Suggestion device, suggestion method, and program
The suggestion device enhances persuasiveness by calculating content scores based on user values and scene/situation matches, and providing reasons for suggestions, addressing the lack of rationale in existing systems.
Patent Information
- Application Number
- JP2023002332
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2023-01-11
- Publication Date
- 2024-11-11
- Estimated Expiration
- 2043-01-11
AI Technical Summary
Existing technologies fail to effectively convey the reasoning behind suggested content to users, making the persuasiveness of suggested lists inadequate.
A suggestion device and method that estimates user values, scene, and situation to calculate content scores based on matching attributes, and provides a list of content with reasons for inclusion, using a system that includes an estimation content providing unit, operation reception unit, scene situation estimating unit, and suggestion unit to enhance persuasiveness.
Improves the persuasiveness of suggested content lists by providing users with a clear rationale for the suggested content, enhancing user engagement and satisfaction.
Smart Images

Figure 0007580502000001 
Figure 0007580502000002 
Figure 0007580502000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a suggestion device, a suggestion method, and a program. [Background technology]
[0002] Conventionally, a technology for providing content based on user attributes is known. This type of technology is described, for example, in Patent Document 1. It is described that the interest analysis device described in Patent Document 1 calculates the importance of content based on user attribute information and interest information indicating that the user has shown interest in the content. In addition, it is described that the travel support system described in Patent Document 2 creates a plan for the user's travel, including items of tourist spots to be passed through and used, transportation means, and routes, based on input information including condition information from a user terminal and user attribute information, in order to support the user's travel. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] Patent Publication No. 2021-157206 [Patent Document 2] JP 2016-139263 A Summary of the Invention [Problem to be solved by the invention]
[0004] The technology described in Patent Document 1 calculates the importance of content based on user attribute information, and the technology described in Patent Document 2 describes creating a plan proposal based on user attribute information. However, the technologies described in Patent Documents 1 and 2 provide content based on input information including user attribute information and condition information from a user terminal, so there is a problem that even if a list of searched content is suggested, it is difficult to convey the reason for the suggestion to the user.
[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a suggestion device, a suggestion method, and a program that can improve the persuasiveness of a list of suggested content. [Means for solving the problem]
[0006] (1) One aspect of the present invention is a content providing unit for estimation that provides a plurality of pieces of content for estimation including images and catch phrases specialized for an application field, which is one field in which a suggestion device realizes an information provision service; Any a scene situation estimation unit that estimates a scene or situation from information including a spot specified from user's location information, a person accompanying the user, a time period, and weather; a reception unit that receives a search request from a user; a search unit that searches for content based on the search request received by the reception unit; and a content score total based on each of the user's multiple values that is higher for content having a content attribute that matches the user's values, and the estimated scene or situation for the user. a suggestion unit that calculates a higher score total for content based on a scene or situation for content with content attributes that match the situation, and transmits to the user terminal device content including a list of content searched by the search unit that has a high score obtained by adding up the score totals of content based on each of the user's multiple values and the score totals of content based on the scene or situation in which the user is placed, and the suggestion unit creates a reason for each piece of content included in the list of contents based on the score total, and transmits information indicating the reason to the user terminal device of the user.
[0007] (2) One aspect of the present invention includes a step of providing a plurality of estimation contents including an image and a catch phrase specialized for an application field, which is one field in which the suggestion device realizes an information provision service; Any a step of receiving a user's operation to select a desired one from the plurality of values of the user; a step of estimating the values of a plurality of users in the field of application based on the received operation; a step of estimating the scene or situation in which the user is placed based on information including a spot identified from the user's location information, the user's companion, the time of day, and the weather; a step of receiving a search request from the user; a step of searching for content based on the search request; a step of calculating a higher score distribution for content based on each of the plurality of values of the user for content that matches the values of the user, and a step of calculating a higher score distribution for content based on a scene or situation for content that matches the content attribute estimated for the user, and a step of transmitting content including a list of content having a high score obtained by adding up the score distribution for content based on each of the plurality of values of the user and the score distribution for content based on the scene or situation in which the user is placed among the searched contents to a user terminal device; and a step of creating a reason for the inclusion of each content in the list of contents based on the score distribution for each content, and transmitting information indicating the reason to the user terminal device of the user.
[0008] (3) One aspect of the present invention includes a step of providing a plurality of estimation contents including images and catch phrases specialized for an application field, which is one field in which the suggestion device realizes an information provision service, to a computer of an information processing device; Anya step of receiving a user's operation to select a desired one from the list of values of the user; a step of estimating the values of multiple users in the field of application based on the received operation; a step of estimating the scene or situation in which the user is placed based on information including a spot identified from the user's location information, the user's companions, the time of day, and the weather; a step of receiving a search request from the user; a step of searching for content based on the search request; a step of calculating a higher score total for content based on each of the user's multiple values for content that matches the user's values, and a step of calculating a higher score total for content based on a scene or situation for content that matches the estimated scene or situation for the user, and a step of transmitting content including a list of content having a high score obtained by adding up the score totals for content based on each of the user's multiple values and the score totals for content based on the scene or situation in which the user is placed among the searched contents to a user terminal device; and a step of creating a reason for the inclusion of each content in the list of contents based on the score total, and transmitting information indicating the reason to the user terminal device of the user. Effect of the Invention
[0009] According to one aspect of the present invention, it is possible to improve the acceptability of a list of suggested content. [Brief description of the drawings]
[0010] [Figure 1] 1 is a block diagram showing a configuration example of a suggestion system according to an embodiment. [Diagram 2] FIG. 2 is a diagram illustrating an example of an overall process of a suggestion system according to an embodiment. [Diagram 3] FIG. 2 is a diagram showing an example of content for value estimation in the embodiment. [Figure 4]11 is a diagram for explaining an example of a process for updating a user's sense of values based on the user's behavior in an embodiment. FIG. [Diagram 5] 11 is a diagram for explaining a process of updating a value by a suggestion device according to an embodiment. FIG. [Figure 6] 11 is a flowchart showing an example of a process for making suggestions according to an embodiment. [Figure 7] FIG. 2 is a diagram for explaining the overall processing of the suggestion system according to the embodiment. [Figure 8] FIG. 1 is a diagram illustrating an overview of a first embodiment. [Figure 9] FIG. 2 is a diagram for explaining a process of the suggestion system in the first embodiment. [Figure 10] FIG. 2 is a diagram for explaining an overview of a process for estimating a spot visited by a user in the first embodiment. [Figure 11] FIG. 2 is a diagram showing an example of elements constituting a scene and a situation in the first embodiment. [Figure 12] FIG. 2 is a diagram showing an overview of a process for predicting a next scene from a current scene in the first embodiment. [Figure 13] FIG. 2 is a diagram for explaining an example of a learning process for dimensionality reduction processing of scene and situation data in the first embodiment. [Figure 14] FIG. 11 is a diagram for explaining another example of the learning process for the dimensionality reduction process of the scene and situation data in the first embodiment. [Figure 15] FIG. 4 is a diagram for explaining an example of a dimensionality compression process of scene / situation data in the first embodiment. [Figure 16] FIG. 2 is a diagram illustrating an example of a process of estimating a scene and a situation using a Bayesian network in the first embodiment. [Figure 17] FIG. 2 is a diagram for explaining a learning process of a probabilistic model using a Bayesian network in the first embodiment. [Figure 18]FIG. 2 is a diagram for explaining a process of estimating a probabilistic model using a Bayesian network in the first embodiment. [Figure 19] FIG. 11 is a diagram illustrating an example of a calculation process of a suggestion score using sequence prediction in the first embodiment. [Figure 20] 13 is a diagram illustrating an example of suggesting travel plans and content related to travel plans, spots, or word-of-mouth information of similar users when the scene or situation is similar between the current user and similar users in the second embodiment. FIG. [Figure 21] FIG. 11 is a diagram for explaining the processing of a suggestion system in the second embodiment. [Figure 22] FIG. 13 is a diagram illustrating an example of a process for extracting similar users for each category in the second embodiment. [Diagram 23] 13A and 13B are diagrams showing an example of a process for calculating user similarity in the second embodiment, in which (a) shows an example of user attribute data, and (b) shows an example of a user similarity matrix. [Figure 24] 13A and 13B are diagrams showing another example of the process of calculating user similarity in the second embodiment, in which (a) shows application operation history aggregate data, and (b) shows an example of a user similarity matrix. [Diagram 25] FIG. 11 shows another example of the process for calculating user similarity in the second embodiment, where (a) shows an example of travel plan data, (b) shows an example of content attribute aggregate data, (c) shows an example of a content attribute similarity matrix, (d) shows an example of visited area aggregate data, (e) shows an example of a visited area similarity matrix, and (f) shows an example of a user similarity matrix. [Figure 26] 13 is a diagram for explaining the similarity of a scene / situation in the scene / situation similarity score calculation according to the second embodiment; FIG. [Figure 27] FIG. 13 is a diagram illustrating an example of a process for calculating a similarity score of a scene / situation in the second embodiment. [Figure 28]FIG. 13 is a diagram illustrating an example of a procedure of a suggestion score calculation process (step S308) based on similar users and similar scenes and situations in the second embodiment. [Figure 29] FIG. 11 is a diagram illustrating word-of-mouth suggestions in the second embodiment. [Diagram 30] 13 is a diagram showing an example of a screen for displaying content on a user terminal device in the second embodiment. [Diagram 31] 13A and 13B are diagrams illustrating an overview of the third embodiment, in which (a) is a diagram showing an example of a screen for displaying content on a user terminal device, and (b) is a diagram showing an example of a screen for displaying a map and content on a user terminal device. [Diagram 32] 13 is a flowchart showing an example of a process for calculating a suggestion reason according to a value and a scene / situation by the suggestion device 200 in the third embodiment. [Diagram 33] FIG. 13 is a diagram for explaining an overview of a process for displaying a suggestion reason in the third embodiment. [Diagram 34] FIG. 13 is a diagram for explaining an example of a process for providing a suggestion reason based on a partial total of suggestion scores in the third embodiment, where (a) is a diagram showing an example of the total suggestion score for each content and a partial total of each suggestion score, (b) is a diagram showing an example of a suggestion reason table, and (c) is a diagram showing an example of a suggestion screen. [Diagram 35] 13A and 13B show an example of the relationship between the search space and the value search objective function in the fourth embodiment, where (a) is a diagram for explaining that the search for values falls into a local solution and updates become less frequent, and (b) is a diagram for explaining that the optimal solution for values can be obtained by suggesting content having unexplored values. [Diagram 36] FIG. 23 is a diagram for explaining an example of an operation of a suggestion unit in the fourth embodiment. [Figure 37] FIG. 23 is a diagram showing the relationship between a user's sense of values, an attribute score, an article title of a content, and a ranking according to fluctuation in the fourth embodiment. [Figure 38] FIG. 23 is a diagram showing the relationship between a user's values, attribute scores, values that temporarily increase the attribute scores, and article titles and rankings of content in the fourth embodiment. [Figure 39] 13 is a flowchart showing an example of a process for calculating a suggestion reason according to values and a scene / situation by the suggestion device in the fourth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0011] A suggestion device, a suggestion method, and a program to which the present invention is applied will be described below with reference to the drawings. The suggestion device, the suggestion method, and the program to which the present invention is applied are realized by a suggestion system that suggests content to a user based on the user's values, the scene in which the user is placed, the situation, etc., in response to a search request based on a user's operation, for example. Suggestion is a process of providing a user with content including a list of content. In this way, the suggestion system realizes an information provision service that is highly convenient for the user. In the following embodiment, the application field of the suggestion system is tourism and walking around town, but is not limited thereto and can be applied to any field such as shopping.
[0012] <Overall explanation of the suggestion system> First, the overall configuration and processing contents of the suggestion system in the embodiment will be described. Fig. 1 is a block diagram showing an example of a configuration of a suggestion system according to an embodiment. The suggestion system includes, for example, a user terminal device 100, a suggestion device 200, a content providing device 300, and an environmental information providing device 400. The user terminal device 100, the content providing device 300, and the environmental information providing device 400 are connected to, for example, a communication network. Each device connected to the communication network includes a communication interface such as a network interface card (NIC) or a wireless communication module (not shown in Fig. 1). The communication network includes, for example, the Internet, a wide area network (WAN), a local area network (LAN), and a cellular network.
[0013] The user terminal device 100 is a portable terminal device such as a smartphone or a tablet terminal. A UA (User Agent) such as a browser or an application program is started in the user terminal device 100. The UA is, for example, an application for receiving an information providing service provided by the suggestion device 200. The user terminal device 100 generates a search request including operation information based on a user's operation, location information of the user terminal device 100 using a GPS (Global Positioning System, Global Positioning Satellite), and a query, and transmits the search request to the suggestion device 200 or the content providing device 300. The user terminal device 100 performs display processing, operation reception processing, and the like using information representing a list of contents provided by the suggestion device 200 and contents received from the content providing device 300.
[0014] The suggestion device 200 is, for example, an information processing device that performs processing for an information provision service. The suggestion device 200 includes, for example, a reception unit 210, a behavior history acquisition unit 220, a scene / situation estimation unit 230, a value estimation unit 240, a search unit 250, a suggestion unit 260, and a database unit 270. The functional units such as the reception unit 210, the behavior history acquisition unit 220, the scene / situation estimation unit 230, the value estimation unit 240, the search unit 250, and the suggestion unit 260 are realized by a processor such as a CPU (Central Processing Unit) executing a program stored in a program memory. In addition, some or all of these functional units may be realized by hardware such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array), or may be realized by cooperation between software and hardware.
[0015] The reception unit 210 receives user operations and the like by acquiring operation information, which is information based on the user's operation, from the user terminal device 100. The behavior history acquisition unit 220 acquires behavior history information including operation information, user location information, content browsing history, favorite registration, and other behavior (action) history related to the user's values. The scene / situation estimation unit 230 acquires scene information representing a scene that influences the user's values and / or situation information representing a situation in which the user is placed. The value estimation unit 240 estimates the user's values in an application field such as tourism. The user's values are, for example, preferences or tastes in tourism. The user's values vary depending on the scene and / or situation, and therefore include values corresponding to the scene and / or situation. The value estimation unit 240 may estimate a persona corresponding to the user based on the user's values. The persona is a customer model that receives an information provision service. The value estimation unit 240 may also acquire, for example, fixed information of the user (address, gender, age, etc.) as the user attributes. The search unit 250 searches for content based on the search request received by the reception unit 210. The suggestion unit 260 suggests content from the content searched by the search unit 250 based on the scene / situation estimated by the scene / situation estimation unit 230, the values estimated by the values estimation unit 240, etc.
[0016] The database unit 270 is realized by, for example, a hard disk drive (HDD), a flash memory, an electrically erasable programmable read only memory (EEPROM), a read only memory (ROM), or a random access memory (RAM), or a hybrid storage device using a plurality of these. The database unit 270 may be realized by an external storage device accessible via various networks. An example of an external storage device is a network attached storage (NAS) device. The database unit 270 manages information required for processing by each unit in the suggestion device 200.
[0017] The content providing device 300 is, for example, a server device that provides various contents in response to a request from the user terminal device 100. The content providing device 300 provides, for example, contents related to restaurants, shops, public facilities (such as parks), transportation, and commercial facilities at a travel destination. The content providing device 300 stores content data of the content to be provided, including, for example, advertising information of restaurants, coupon information, and URLs of restaurants, in the content storage unit 300A. Content attribute information indicating the attributes of the content is added to the content data. The attributes of the content are information indicating the nature of the content, and are, for example, information corresponding to the user's values, information corresponding to the persona, and information corresponding to a scene or situation.
[0018] The environmental information providing device 400 is a server device that distributes ever-changing environmental information such as weather information, traffic information, and disaster information. The environmental information providing device 400 transmits environmental information in response to a request from, for example, the suggestion device 200. The environmental information may include user information such as current location information, spot information indicating the spot of the current location (e.g., store name, tourist spot name), and electronic money balance information.
[0019] FIG. 2 is a diagram illustrating an example of the overall process of the suggestion system according to the embodiment. First, when the user terminal device 100 installs or starts an application for receiving an information provision service, the user terminal device 100 transmits a connection request S10 to the suggestion device 200. In response to the connection request S10, the suggestion device 200 transmits content S11 for value estimation stored in the database unit 270 to the user terminal device 100. The user terminal device 100 accepts a selection operation on the content S11 for value estimation and transmits operation information S12 to the suggestion device 200. The reception unit 210 accepts the operation information S12 based on the selection operation, and the value estimation unit 240 estimates the user's values based on the selection operation and assigns the estimated values to the user information (step ST10).
[0020] After that, the user terminal device 100 transmits various operation information S13a, position information S13b, or information on scene / situation S13c to the suggest device 200. In response, the behavior history acquisition unit 220 updates the behavior history based on the operation information S13a, and the value estimation unit 240 updates the value based on the transmitted information (step ST11). In addition, when a condition is met, such as when a preset timing arrives or when environmental information changes, the environmental information providing device 400 transmits environmental information S14 related to scene / situation to the suggest device 200. In response, the value estimation unit 240 updates the value based on the transmitted environmental information S14 (step ST12).
[0021] Thereafter, the user terminal device 100 transmits a search request S15 including a query for searching for sightseeing-related content to the suggestion device 200. The reception unit 210 receives the search request S15, the search unit 250 acquires content data S16 from the content providing device 300 in response to the search request S15, and the suggestion unit 260 calculates a suggestion score based on the user's current sense of values and the scene and situation (step ST13). The suggestion unit 260 transmits to the user terminal device 100 suggestion data S17 for displaying a content list based on the calculated suggestion score. The user terminal device 100 receives the suggestion data S17 and displays the content list. When the user terminal device 100 receives an operation resulting from displaying the content list, it transmits operation information S18 to the suggestion device 200. The operation resulting from displaying the content list includes, for example, viewing content based on selecting content included in the content list, adding the content to favorites, and adding to a plan. The value estimation unit 240 updates the value based on the transmitted operation information S18 (step ST14).
[0022] FIG. 3 is a diagram showing an example of content for value estimation in the embodiment. The content for value estimation includes a plurality of sets of images specialized for tourism and catch phrases specialized for tourism. When the user terminal device 100 receives the content for estimation, it displays a value estimation screen. The value estimation screen includes four sets of images and catch phrases corresponding to the images, and a question message such as "What do you like?" The images and catch phrases included in the content for value estimation are images and catch phrases corresponding to the attributes of the content. When the user terminal device 100 receives an operation to select one of the four sets, it transmits selection result information indicating the selected image and catch phrase to the suggestion device 200. The value estimation unit 240 estimates the user's values corresponding to the attributes of the content based on the selection result information, and the suggestion unit 260 transmits list information of the content to the user terminal device 100 based on the estimated values. As a result, the user terminal device 100 displays a recommended spot list screen including a list of contents. The user terminal device 100 may transmit filter information such as categories and tourist spots to the suggestion device 200, and the suggestion device 200 may transmit list information of contents matching the filter information to the user terminal device 100. The contents for value estimation may include images and catchy slogans that update the values of model plans specialized for tourism.
[0023] Although the value estimation screen shown in FIG. 3 displays four sets of images and catch phrases, the present invention is not limited to this, and eight sets may be displayed at once. Furthermore, the sets of images and catch phrases displayed at once may be displayed in a random order or position without setting a priority or ranking. Furthermore, it is preferable that each set of images and catch phrases displayed at once has the same size and is aligned on the screen so that the user can view them all at once, thereby maintaining fairness between the images.
[0024] FIG. 4 is a diagram for explaining an example of a process of updating a user's values based on the user's actions in the embodiment. As shown in FIG. 5, the value estimation unit 240 of the embodiment updates the user's values corresponding to the attributes of the content when an operation (action) is performed. When an operation such as viewing of content is performed on content belonging to "local" and "traditional", the value estimation unit 240 updates the attribute scores of the user's values corresponding to "local" and "traditional" to be high. On the other hand, since no operation is performed on content belonging to "photogenic", the value estimation unit 240 updates the attribute scores of the user's values corresponding to "photogenic" to be low. In this way, the value estimation unit 240 can more accurately grasp the user's values by reflecting the user's operations in the user's values estimated using the content for value estimation.
[0025] FIG. 5 is a diagram for explaining a process (step ST14) of updating a value by the suggestion device 200 in the embodiment. The user terminal device 100 displays a content list screen received from the suggestion device 200 (step ST20). The reception unit 210 receives operation information based on an action such as browsing a content on the content list, and acquires content data on the content operated by the operation information from the content providing device 300. The suggestion unit 260 checks the attribute of the content related to the action (corresponding to the user's value) (step ST30). The suggestion device 200 may calculate the attribute of the content from information such as a genre included in the content data. Next, the value estimation unit 240 calculates the user's value by combining the attribute of the content, the user's action, and the user's value before the update (step ST31), and updates the calculated user's value as the user's value after the update (step ST32).
[0026] FIG. 6 is a flowchart showing an example of a process of making suggestions in the embodiment. First, in response to receiving a search request from the user terminal device 100, the reception unit 210 causes the search unit 250 to acquire content data based on the search request, and the suggestion unit 260 calculates a suggestion score based on the user's values (step ST40). The suggestion unit 260 acquires operation information of the user terminal device 100 and information on a scene and / or situation from the environmental information providing device 400, acquires content data based on the scene information and / or situation information, and calculates a suggestion score (step ST41). The suggestion unit 260 increases the suggestion score for content that matches the user's values, and increases the suggestion score for content that matches the scene and / or situation in which the user is placed. The suggestion unit 260 calculates a final suggestion score by summing up the suggestion scores (step ST42). The suggestion unit 260 creates content list information so that contents with higher suggestion scores are placed at the top of the content list screen according to the final suggestion scores (step S43). The suggestion unit 260 suggests contents by transmitting the content list information to the user terminal device 100. This enables the suggestion device 200 to suggest a content list in consideration of the user's values, scene, and / or situation.
[0027] Fig. 7 is a diagram for explaining the overall processing of the suggestion system in the embodiment. In Fig. 7, "UI (user)" indicates the processing of the user terminal device 100, "processing" indicates the processing of the suggestion device 200, and "data" indicates data used in the suggestion device 200 and recorded in the database unit 270. First, the user terminal device 100 accepts an operation to select content for value estimation provided by the suggestion device 200 (step S100), and the suggestion device 200 assigns a value in the user attribute by content (image) diagnosis based on the user's operation (step S102), and updates the user attribute data.
[0028] Next, the user terminal device 100 requests a list of contents (step S104), and the suggestion unit 260 calculates a suggestion score using at least one of information such as user attribute data, content data, scene / situation data, user location information (user GPS data), operation information (UI operation log data), and travel plan data (step S106). The suggestion unit 260 calculates a suggestion score according to various values, scenes, and / or situations, and suggests content based on the calculated suggestion score. The suggestion unit 260 may predict a next scene or situation to which the current scene or situation transitions based on the current scene or situation, and suggest content corresponding to the predicted next scene or situation (first embodiment). Furthermore, the suggestion unit 260 may suggest content based on the behavior history of a similar user who has values similar to those of the user in a scene or situation that is the same as or similar to the scene or situation in which the user is placed, among the contents searched for by the search unit 250 (second embodiment). The suggestion unit 260 may provide the user with information indicating the reason for the suggested content (third embodiment). Furthermore, the suggestion unit 260 may suggest content based on values lower than a standard among the values of the user (fourth embodiment).
[0029] The suggestion unit 260 provides a list of contents to the user terminal device 100 based on the suggestion score, and the user terminal device 100 can display the list of contents and the reason for the suggested content (step S108). The user terminal device 100 accepts operations such as browsing of contents, adding of contents to favorites, or adding of contents to a travel plan based on a user operation (step S110). The suggestion device 200 updates UI operation log data, travel plan data, etc. based on the operation accepted by the user terminal device 100. The value estimation unit 240 may perform a process of updating values and a process of estimating a persona based on a user operation (step S112).
[0030] <First embodiment> Hereinafter, a first embodiment of the above-mentioned suggestion system will be described. Note that in the above-mentioned embodiment, the same parts as those in the above-mentioned suggestion system are given the same reference numerals.
[0031] As shown in FIG. 1, the suggestion device 200 in the first embodiment includes, for example, a reception unit 210 that receives a search request, a search unit 250 that searches for content based on the search request received by the reception unit 210, a scene / situation estimation unit 230 that estimates a scene or situation in which a user is currently placed and predicts a next scene or situation to which the current scene or situation will transition based on the estimated current scene or situation, and a suggestion unit 260 that suggests content corresponding to the next scene or situation predicted by the scene / situation estimation unit 230 from the content searched by the search unit 250.
[0032] FIG. 8 is a diagram for explaining an overview of the first embodiment. The suggestion system of the first embodiment estimates the scene or situation (including the user's behavior) in which the user is currently placed, and predicts the next scene or situation to which the current scene or situation will transition based on the estimated current scene or situation. This allows the suggestion system to suggest content related to spots that match the predicted next scene, for example. Specifically, if the current scene is "shopping at a shopping mall with friends in the afternoon," the suggestion system predicts that there is a high probability that the next scene will be "taking a break at a cafe," and can suggest content related to "taking a break at a cafe."
[0033] FIG. 9 is a diagram for explaining the processing of the suggestion system in the first embodiment. In FIG. 9, "UI (user)" indicates the processing of the user terminal device 100, "processing" indicates the processing of the suggestion device 200, and "data" indicates the data used in the suggestion device 200 and recorded in the database unit 270. First, the user terminal device 100 accepts a search request requesting a list of contents (step S200), and the scene / situation estimation unit 230 estimates the spots visited by the user based on the UI operation log data and the travel plan data (step S202). The travel plan data is plan information including the location and time period of the user. Next, the scene / situation estimation unit 230 compresses the dimension of the scene / situation data (step S204). The scene / situation estimation unit 230 performs the dimension compression (step S204) using a trained model constructed by the learning process (step S206).
[0034] Next, the scene / situation estimation unit 230 outputs a scene at a prediction target time point (next) that is a predetermined time from the present time, using a probabilistic model (step S210). The probabilistic model is a trained model constructed by the learning process (step S208). The scene / situation estimation unit 230 may predict a scene at a prediction target time point based on the user's future time zone and location included in the travel plan data, for example. Next, the suggestion unit 260 calculates a suggestion score using sequence prediction (step S212). Next, the suggestion unit 260 sums up the suggestion score using sequence prediction with respect to a list of suggestion scores corresponding to the list of contents searched by the search unit 250 (step S214), and provides the user terminal device 100 with suggestion data for displaying a list of contents based on the suggestion score as a result of the summation. As a result, the user terminal device 100 displays the list of contents (step S216). The suggestion unit 260 may provide the user terminal device 100 with information indicating the reason for the suggested content based on the probability of transitioning from the current scene or situation to the next scene or situation output from the graphical model (scene / situation estimation unit 230).
[0035] FIG. 10 is a diagram for explaining an outline of the process (step S202) of estimating the spot visited by the user in the first embodiment. The scene / situation estimation unit 230 identifies the spot visited by the user based on the location information as shown in FIG. 10(a). Next, the scene / situation estimation unit 230 estimates the scene or situation from the identified spot, the user's companion, the time period, the location, and the weather. For example, as shown in FIG. 10(b), the scene / situation estimation unit 230 can extract a group of scenes of "shopping mall," "family," and "12:00" by clustering, and assign a scene label of "having lunch" to the group of scenes. When the scene / situation estimation unit 230 estimates the current scene to be "shopping," as shown in FIG. 10(c), it estimates the next scene to be "taking a break" by sequence prediction processing. The sequence prediction processing may be performed using, for example, a prediction model that inputs the granularity and amount of data for identifying the current scene and outputs data for identifying the next scene. When the data output from the prediction model is "cafe," "friends," and "15:00," the scene / situation estimation unit 230 can estimate the scene of "taking a break" based on the output data and suggest content of "cafe" that has a high scene attribute score for "taking a break."
[0036] FIG. 11 is a diagram showing an example of elements constituting a scene and a situation in the first embodiment, and FIG. 12 is a diagram showing an outline of a process (steps S204, S210) for predicting a next scene from a current scene in the first embodiment. As shown in FIG. 11, the scene / situation estimation unit 230 acquires scene information including information indicating an action, a companion, a time period, and a location, and situation information including information indicating weather and temperature. The scene / situation estimation unit 230 first acquires scene information and situation information at time t-1 by collecting the scene information and situation information, classifies the scene at time t-1, and predicts and outputs a scene at time t from the classified scene at time t-1 (sequential prediction of a scene). The scene / situation estimation unit 230 may predict and output a user's behavior at time t from the scene information and situation information at time t-1 (sequential prediction of a behavior).
[0037] Fig. 13 is a diagram for explaining an example of the learning process (step S206) for the dimensionality compression process of the scene / situation data in the first embodiment. The scene / situation estimation unit 230 performs matrix decomposition on the user action history data shown in Fig. 13(a) and obtains a matrix representing the location (Where), the action (What), the feature amount of the location, and the feature amount of the action, as shown in Fig. 13(b). The scene / situation estimation unit 230 may generate a matrix representing the user ID and the feature amount shown in Fig. 13(c).
[0038] FIG. 14 is a diagram for explaining another example of the learning process (step S206) for the dimensionality compression process of the scene / situation data in the first embodiment. The scene / situation estimation unit 230 performs dimensionality compression based on the user's behavior history (e.g., location information) to pattern the location (Where) and the behavior (What), and creates a definition of a classified scene. The scene / situation estimation unit 230 defines five classified scenes including a classification number, a location (Where), an action (What), and an interpretation of a feature amount, as shown in FIG. 14, using, for example, NMF (non-negative matrix factorization) or LDA (latent Dirichlet allocation). The scene / situation estimation unit 230 inputs scene information and situation information to the trained model, and trains the parameters of the trained model so as to output one of the five classified scenes from the trained model.
[0039] FIG. 15 is a diagram for explaining an example of the dimensionality reduction process (step S204) of the scene / situation data in the first embodiment. The scene / situation estimation unit 230 adopts a classified scene using a trained model trained by the learning process (step S206). Specifically, the scene / situation estimation unit 230 inputs a place (Where) and an action (What) into the trained model, calculates the score of the classified scene from the output of the trained model, and adopts the classified scene with the highest score (for example, "taking a leisurely break") among the multiple classified scenes. As a result, the scene / situation estimation unit 230 can dimensionally compress the place (Where) and the action (What) into one classified scene. That is, the scene / situation estimation unit 230 can estimate the time period, the accompanying person, the weather, and the classified scene as the scene / situation data. The scene / situation estimation unit 230 can calculate the score of the classified scene even for an unknown combination of scene / situation data by using, for example, LDA (Latent Dirichlet Allocation).
[0040] FIG. 16 is a diagram showing an example of a process of estimating a scene and a situation using a Bayesian network in the first embodiment. The scene / situation estimation unit 230 as a prediction unit may input a current scene or situation into a graphical model expressing the directionality and probability of a plurality of scenes or situations (A, B, C), and predict the next scene or situation based on the probability of transition from the current scene or situation to the next scene or situation output from the graphical model. The scene / situation estimation unit 230 uses a Bayesian network to estimate a probability distribution for a combination of elements such as scene A transitioning to A1 or A2, scene B transitioning to B1 or B2, and scene C transitioning to C1 or C2. Similarly, the scene / situation estimation unit 230 can estimate a probability distribution for a combination of elements such as scene information and situation information.
[0041] FIG. 17 is a diagram for explaining the learning process (step S208) of the probability model using the Bayesian network in the first embodiment. The scene / situation estimation unit 230 obtains the probability distribution of the classified scene at time t from the classified scene at time t-1, and constructs a learned model that obtains the probability distribution of the classified scene at time t from the probability distribution of the classified scene at time t+1. The probability distribution of the classified scene at each time varies depending on the direction of causality between the classified scenes by creating a Bayesian network as described above. In the learning process, nodes that have no causal relationship (arrows in the figure) are deleted from the probability model, so that elements that do not contribute to the prediction of the scene or situation at the prediction target time can be found for each probability model. This allows the scene / situation estimation unit 230 to predict the next scene or situation with a high probability from the current scene or situation. The trained model is not limited to the above example, and may be a trained model that calculates the probability distribution of the current classified scene from past classified scenes, and calculates the probability distribution of the current classified scene from the probability distribution of future classified scenes.For example, processing may be performed according to a probability model (trained model) that uses classified scenes before time t-2 and classified scenes after time t+2.
[0042] FIG. 18 is a diagram for explaining the estimation process (step S210) of the probability model using the Bayesian network in the first embodiment. The scene and situation estimation unit 230 inputs the past scene and situation data dimensionally compressed in step S204 shown in FIG. 18(a), the future scene and situation data shown in FIG. 18(c), and the scene and situation data at the prediction target time shown in FIG. 18(b) to the learned model constructed by the learning process (step S208) of the probability model using the Bayesian network. The scene and situation estimation unit 230 selects a learned model from the time period, the accompanying person, and the weather as in process (1). Next, the scene and situation estimation unit 230 outputs the probability distribution of the classified scene at the prediction target time as shown in FIG. 18(d) based on the output of the learned model as in process (2).
[0043] FIG. 19 is a diagram showing an example of the calculation process (step S212) of the suggestion score using sequence prediction in the first embodiment. The suggestion unit 260 calculates the suggestion score according to the content attribute using the probability distribution of the classified scene estimated in step S210. For example, in the classified scene shown in FIG. 19(a), when the probability distribution shown in FIG. 19(b) is output, the suggestion unit 260 adds 0.7 to the suggestion score of the content having the attribute "shopping·high class" since the probability of the classification number "5" is 70%. The suggestion unit 260 adds 0.05 to the suggestion score of the content having the attribute "dining·cafe" since the probability of the classification number "1" is 5%. In this way, the suggestion unit 260 performs calculations for the content having the attribute corresponding to the classified scene so that the higher the probability, the higher the suggestion score becomes.
[0044] As described above, the suggestion system of the first embodiment can predict the next scene or situation to which the user will transition from the current scene or situation, and can suggest content corresponding to the predicted next scene or situation from among the content searched for by the search request. This allows the suggestion system to predict the next action that the user is likely to take and make suggestions.
[0045] In addition, according to the suggestion system of the first embodiment, a graphical model such as a Bayesian network is used, so that suggestions that allow the prediction process (reason) to be interpreted can be made. Furthermore, according to the suggestion system of the first embodiment, the predicted next scene or situation can be provided to the user terminal device 100 as the reason for the suggestion together with a list of contents, so that the user can recognize that the content is suggested according to the scene or situation, and the persuasiveness of the suggestion result can be improved.
[0046] Furthermore, according to the suggestion system of the first embodiment, the current scene or situation is predicted using the travel plan data, and the next scene or situation is predicted using information on future locations and time periods contained in the user's travel plan data, so that content can be suggested taking into account information in both directions, both past and future.
[0047] Furthermore, according to the suggestion system of the first embodiment, the number of combination patterns of scenes or situations can be reduced by capturing the scenes or situations as abstract concepts such as "taking a break" or "having a drinking party" through dimensionality reduction processing. As a result, the suggestion system can reduce the amount of data required to make suggestions.
[0048] <Second embodiment> A second embodiment of the above-mentioned suggestion system will be described below. Note that in the above-mentioned embodiment, the same parts as those in the above-mentioned suggestion system are given the same reference numerals.
[0049] As shown in FIG. 1, the suggestion device 200 in the second embodiment includes a value estimation unit 240 that estimates the user's values, a scene / situation estimation unit 230 that estimates the scene or situation in which the user is placed, a behavioral history acquisition unit 220 that acquires the user's behavioral history, a reception unit 210 that receives a search request from the user, a search unit 250 that searches for content based on the search request received by the reception unit 210, and a suggestion unit 260 that, when a search request is received from the user, suggests content to be provided to the user from among the contents searched by the search unit 250 based on behavioral history of similar users who have values similar to those of the user, in scenes or situations that are the same or similar to the scene or situation in which the user is placed.
[0050] FIG. 20 is a diagram for explaining an example of suggesting a travel plan or content related to the travel plan, spot, or word-of-mouth information of the similar user when the scene or situation of the user and the similar user is similar in the second embodiment. The suggestion device 200 of the second embodiment estimates a similar user having values similar to those of the user in order to suggest a list of contents to the user. The suggestion device 200 acquires the user's values, scene and situation data, travel plan data, spot information, and word-of-mouth information related to the similar user, and accumulates them in the database unit 270. When the suggestion device 200 receives a search request from the user terminal device 100 based on the operation of the user, and determines that the scene and situation data of the user and the scene and situation data of the similar user are the same or similar, the suggestion device 200 suggests the content included in the travel plan data created by the similar user, the content related to the spot information, or the content related to the word-of-mouth information.
[0051] FIG. 21 is a diagram for explaining the processing of the suggestion system in the second embodiment. In FIG. 21, "UI (user)" indicates the processing of the user terminal device 100, "processing" indicates the processing of the suggestion device 200, and "data" indicates data used in the suggestion device 200 and recorded in the database unit 270. First, the user terminal device 100 accepts a search request requesting a list of contents (step S300), and the value estimation unit 240 calculates the similarity between the user and other users by referring to at least a part of the user attribute data, UI operation log data, travel plan data, and content data browsed by the user (step S302). The reception unit 210 acquires travel plan data including the location and time period of the user, and the value estimation unit 240 may estimate the similarity between the user and similar users based on the travel plan data created by the user and the action history of the similar users. The value estimation unit 240 determines users whose similarity is equal to or higher than a standard as similar users.
[0052] Next, the scene / situation estimation unit 230 refers to at least a part of the UI operation log data and the travel plan data to calculate the similarity between the scene / situation of the user and the scene / situation of the similar user (step S304). The behavior history acquisition unit 220 extracts word-of-mouth information in the similar scene / situation data by the similar user based on the similar user and the similar scene / situation data (step S306). The suggestion unit 260 calculates a suggestion score based on the similar user and the similar scene / situation data (step S308). Next, the suggestion unit 260 sums up the suggestion score calculated in step S308 with the list of suggestion scores corresponding to the list of contents searched by the search unit 250 (step S312), and provides the user terminal device 100 with suggestion data for displaying a list of contents based on the summed suggestion score. As a result, the user terminal device 100 displays the word-of-mouth information extracted in step S306 and the list of contents (step S314).
[0053] FIG. 22 is a diagram showing an example of the process (step S302) of extracting similar users for each category in the second embodiment. Since user preferences differ for each category, the value estimation unit 240 extracts similar users for each category of suggested targets. A category is, for example, information indicating a user's behavior such as eating. When the value estimation unit 240 calculates similarity for all categories, it extracts user B as a similar user to the current user A. For example, the value estimation unit 240 calculates similarity focusing on the category "sightseeing" to extract user C as a similar user to the current user A.
[0054] FIG. 23 is a diagram showing an example of a process (step S302) for calculating user similarity in the second embodiment, where (a) shows an example of user attribute data, and (b) shows an example of a user similarity matrix. In the database unit 270, user attribute data for a plurality of users is stored as shown in FIG. 23(a). In the user attribute data, a value score for each category is registered for each user. In the user attribute data, the value estimation unit 240 creates a user similarity matrix as shown in FIG. 23(b) by calculating a correlation coefficient between users for each category in the user attribute data. The value estimation unit 240 extracts users with a high correlation coefficient as similar users from among the users registered in the database unit 270. In this way, the value estimation unit 240 can extract similar users based on the similarity of various values of users.
[0055] FIG. 24 is a diagram showing another example of the process (step S302) for calculating the similarity of users in the second embodiment, where (a) shows application operation history aggregate data, and (b) shows an example of a user similarity matrix. The database unit 270 stores application operation history aggregate data for multiple users as shown in FIG. 24(a). The application operation history aggregate data is updated by the behavior history acquisition unit 220, for example, by adding 0.2 when the user views content at a certain spot once, and by adding 1.0 when the user adds the content to a favorite or a travel plan. The value estimation unit 240 creates a user similarity matrix as shown in FIG. 24(b) by calculating a correlation coefficient between users for each category in the application operation history aggregate data. Since the application operation history aggregate data is sparse data with many dimensions, the correlation coefficient may be calculated after dimension compression using, for example, NMF (non-negative matrix factorization) or LDA (latent Dirichlet allocation). The value estimation unit 240 extracts users with high correlation coefficients from among users registered in the database unit 270 as similar users. In this way, the value estimation unit 240 can extract similar users based on the operation history of the user.
[0056] FIG. 25 is a diagram showing another example of the process (step S302) for calculating the similarity of users in the second embodiment, where (a) shows an example of travel plan data, (b) shows an example of content attribute aggregate data, (c) shows an example of a content attribute similarity matrix, (d) shows an example of visited area aggregate data, (e) shows an example of visited area similarity matrix, and (f) shows an example of a user similarity matrix. As shown in FIG. 25(a), the database unit 270 stores travel plan data including time periods and places (spots) for multiple users. The value estimation unit 240 creates the content attribute aggregate data shown in FIG. 25(b) by aggregating the average of the content attributes of the spots included in the travel plan shown in FIG. 25(a). The content attribute aggregate data includes the average value of the attributes of the content related to the places in the travel plan for each user. Next, the value estimation unit 240 creates the content attribute similarity matrix shown in FIG. 25(c) by calculating the cosine similarity between users. The content attribute similarity matrix includes the similarity between users. The value estimation unit 240 creates the visited area aggregate data shown in FIG. 25(d) by adding flags to the cities, towns, and villages included in the travel plan shown in FIG. 25(a). The visited area aggregate data includes the visited areas with flags for each user. Next, the value estimation unit 240 creates the visited area similarity matrix shown in FIG. 25(e) by calculating the cosine similarity between the users. The visited area similarity matrix includes the similarity between the users. The value estimation unit 240 creates the user attribute similarity matrix shown in FIG. 25(f) by calculating the geometric mean of the content attribute similarity matrix and the visited area similarity. The value estimation unit 240 extracts users with a high correlation coefficient from among the users registered in the database unit 270 as similar users. In this way, the value estimation unit 240 can extract similar users based on the similarity of the content attributes and visited areas in the travel plan.
[0057] FIG. 26 is a diagram for explaining the similarity of the scene / situation in the scene / situation similarity score calculation (step S304) of the second embodiment. The scene / situation estimation unit 230 calculates the similarity of the scene / situation by performing processing such as a similarity measurement method on the items and values of the scene / situation. If the scenes / situations of the user and the similar user match in "accompanying person" and the value matches in "with friend", the scene / situation estimation unit 230 calculates the similarity as "1", and if they are different, calculates the similarity as "0". In this way, the scene / situation estimation unit 230 can register information in the database unit 270 that corresponds to the items, values, and similarity measurement methods of the scene / situation that are set in advance, and calculate the similarity of the scene / situation of the user and the similar user according to the registered information.
[0058] 27 is a diagram showing an example of the process (step S304) of calculating the similarity score of the scene / situation in the second embodiment. The scene / situation estimation unit 230 compares the scene / situation of the target user to be suggested with the scene / situation of similar users, and calculates the similarity score of the scene / situation for each content related to the destination. The scene / situation estimation unit 230 calculates a weighted average, for example, as shown in the following formula. (Scene-situation similarity score) = Σ((individual similarity) × (weight)) / Σ(weight) As shown in Fig. 27(a), when the user is searching for a place to visit, it is assumed that the similar user's visit destination is "XX Cafe." In this case, the scene / situation estimation unit 230 calculates an individual similarity for each scene / situation as shown in Fig. 27(b) by comparing the scene / situation of the user shown in Fig. 27(a) with the scene / situation of the similar user. The scene / situation estimation unit 230 multiplies each individual similarity by a weight for each scene / situation set in advance, and calculates a scene / situation similarity score by summing up the multiplied values of all similarities, as shown below. (Scene-situation similarity score) = (1.0 x 1.0 + 0.5 x 1.0 + 0.9 x 1.0 + 1.0 x 0.2 + 1.0 x 0.3 + 0.0 x 0.3 + 0.0 x 0.8 + 0.0 x 0.5 + 0.6 x 0.5 + 1.0 x 0.5) / (1.0 + 1.0 + 1.0 + 0.2 + 0.3 + 0.3 + 0.8 + 0.5 + 0.5 + 0.5) = 0.606
[0059] FIG. 28 is a diagram showing an example of a processing procedure of the suggestion score calculation process (step S308) based on similar users and similar scenes and situations in the second embodiment. As shown in FIG. 28(a), for example, when a user A wants to search for a dining spot, it is assumed that the similarities of similar users B and C to the user A in the dining category are 0.7 and 0.1, respectively. Note that the similarities of similar users B and C to the user A in the sightseeing category as shown in FIG. 28(c) are not used in the calculation of the suggestion score. In addition, as shown in FIG. 28(d), it is assumed that the scene and situation similarities of similar user B for each of the visited destinations (○○ cafe, △△ restaurant) are 0.6 and 0.4, and the scene and situation similarities of similar user C for each of the visited destinations (○○ cafe, □□ pub) are 0.7 and 0.2. In this case, the suggestion unit 260 calculates a value (0.42) by multiplying the similarity score of similar user B (0.7) by the scene / situation similarity of similar user B at ○○ Cafe (0.6), a value (0.28) by multiplying the similarity score of similar user B (0.7) by the scene / situation similarity of similar user B at △△ Restaurant, a value (0.07) by multiplying the similarity score of similar user C (0.1) by the scene / situation similarity of similar user C at ○○ Cafe (0.7), and a value (0.02) by multiplying the similarity score of similar user C (0.1) by the scene / situation similarity of similar user C at □□ Izakaya (0.2). The suggestion unit 260 adds the value (0.42) obtained by multiplying the similarity score (0.7) of similar user B by the scene / situation similarity (0.6) of similar user B's XX cafe to the suggestion score of XX cafe, and adds the value (0.28) obtained by multiplying the similarity score (0.7) of similar user B's scene / situation similarity (0.4) of △△ diner to the suggestion score of △△ diner. The suggestion unit 260 does not add the value (0.07) obtained by multiplying the similarity score (0.1) of similar user C by the scene / situation similarity (0.7) of XX cafe of similar user C to the suggestion score of XX cafe, but adds the value (0.02) obtained by multiplying the similarity score (0.1) of similar user C by the scene / situation similarity (0.2) of similar user C's □□ izakaya to the suggestion score of □□ izakaya.This allows the suggestion unit 260 to calculate a suggestion score based on the similarity scores of similar users and the similarity scores of similar scenes / situations.
[0060] FIG. 29 is a diagram for explaining word-of-mouth suggestion (step S306) in the second embodiment. The suggestion unit 260 may suggest word-of-mouth information narrowed down by similar users, similar scenes and situations, and user ratings from among the word-of-mouth information stored in the database unit 270. For example, when a list of contents is requested by a user, the suggestion unit 260 first extracts word-of-mouth information posted by similar users, then narrows down the extracted word-of-mouth information to word-of-mouth information of similar scenes and situations, and further narrows down the word-of-mouth information narrowed down by similar scenes and situations to highly rated word-of-mouth information.
[0061] Fig. 30 is an example of a screen displaying content on the user terminal device 100 in the second embodiment. As shown in Fig. 30, the suggestion unit 260 can suggest content based on a suggestion score calculated from similar users and similar scenes and situations, and can also suggest word-of-mouth information narrowed down by similar users, similar scenes and situations, and evaluations.
[0062] As described above, according to the suggestion system of the second embodiment, content to be provided to the user is suggested based on the behavior history in the same or similar scene or situation as the scene or situation in which the user is placed, among the behavior history of similar users who have values similar to those of the user, so that the user's satisfaction with the suggested content can be improved. In addition, according to the suggestion system, since the suggestion is performed in combination with user-based collaborative filtering, the suggested content is more likely to be unexpected, and the user is more likely to make new discoveries. Furthermore, according to the suggestion system, the fact that the scene or situation of the user and the scene or situation of the similar user are the same or similar can be reflected in the suggestion score, so that user-based collaborative filtering suggestions according to the scene or situation can be performed.
[0063] In addition, according to the suggestion system of the third embodiment, content is suggested using data with little noise, such as travel plan data created by the user, so that content that matches the preferences of the user can be suggested. Furthermore, according to the suggestion system of the third embodiment, word-of-mouth information of similar users related to the suggested content is provided, so that a third-party opinion can be provided to the user, and as a result, the persuasiveness of the suggested content can be improved.
[0064] <Third embodiment> A third embodiment of the suggestion system described above will be described below. Note that in the above embodiment, the same parts as those in the suggestion system described above are given the same reference numerals.
[0065] 1, the suggestion device 200 in the third embodiment includes a value estimation unit 240 that estimates the user's values, a scene situation estimation unit 230 that estimates the scene or situation in which the user is placed, a reception unit 210 that receives a search request from the user, a search unit 250 that searches for content based on the search request received by the reception unit 210, and a suggestion unit 260 that, when a search request is received from the user, suggests content based on a score obtained by adding up a score distribution of content based on the user's values and a score distribution of content based on the scene or situation in which the user is placed among the content searched by the search unit 250. The suggestion unit 260 provides the user with information indicating the reason for the suggested content based on the score distribution.
[0066] FIG. 31 is a diagram for explaining an overview of the third embodiment, where (a) is a diagram showing an example of a screen displaying contents on the user terminal device 100, and (b) is a diagram showing an example of a screen displaying a map and contents on the user terminal device 100. When the suggestion device 200 suggests a list of contents, for example, the suggestion device 200 transmits information showing the list of contents and information showing the reason for the suggested contents to the user terminal device 100. As shown in FIG. 31(a), the user terminal device 100 can display a list of contents including the contents of XX cafe, and display text images such as "with a partner" and "recommended for photogenic lovers" as the reason for suggesting the contents of XX cafe. As shown in FIG. 31(b), the user terminal device 100 can display the contents of the cafe and the reason for suggesting the contents redundantly on the map.
[0067] Fig. 32 is a flowchart showing an example of a process of calculating a suggestion reason according to a value and a scene / situation by the suggestion device 200 in the third embodiment. In Fig. 32, "UI (user)" indicates the process of the user terminal device 100, "processing" indicates the process of the suggestion device 200, and "data" indicates data used in the suggestion device 200 and recorded in the database unit 270. First, the user terminal device 100 accepts a search request requesting a list of contents (step S400) and transmits the search request to the suggestion device 200. The search unit 250 searches for contents matching the search request and obtains a list of contents based on the search request.
[0068] The suggestion unit 260 refers to the user attribute data and at least a part of the content data for each of the contents included in the list of contents acquired by the search unit 250, and calculates a suggestion score distribution based on various values estimated by the value estimation unit 240 (step S402). Next, the suggestion unit 260 refers to the UI operation log data, the scene / situation rule engine, and at least a part of the content data for each of the contents included in the list of contents acquired by the search unit 250, and calculates a suggestion score distribution based on a combination of scenes and situations estimated by the scene / situation estimation unit 230 (step S404). Next, the suggestion unit 260 refers to the user attribute data and at least a part of the content data for each of the contents included in the list of contents acquired by the search unit 250, and calculates a suggestion score distribution for each persona corresponding to the user (step S406).
[0069] Next, the suggestion unit 260 calculates the sum of the suggestion score totals calculated in steps S402, S404, and S406 for each of the contents included in the list of contents acquired by the search unit 250 (step S408). At this time, the suggestion unit 260 may calculate the sum of the suggestion scores by taking into consideration the suggestion scores using the sequence prediction described in step S212 of Fig. 9 as the score total, and / or the suggestion scores based on similar users and similar scenes / situations described in step S308 of Fig. 21 as the score total.
[0070] Next, the suggestion unit 260 provides the user terminal device 100 with suggestion data for displaying a list of contents based on the totalized suggestion scores. The suggestion unit 260 also creates a suggestion reason based on the totalized suggestion scores, and provides the user terminal device 100 with data for displaying the suggestion reason for the contents included in the list of contents. As a result, the user terminal device 100 displays the list of contents and the suggestion reason (step S412).
[0071] In the process of displaying the suggestion reasons described above, the suggestion device 200 displays candidate suggestion reasons for each of the contents included in the list of contents in order of priority. The suggestion device 200 determines the priority of the suggestion reasons based on the score distribution of the suggestion reasons. When displaying the contents included in the list of suggested contents, the user terminal device 100 controls the number of suggestion reasons corresponding to the contents to be displayed.
[0072] FIG. 33 is a diagram for explaining an outline of a process for displaying a suggestion reason in the third embodiment. It is desirable that the suggestion device 200 provides the user with, for example, the suggestion reason that the content captures a scene / situation, that the content captures various values, or that the content captures a persona. For example, when making a suggestion capturing a scene / situation (layer), the suggestion device 200 may provide the user with a suggestion reason such as "It's sunny and the sunset is beautiful" or "The scenery is nice" when a rule such as "Suggest a sunset on a fine day" set in a rule engine of the scene / situation is activated (item, summary). Furthermore, when making a suggestion capturing various values (layer), if there is content that has a high degree of matching with the user's values, the suggestion device 200 may provide the user with a suggestion reason such as "Recommended for those who like high class" for example.
[0073] The suggestion device 200 may provide the suggestion reasons in a format that allows the user terminal device 100 to change the display order of the suggestion reasons. The suggestion device 200 may provide the user terminal device 100 with data in which sets of items and phrases representing the suggestion reasons shown in FIG. 33 are arranged in order of priority, for example, and the user terminal device 100 may select items for the suggested content and display the suggestion reasons corresponding to the items. Furthermore, the suggestion device 200 may provide the suggestion reasons for items with score totals higher than the standard in order of priority to the user terminal device 100. For example, when the score totals are A>B>C>D>E, the suggestion device 200 may: {[“It's sunny and the sunset is beautiful”, Score_minutes_total_A, Scene_situation_rule], [“Within 300m”, Score Minute B, Scene Status_Distance], [“Recommended for high class lovers”, Score C, Diverse values_values], [“With friends”, score total D, scene situation_companion], [“Ocean View”, Score E, Diverse Values_Details]} As such, information including the suggestion reason, score total, and item may be provided. For example, the suggestion device 200 considers various values, companions, and persona rules on the information collection screen, and therefore, when the score total is F>G>H, {[“Recommended for those who like high class”, score total F, diverse values_values], [“With friends”, score total G, scene situation_companion], [“Ocean View”, Score H, Diverse Values_Details]} As such, information including elements that correspond to various values, companions, and persona rules and contribute to the suggestions may be provided.
[0074] FIG. 34 is a diagram for explaining an example of a process for providing a suggestion reason based on a divisional total of the suggestion score in the third embodiment, where (a) is a diagram showing an example of the total of the suggestion score for each content and the divisional total of each suggestion score, (b) is a diagram showing an example of a suggestion reason table, and (c) is a diagram showing an example of a suggestion screen. For example, as shown in FIG. 34(a), the suggestion score of the value corresponding to one content included in the list of contents is 2.2, the suggestion score of the scene / situation is 1.5, the suggestion score of the sequence prediction is 0.7, the suggestion score of the similar user and similar scene / situation is 0.4, and the total of the suggestion scores is 4.8. Each suggestion score includes multiple score divisions. The suggestion unit 260 creates a suggestion reason table in which the score types and suggestion reasons are arranged in descending order of the score divisions, as shown in FIG. 34(b). The suggestion unit 260 transmits the suggestion reasons to the user terminal device 100 so as to display the top two suggestion reasons among the score totals included in the suggestion reason table. This allows the user terminal device 100 to display the two suggestion reasons together with the content of XX Cafe, as shown in FIG. 34(c).
[0075] The suggestion device 200 may provide a suggestion reason that is not related to the score calculation. For example, the suggestion device 200 may display a suggestion reason such as "Many users are interested in this content" for a content with a high popularity in the content list.
[0076] As described above, according to the suggestion device 200 of the third embodiment, when a search request is received, among the contents searched by the search unit 250, content is suggested based on a score obtained by adding up the score distribution of the content based on the user's values and the score distribution of the content based on the scene or situation in which the user is placed, and information indicating the reason for the suggested content can be provided to the user based on the score distribution. According to the suggestion system of the third embodiment, the value, scene, or situation that has a large contribution rate (=score distribution) to the suggestion score can be displayed on the user terminal device 100 together with a list of contents as the suggestion reason. As a result, according to the suggestion system of the third embodiment, the user can recognize that content that matches the value, scene, or situation is being suggested, and the acceptability of the suggestion result can be improved.
[0077] <Fourth embodiment> A fourth embodiment of the suggestion system described above will be described below. Note that in the above embodiment, the same parts as those in the suggestion system described above are given the same reference numerals.
[0078] As shown in FIG. 1, the suggestion device 200 in the fourth embodiment includes a reception unit 210 that receives a search request from a user, a search unit 250 that searches for content based on the search request received by the reception unit 210, a behavioral history acquisition unit 220 that acquires a behavioral history of the user, a value estimation unit 240 that updates attribute scores corresponding to each of a plurality of user values based on the behavioral history acquired by the behavioral history acquisition unit 220 and estimates the user's values based on the attribute scores, and a suggestion unit 260 that suggests content that matches the user values based on the attribute scores of the user values from among the content searched by the search unit 250, and the suggestion unit 260 suggests content by temporarily changing the attribute scores of the plurality of user values that are lower than a standard to higher values.
[0079] The suggestion device 200 in the above-described embodiment attributes the user's values to grasp the user's preferences. In order to correctly grasp the user's values, the value estimation unit 240 performs processing to estimate and update the user's values and persona based on the attributes of the content when acquiring the user's actions such as browsing content, adding to favorites, and adding to a travel plan. The suggestion device 200 updates the user's values to preferentially suggest content that matches the user's preferences.
[0080] However, there are times when a state occurs in which a particular value is difficult to update in the suggestion device 200. For example, when a user who has a high preference for photogenicity and a high preference for locality (climate of the area) frequently browses content with a photogenic value, the suggestion device 200 preferentially suggests photogenic content. In this state, even if the user has a high preference for locality, local content is not suggested, and the user may not realize that the local content matches his / her preference. In order to solve this problem, it is desirable to suggest content so as to search the entirety of the user's values that are set in advance. Therefore, the suggestion unit 260 of the fourth embodiment has a function of suggesting content that matches the user's value based on the attribute score of the user's value among the contents searched by the search unit 250, and a function of suggesting content by temporarily changing the attribute scores of a plurality of user's values that are lower than a reference to a higher value. In this way, the suggestion unit 260 can provide content having values (values that have not been explored) with attribute scores updated less frequently among a plurality of user values, and update the values that have not been explored when the user discovers the content and acts on it. In other words, the suggestion unit 260 can make serendipity suggestions that are both useful and unexpected by suggesting content as if searching the values of the user for the entire set of values that have been set in advance.
[0081] FIG. 35 is a diagram showing an example of the relationship between the search space and the value search objective function in the fourth embodiment, where (a) is a diagram showing a state where the search for values has progressed, and (b) is a diagram showing a state when content for which the value search has not progressed is suggested. In FIG. 35, the vertical axis is the value search objective function, and the horizontal axis is the search space. The value search objective function is a function whose value becomes lower as the search for values set in the suggestion system progresses, and the search space is a space that represents the values held by the user. As shown in FIG. 35(a), when content having a specific value is provided multiple times, the user's value regarding the specific value is updated multiple times, so that even if content having the specific value is further provided, the degree of update of the user's value becomes gentle, and the value objective function for the search space is in a state where a local solution is obtained. However, it is desirable to provide content having values other than the specific value and capture the original values held by the user, thereby bringing the value for the search space closer to the optimal solution. In response to this, the suggestion unit 260 provides content having values corresponding to the user's specific values that have been provided less frequently (the user's values have not been updated much), thereby bringing the values for the search space closer to the optimal solution, as shown in FIG. 35(b).
[0082] FIG. 36 is a diagram for explaining an example of the operation of the suggestion unit 260 in the fourth embodiment. For example, when the attribute score of the user value of photogenic for user A is high and the attribute score of the user value of local is low, the content having the value of photogenic is frequently suggested, but the content having the value of local is not easily suggested. In response to this, the suggestion unit 260 temporarily increases the attribute score of the user value of local, so that the content having the value of local is easily suggested. That is, the suggestion unit 260 changes the value having a low attribute score to "the same value as the value having the maximum value" by fluctuating the attribute score of the user value, and calculates the suggestion score. The suggestion unit 260 may select one of the attribute scores of the user value to be changed in a round robin manner from the values other than the attribute score of the user value having the maximum value. In this way, the suggestion unit 260 can accurately grasp the user value by including in the list of contents the content for which the exploration of the user value may not have progressed.
[0083] FIG. 37 is a diagram showing the relationship between the user's values, the attribute score, the article name of the content, and the ranking according to the fluctuation in the fourth embodiment. When the suggestion unit 260 does not fluctuate the attribute score of the user's values, a list of article names (contents) (1st to 12th) having the value "looks good" as an attribute, which has a high attribute score for user A, is suggested. For example, when the suggestion unit 260 increases the attribute score of the value "learning" by fluctuating the value, it can vary the ranking included in the article name list in the case where the attribute score of the user's values is not fluctuated. Note that the suggestion unit 260 may, for example, not fluctuate an article in a specific ranking (6th place) in the list of contents, but fluctuate the ranking between 1st and 5th place according to the fluctuation, and fluctuate the ranking between 7th and 12th place according to the fluctuation.
[0084] FIG. 38 is a diagram showing the relationship between a user's values, attribute scores, values whose attribute scores have been temporarily increased, and the article titles and rankings of contents in the fourth embodiment. For example, when the attribute score of the value "classic" is the highest among the attribute scores of the values of user A as shown in FIG. 38(a), the suggestion unit 260 suggests a list of contents from 1st to 10th place in "classic" as shown in FIG. 38(b). For example, when the attribute score of the value "local" is increased, the suggestion unit 260 can suggest the 13th content between 7th and 8th place when the attribute score of the value "classic" is the highest, as shown in "local" in FIG. 38(b).
[0085] FIG. 39 is a flowchart showing an example of a process of calculating a suggestion reason according to a value and a scene / situation by the suggestion device 200 in the fourth embodiment. In FIG. 39, "UI (user)" indicates the process of the user terminal device 100, "processing" indicates the process of the suggestion device 200, and "data" indicates data used in the suggestion device 200 and recorded in the database unit 270. First, the user terminal device 100 accepts a search request requesting a list of contents (step S500) and transmits the search request to the suggestion device 200. The search unit 250 searches for contents matching the search request and acquires a list of contents based on the search request. The suggestion unit 260 can suggest contents matching the user's value based on the attribute score of the user's value among the contents searched by the search unit 250 as in the above-mentioned embodiment. That is, the higher the attribute score of the user's value is above a standard such as the maximum value or average value, the more frequently the suggestion unit 260 can suggest contents.
[0086] In the fourth embodiment, the suggestion unit 260 refers to the user attribute data and temporarily raises the attribute score of at least one of the user's values (step S502). At this time, the suggestion unit 260 may temporarily raise the attribute score of a user's value system whose attribute score has been updated less frequently among the multiple user's values. The suggestion unit 260 may temporarily raise the attribute score of a user's value system whose attribute score has been updated less frequently among the multiple user's values.
[0087] Next, the suggestion unit 260 refers to at least a part of the content data for each of the contents included in the list of contents acquired by the search unit 250, and calculates a suggestion score total according to the various values estimated by the value estimation unit 240 (step S504). Next, the suggestion unit 260 sums up the suggestion score calculated in step S504 for the list of suggestion scores corresponding to the list of contents searched by the search unit 250 (step S506), and provides the user terminal device 100 with suggestion data for displaying a list of contents based on the suggestion score as a result of the summation. As a result, the user terminal device 100 displays the list of contents (step S508).
[0088] As described above, according to the suggestion device 200 of the fourth embodiment, the attribute scores corresponding to the plurality of user values are updated based on the behavior history acquired by the behavior history acquisition unit 220, the user values are estimated based on the attribute scores, content matching the user values is suggested based on the attribute scores of the user values, and attribute scores lower than a reference among the plurality of user values are temporarily changed to higher values to suggest content. As a result, according to the suggestion device 200, content that may not have been explored in the user values is suggested at a higher level, and thus suggestions can be made according to the user's potentially preferred values.
[0089] Although each embodiment and each variant has been described, these are merely examples and are not intended to be limiting. For example, any of the embodiments or variants, or a part of each embodiment or a part of each variant, may be combined with one or more other embodiments or one or more other variants to realize one aspect of the present invention.
[0090] In addition, programs for executing the processes of the user terminal device 100, the suggestion device 200, the content providing device 300, and the environmental information providing device 400 in this embodiment may be recorded on a computer-readable recording medium, and the programs recorded on the recording medium may be read into and executed by a computer system, thereby performing the various processes described above related to the user terminal device 100, the suggestion device 200, the content providing device 300, and the environmental information providing device 400.
[0091] Note that the "computer system" referred to here may include hardware such as the OS and peripheral devices. Furthermore, if a WWW system is used, the "computer system" also includes the homepage provision environment (or display environment). Furthermore, "computer-readable recording medium" refers to storage devices such as flexible disks, magneto-optical disks, ROMs, writable non-volatile memories such as flash memory, portable media such as CD-ROMs, and hard disks built into computer systems.
[0092] Furthermore, the term "computer-readable recording medium" refers to the volatile memory (e.g., DRAM (Dynamic Random Access Memory)) within a computer system that is a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. This includes devices that hold a program for a certain period of time, such as a random access memory (Random Access Memory). The program may be transmitted from a computer system in which the program is stored in a storage device to another computer system via a transmission medium, or by a transmission wave in the transmission medium.
[0093] Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The above program may also be one that realizes part of the above-mentioned functions. Furthermore, it may be one that can realize the above-mentioned functions in combination with a program already recorded in the computer system, a so-called difference file (difference program).
[0094] Although the embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and the present invention also includes designs within the scope of the gist of the present invention. [Explanation of symbols]
[0095] 100 User terminal device 200 Suggestion Device 210 Reception 220 Behavioral History Acquisition Department 230 Scene and Situation Estimation Department 240 Value Estimation Department 250 Search Department 260 Suggestion Department 270 Database Department 300 Content providing device 300A Content storage unit 400 Environmental information provision equipment
Claims
1. an estimation content providing unit that provides a plurality of estimation content including images and catch phrases specialized in an application field, which is one field in which the suggestion device realizes an information provision service; an operation receiving unit that receives a user's operation to select one of the plurality of estimation contents; a value estimation unit that estimates values of a plurality of users in the application field based on the operation received by the operation reception unit; a scene situation estimation unit that estimates a scene or situation from a spot identified from the user's position information and information including the user's companion, time period, and weather; A reception unit that receives a search request from a user; a search unit that searches for content based on the search request received by the reception unit; a suggestion unit which calculates a higher score for content based on each of the user's multiple values for a content that has a content attribute that matches the user's values, calculates a higher score for content based on a scene or situation for a content that has a content attribute that matches a scene or situation estimated for the user, and transmits to a user terminal device a content including a list of content that has a high score obtained by adding up the score for content based on each of the user's multiple values and the score for content based on a scene or situation in which the user is placed among the content searched for by the search unit, The suggestion unit creates a reason for the inclusion of each piece of content in the content list based on a score summary, and transmits information indicating the reason to the user terminal device of the user.
2. Providing a plurality of estimation contents including images and catch phrases specialized in an application field, which is one field in which the suggestion device realizes an information provision service; receiving a user operation to select one of the plurality of estimation contents; estimating values of a plurality of users in the application field based on the received operation; A step of estimating a scene or situation in which the user is placed based on information including a spot identified from the user's location information, the user's companion, a time period, and weather; receiving a search request from a user; searching for content based on the search request; a step of calculating a higher score total for content based on each of the user's multiple values for a content that has a content attribute that matches the user's values, calculating a higher score total for content based on a scene or situation for a content that has a content attribute that matches the scene or situation estimated for the user, and transmitting to a user terminal device content including a list of content that has a high score among the searched contents, calculated by adding up the score total for content based on each of the user's multiple values and the score total for content based on the scene or situation in which the user is placed; generating a reason for the inclusion of each piece of content in the list of contents based on the score distribution, and transmitting information indicating the reason to a user terminal device of the user; Suggestion methods, including:
3. The computer of the information processing device Providing a plurality of estimation contents including images and catch phrases specialized in an application field, which is one field in which the suggestion device realizes an information provision service; receiving a user operation to select one of the plurality of estimation contents; estimating values of a plurality of users in the application field based on the received operation; A step of estimating a scene or situation in which the user is placed based on information including a spot identified from the user's location information, the user's companion, a time period, and weather; receiving a search request from a user; searching for content based on the search request; a step of calculating a higher score total for content based on each of the user's multiple values for a content that has a content attribute that matches the user's values, calculating a higher score total for content based on a scene or situation for a content that has a content attribute that matches the scene or situation estimated for the user, and transmitting to a user terminal device content including a list of content that has a high score among the searched contents, calculated by adding up the score total for content based on each of the user's multiple values and the score total for content based on the scene or situation in which the user is placed; generating a reason for the inclusion of each piece of content in the list of contents based on the score distribution, and transmitting information indicating the reason to a user terminal device of the user; A program to execute.
Citation Information
Patent Citations
Information display method
JP1996087509A
On-line shopping system
JP1998021304A
Method for introducing object and method for supporting communication construction
JP2003337828A
Information processing unit
JP2012212309A
Information presentation device
JP2012256183A
