and position P _X (t ₀ ) is the position of object O _X in the latest space of interests.

8, the transition of target O1 from position P ₁ (t _-3 ) to position P ₁ (t ₀ ) is similar to the transition of target O _X from position P _X (t _-3 ) to position P _X (t ₀ ). That is, the transition of target O1 from the space of interest three periods ago to the latest space of interest is similar to the transition of target O _X from the space of interest three periods ago to the latest space of interest. Therefore, the second identification unit 41 identifies target O _X as the second target.

Fig. 9 is a diagram showing another example of a second target identified by the second identification unit 41 of the information processing device 1 according to the embodiment. In Fig. 9, in addition to the state of Fig. 8, a position P ₁ (t _-4 ) of target O ₁ in the space of interest four periods ago is added, and positions P _X (t _-3 ), P X (t _-2 ), P _X (t _-1 ), and P _X (t ₀ ) of target _{O X} are different from the positions shown in Fig. 8.

In the example shown in Figure 9, the transition pattern of target O1 from position _P1 (t _-3 ) to position _P1 (t _-1 ) is similar to the transition pattern of target _OX from position _Px (t _-2 ) to position _Px ( _t0 ). In other words, the transition pattern of target O1 from the space of interest three periods ago to the space of interest one period ago is similar to the transition pattern of target _OX from the space of interest two periods ago to the latest space of interest. Therefore, the second identification unit 41 identifies target _OX as the second target. In this way, a target that is partially similar to the transition pattern of the second identification unit 41 can be identified as the second target.

Furthermore, instead of or in addition to the change in the difference in the vector of the object between two spaces of interest of successive periods, the second identification unit 41 can also treat the difference in the vector of the object between the space of interest K periods ago and the latest space of interest as the transition mode.

A change in the difference between the vectors of an object between two consecutive interest spaces is a change in the vector of the object, but either a change in the magnitude of the object vector or a change in the direction of the object vector can also be treated as the transition mode of the object. A change in the magnitude of the object vector indicates a change in the amount of movement of the object vector or a change in the speed of movement of the object vector, and a change in the direction of the object vector indicates a change in the direction of movement of the object vector.

[3.3.5. Estimation unit 34]
The estimation unit 34 estimates the reason for the transition of the second object based on the commonality between the first object and the second object identified by the identification unit 33. The commonality between the first object and the second object is, for example, the commonality of the attributes of the objects.

The attributes of an object include various attributes corresponding to the object, such as the category, usage scenario, specifications (such as the specifications or functions of the traded object), the date of appearance of the object (such as the product's release date), issues that can be solved by the object, and human concerns that can be resolved by the object. An object is, for example, a trade object, and a trade object is, for example, a product or service.

For example, if the first object and the second object are in the same category, the estimation unit 34 estimates that the change in the second object is similar to the change in the first object due to the common category. For example, suppose that an object that approaches the interest vector of user U in the interest space and enters the interest range, which is a range similar to the interest vector of user U, is identified as the first object. In this case, the estimation unit 34 estimates that the reason for the change in the second object is that user U's interest in the category common to the first and second objects is increasing.

Also, for example, suppose an object that has moved away from user U's interest vector in the interest space and from within the interest range to outside the interest range is identified as the first object. In this case, the estimation unit 34 estimates that the reason for the change in the second object is that user U's interest in a category common to the first object and the second object has decreased.

Furthermore, when the first object and the second object have multiple attributes in common, the estimation unit 34 estimates that the reason for the change in the second object is that the second object shares multiple attributes with the first object. For example, suppose that the first object and the second object are both bags and are the same brand, Brand A, and that Brand A's products include wallets and clothes in addition to bags. Then, suppose that the first object is moving away from the interest vector of user U. In this case, the estimation unit 34 estimates that the reason for the change in the second object is that user U is losing interest in Brand A and shoes.

In the above example, the estimation unit 34 estimated the reason for the change of the second target from the change in the interest vector and interest range of user U, but this is not limited to such an example. For example, suppose that the first identification unit 40 identifies a target with a specific change as the first target, or that the first identification unit 40 identifies a target identified by user U as the first target. In this case, too, the estimation unit 34 can estimate the reason for the change of the second target based on the commonality between the first target and the second target.

[3.3.6. Providing Department 35]
The providing unit 35 provides the user UA with transition reason information, which is information indicating the transition reason of the second object estimated by the estimating unit 34. The transition reason information is provided to the user UA by transmitting it from the providing unit 35 to the terminal device 3 via the communication unit 10 and the network N. This allows the providing unit 35 to provide the user UA with information indicating the transition reason of the second object.

The transition reason information is text information indicating the second object and the reason for the transition of the second object. In addition to or instead of such text information, it may also include information illustrating the transition between the first object and the second object in the visualized interest space, for example, as shown in Figure 8 or Figure 9.

The providing unit 35 can provide the transition reason information to the user UA in a push-type manner or in a pull-type manner. For example, the providing unit 35 can display the transition reason information as a pop-up on the terminal device 3 using an application installed on the terminal device 3, or send it to the user UA's email address by email. The providing unit 35 can also provide the transition reason information to the user UA by sending the transition reason information to the terminal device 3 when the user UA accesses the information processing device 1 using the terminal device 3. The method of providing the transition reason information to the user UA is not limited to these methods.

4. Processing Procedure
Next, a processing procedure by the information processing device 1 according to the embodiment will be described with reference to Fig. 10. Fig. 10 is a flowchart showing a processing procedure by the processing unit 12 of the information processing device 1 according to the embodiment.

As shown in FIG. 10, the processing unit 12 of the information processing device 1 determines whether it is time to perform the learning process for the interest model (step S10). The learning process for the interest model occurs, for example, every period TA, but is not limited to this example.

When the processing unit 12 determines that it is time to perform the interest model learning process (Step S10: Yes), it generates an interest model (Step S11). Then, the processing unit 12 generates interest space information using the interest model generated in Step S11, and stores the generated interest space information in the memory unit 11 (Step S12).

When the processing of step S12 is completed, or when it is determined that it is not time to perform the interest model learning process (step S10: No), the processing unit 12 determines whether it is time to determine user interests (step S13). The user interest determination timing is, for example, a timing that occurs every period TA, but is not limited to this example.

If the processing unit 12 determines that it is time to determine the user's interests (step S13: Yes), it identifies the user U's points of interest based on the user U's search query (step S14).

When the processing of step S14 is completed, or when it is determined that the timing for determining user interests has not yet arrived (step S13: No), the processing unit 12 determines whether the timing for determining whether the user is a target of information provision has arrived (step S15). The timing for determining whether the user is a target of information provision is the timing after the user U's location of interests has been identified in step S14. For example, the timing for determining whether the user is a target of information provision is the timing immediately after the user U's location of interests has been identified in step S14, or the timing when the user U accesses the information processing device 1 using the terminal device 2.

If the processing unit 12 determines that the time has come to determine whether or not information should be provided (step S15: Yes), it performs information provision processing (step S16). The processing in step S16 corresponds to steps S20 to S25 shown in Figure 11 and will be described in detail later.

When the processing of step S16 is completed, or when it is determined that the time to determine whether information should be provided has arrived (step S15: No), the processing unit 12 determines whether the time to end the operation has arrived (step S17). The processing unit 12 determines that the time to end the operation has arrived, for example, when the power to the information processing device 1 is turned off.

If the processing unit 12 determines that the operation end timing has not yet arrived (step S17: No), it proceeds to step S10. If the processing unit 12 determines that the operation end timing has arrived (step S17: Yes), it terminates the processing shown in FIG. 10.

Figure 11 is a flowchart showing the information provision procedure performed by the processing unit 12 of the information processing device 1 according to the embodiment. As shown in Figure 11, the processing unit 12 identifies, as a first object, an object that satisfies a predetermined condition among the number of objects in the interest space indicated by the interest space information generated in step S12 of Figure 11 and in which the vectors of each object are updated over a period TA (step S20).

Next, the processing unit 12 identifies a second object whose transition pattern is similar to that of the first object identified in step S20 (step S21). The processing unit 12 then estimates the reason for the transition of the second object based on the commonality between the first object identified in step S20 and the second object identified in step S21 (step S22). The processing unit 12 then provides the user UA with information indicating the reason for the transition of the second object estimated in step S22 (step S23), and terminates the processing shown in FIG. 11.

5. Modifications
The information processing device 1 described above may be implemented in various different forms other than the embodiment described above. Therefore, other embodiments of the information processing device 1 will be described below.

The processing unit 12 of the information processing device 1 can generate multiple types of interest models. For example, the processing unit 12 can generate an interest model for each region and create an interest space for each region based on the interest model for each region. For example, the processing unit 12 can generate an interest model for each region that learns the characteristics of each of multiple search terms, assuming that two or more search terms that satisfy predetermined conditions out of multiple search terms used by multiple users U in the target region have similar characteristics.

The processing unit 12 can also create an interest space for each attribute of user U based on an interest model for each attribute of user U. For example, the processing unit 12 can generate an interest model for each specific attribute by learning the characteristics of each of multiple search terms, assuming that two or more search terms that satisfy predetermined conditions among multiple search terms used by multiple users U with a specific attribute have similar characteristics.

Furthermore, the processing unit 12 generates an interest model for a period TA based on multiple search queries sent by multiple users U from multiple terminal devices 2 during that period TA, but this example is not limited to this. For example, the processing unit 12 can also generate an interest model for the latest period TA based on multiple search queries sent by multiple users U from multiple terminal devices 2 during the period TA, in a period including the period TA P periods before and the latest period TA. P is an integer greater than or equal to 1.

The processing unit 12 can also identify the range of interests of each user U for each period TB based on multiple search queries sent by the same user UA from the terminal device 2 during that period TB. The period TB is, for example, a period longer or shorter than the period TA. Note that the period TB may be a period during which the number of search queries or new search terms newly received by the information processing device 1 from the same user U becomes equal to or exceeds a predetermined threshold.

6. Hardware Configuration
The information processing device 1 according to the embodiment described above is realized by, for example, a computer 80 configured as shown in Fig. 12. Fig. 12 is a hardware configuration diagram showing an example of the computer 80 that realizes the functions of the information processing device 1 according to the embodiment. The computer 80 has a CPU 81, a RAM 82, a ROM (Read Only Memory) 83, an HDD (Hard Disk Drive) 84, a communication interface (I/F) 85, an input/output interface (I/F) 86, and a media interface (I/F) 87.

The CPU 81 operates based on programs stored in the ROM 83 or HDD 84 and controls each component. The ROM 83 stores a boot program executed by the CPU 81 when the computer 80 starts up, as well as programs that depend on the computer 80's hardware.

The HDD 84 stores programs executed by the CPU 81 and data used by such programs. The communication interface 85 receives data from other devices via the network N (see Figure 2) and sends it to the CPU 81, and transmits data generated by the CPU 81 to other devices via the network N.

The CPU 81 controls output devices such as displays and printers, and input devices such as keyboards and mice, via the input/output interface 86. The CPU 81 acquires data from the input devices via the input/output interface 86. The CPU 81 also outputs data generated via the input/output interface 86 to the output devices.

The media interface 87 reads programs or data stored on the recording medium 88 and provides them to the CPU 81 via the RAM 82. The CPU 81 loads the programs from the recording medium 88 onto the RAM 82 via the media interface 87 and executes the loaded programs. The recording medium 88 may be, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or a PD (Phase Change Rewritable Disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory.

The CPU 81 of the computer 80 executes programs loaded onto the RAM 82 to realize the functions of the processing unit 12. In addition, the HDD 84 stores data in the memory unit 11. The CPU 81 of the computer 80 reads and executes these programs from the recording medium 88, but as another example, these programs may be obtained from another device via the network N.

[7. Other]
Furthermore, among the processes described in the above embodiments and modifications, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods. In addition, the information including the processing procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, the various information shown in each drawing is not limited to the information shown in the drawings.

Furthermore, the components of each device shown in the figure are functional concepts and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc.

Furthermore, the above-described embodiments and variations can be combined as appropriate to the extent that the processing content is not contradictory.

8. Effects
As described above, the information processing device 1 according to the embodiment includes the identification unit 33 and the estimation unit 34. The identification unit 33 identifies a second object whose transition pattern is similar to that of the first object in an interest space in which each of a plurality of objects, including a first object, is represented by a vector according to the relevance to the interests of a plurality of users U and the vector is updated every predetermined period. The estimation unit 34 estimates the reason for the transition of the second object based on the commonality between the first object and the second object. This allows the information processing device 1 to accurately analyze the transition of the user U's interests in objects.

The estimation unit 34 also estimates the reason for the change in the second object by considering at least one of the categories, specifications, and brands of the first and second objects as commonalities between the first and second objects. This allows the information processing device 1 to accurately analyze the changes in the user U's interests in objects.

The identification unit 33 also identifies, as a first object, an object that has a predetermined transition pattern based on a vector indicating the interests of the target user U in the interest space of the target user U among the multiple users U. This allows the information processing device 1 to accurately analyze the transition of the user U's interests in the object.

In addition, the identification unit 33 identifies, as the first object, an object that approaches the reference in the interest space and falls within a range of interests similar to the reference. This allows the information processing device 1 to accurately analyze changes in user U's interests in objects.

In addition, the identification unit 33 identifies as a first object an object that has moved away from the reference in the interest space from within a range of interests similar to the reference to outside the range of interests. This allows the information processing device 1 to accurately analyze changes in user U's interests in objects.

The information processing device 1 also includes a provision unit 35 that provides information indicating the reason for the transition of the second object estimated by the estimation unit 34. This allows the information processing device 1 to provide information indicating the reason for the transition of the second object.

The interest space is also formed based on a trained model that has learned the characteristics of each of a plurality of objects, regarding two or more objects that satisfy predetermined conditions as having similar characteristics. This allows the information processing device 1 to more accurately analyze the changes in user U's interests in objects.

The information processing device 1 also includes a learning unit 32 that generates a trained model using multiple search terms used by multiple users U. This allows the information processing device 1 to more accurately analyze changes in user U's interests in subjects.

The above describes in detail the embodiments of the present application based on the drawings, but this is merely an example, and the present invention can be implemented in other forms that incorporate various modifications and improvements based on the knowledge of those skilled in the art, including the aspects described in the Disclosure of the Invention section.

Furthermore, the "section, module, unit" mentioned above can be read as "means" or "circuit." For example, an acquisition unit can be read as an acquisition means or an acquisition circuit.

REFERENCE SIGNS LIST 1 Information processing device 2, _{2 1} to 2 _n , 3 Terminal device 10 Communication unit 11 Storage unit 12 Processing unit 20 Search information storage unit 21 User information storage unit 22 Content storage unit 23 Interest space information storage unit 30 Reception unit 31 Search unit 32 Learning unit 33 Identification unit 34 Estimation unit 35 Provision unit 40 First identification unit 41 Second identification unit 100 Information processing system N Network

Claims (9)

An identification unit that identifies a second object whose vector change is similar to that of the first object in an interest space in which each of a plurality of objects including a first object is represented by a vector and the vector is updated every predetermined period;
an estimation unit that estimates a reason for the change in the second object based on a commonality between the first object and the second object ,
The interest space is
The database includes vectors of the plurality of objects obtained by inputting information of the plurality of objects into a trained model that has learned the characteristics of each of the plurality of search terms, which are two or more search terms that satisfy predetermined conditions among the plurality of search terms used by a plurality of users, as having similar characteristics.
1. An information processing device comprising: The estimation unit
The information processing device according to claim 1, characterized in that the reason for the change in the second object is estimated by using at least one of the category, specifications, and brand of the first object and the second object as a commonality between the first object and the second object. The identification unit
The information processing device according to claim 1 or 2, characterized in that an object having a predetermined vector change based on a vector indicating the interests in the interest space of a user to be determined among the plurality of users is identified as the first object. The identification unit
The information processing apparatus according to claim 3 , wherein an object that approaches the reference in the space of interest and falls within a range similar to the reference is identified as the first object. The identification unit
The information processing device according to claim 3 or 4, wherein an object that moves away from the reference in the space of interest from within a range similar to the reference to outside the range is identified as the first object. The information processing device according to any one of claims 1 to 5, further comprising a providing unit that provides information indicating a reason for the change in the second object estimated by the estimation unit. The information processing device according to any one of claims 1 to 5, further comprising a learning unit that generates the trained model using a plurality of search terms used by the plurality of users. 1. A computer-implemented information processing method, comprising:
An identification step of identifying a second object whose vector change is similar to that of the first object in an interest space in which each of a plurality of objects including a first object is represented by a vector and the vector is updated every predetermined period;
an estimation step of estimating a reason for the change in the second object based on a commonality between the first object and the second object ,
The interest space is
The database includes vectors of the plurality of objects obtained by inputting information of the plurality of objects into a trained model that has learned the characteristics of each of the plurality of search terms, which are two or more search terms that satisfy predetermined conditions among the plurality of search terms used by a plurality of users, as having similar characteristics.
1. An information processing method comprising: an identification step of identifying a second object whose vector change is similar to that of the first object in an interest space in which each of a plurality of objects including a first object is represented by a vector and the vector is updated every predetermined period;
an estimation procedure for estimating a reason for the change in the second object based on a commonality between the first object and the second object ;
The interest space is
The database includes vectors of the plurality of objects obtained by inputting information of the plurality of objects into a trained model that has learned the characteristics of each of the plurality of search terms, which are two or more search terms that satisfy predetermined conditions among the plurality of search terms used by a plurality of users, as having similar characteristics.
An information processing program characterized by: