JP2012242844A - Recommendation information generation device and recommendation information generation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a recommendation information generation device and a recommendation information generation method for selecting content matching the characteristics of a similar user without using the browsing history of the user.SOLUTION: An information distribution server 200 comprises: a user characteristic vector calculation unit 204 which calculates the user characteristic vector of each user; a similar user calculation unit 209 which calculates similarity between users on the basis of the user characteristic vector of the users calculated, and selects a prescribed number S of top users who have higher similarity to one user on the basis of the calculated similarity; a new user characteristic vector calculation unit 209a which generates a new user characteristic vector on the basis of the user characteristic vectors of the prescribed number S of similar users and their similarity to the one user; and a recommendation information generation unit 210 which generates recommendation information for the one user on the basis of the generated new user characteristic vector.

Description

  The present invention relates to a recommendation information generating apparatus and a recommendation information generating method for generating recommendation information.

  A collaborative filtering method is considered in which similar users are obtained based on a comparison between the user's history and the other users' history, and content that is often viewed by similar users is recommended. For example, in the following Patent Document 1 (Japanese Patent Laid-Open No. 2002-215665), a collaborative user having a taste most similar to a target user is selected, and content to be recommended is selected based on the profile of the collaborative user. It is described. Further, in the following Patent Document 2 (Japanese Patent Laid-Open No. 2009-252177), the similarity between users is calculated based on the user characteristic vector of each user, and among the browsed contents of similar users with high similarity, It is described that the content close to the user characteristic vector is recommended.

JP 2002-215665 A JP 2009-252177 A

  However, in the technique described in Patent Document 1, since content selection is performed based on the profile of a collaborative user, only the same content as that of the collaborative user is recommended, and appropriate recommendation information cannot be provided. There was a problem.

  Also, in the technique described in Patent Document 2 described above, a recommendation is made using the browsing history of a similar user, so that a large amount of content exists and the similar user is browsing only a part of the content. Then, only a part of the content can be recommended, and the content that has not been browsed is now the target of the recommendation.

  Therefore, an object of the present invention is to provide a recommendation information generation apparatus and a recommendation information generation method that can select content that matches characteristics of a similar user without using the browsing history of the similar user.

  In order to solve the above-described problem, the recommendation information generating apparatus of the present invention is based on a vector acquisition unit that acquires a user characteristic vector of each user, and a user characteristic vector of each user acquired by the vector acquisition unit. Similarity calculation means for calculating similarity between users, selection means for selecting a plurality of similar users similar to one user based on the similarity calculated by the similarity calculation means, and selection by the selection means Calculation means for calculating a new user characteristic vector of the one user based on respective similarities with the plurality of similar users and user characteristic vectors of the plurality of similar users, and a new value calculated by the calculation means Generating means for generating recommendation information for the one user based on a user characteristic vector.

  The recommendation information generation method according to the present invention includes a vector acquisition step of acquiring a user characteristic vector of each user, and calculates a similarity between users based on the user characteristic vector of each user acquired by the vector acquisition step. A similarity calculating step, a selection step for selecting a plurality of similar users similar to one user based on the similarity calculated by the similarity calculation step, and a plurality of similar users selected by the selection step; Based on the respective similarity and user characteristic vectors of a plurality of similar users, a calculation step of calculating a new user characteristic vector of the one user, and a new user characteristic vector calculated by the calculation step, Based on the characteristic vector of the similar user selected by the selection step And a, a generation step of generating recommendation information for the one user.

  According to this invention, the similarity between users is calculated based on the user characteristic vector of each user, and a plurality of similar users similar to one user are selected based on the calculated similarity. Then, based on the respective similarities with the plurality of selected similar users and the user characteristic vectors of the plurality of similar users, a new user characteristic vector of the one user is calculated, and based on the calculated new user characteristic vector Then, recommendation information for the one user is generated. Accordingly, appropriate recommendation information can be generated with reference to the user characteristic vectors of other similar users having a user characteristic vector close to that of one user. That is, it is possible to obtain a preference that is not in the user's hobby preference but may become a favorite. Furthermore, recommendation information that matches the characteristics of similar users can be provided without using the browsing history of similar users. In addition, since the user characteristic vector of only the similar user is not used, there is no same information as the recommendation information for the similar user, and specific recommendation information for one user can be provided.

  Further, in the recommendation information generating apparatus of the present invention, the selection means selects a plurality of upper predetermined number of similar users, and the calculation means determines the user characteristics of the similar users according to the respective similarities of the plurality of similar users. A multiplication value obtained by multiplying each of the parameters constituting the vector is calculated, and a new user characteristic vector of the one user is calculated based on a total value of the multiplication values calculated for each similar user. Also good.

  According to the present invention, the multiplication value obtained by multiplying the similarity of each of the plurality of similar users by each of the parameters constituting the user characteristic vector of the similar user is calculated, and the multiplication value calculated for each similar user is calculated. By calculating a new user characteristic vector of one user based on the total value, parameters constituting the user characteristic vector according to the similarity with the similar user can be processed, and the influence of the similarity of the similar user A new user characteristic vector for one user can be generated while appropriately giving Therefore, appropriate recommendation information can be provided to one user.

  Further, in the recommendation information generating apparatus of the present invention, the calculating means calculates a new user characteristic vector of one user based on a parameter obtained by dividing the calculated total value by a total value of absolute values of similarity with similar users. You may comprise so that it may calculate.

  According to the present invention, a new user characteristic vector for one user is further calculated by dividing the sum of absolute values of similarities with similar users, thereby generating a more appropriate new user characteristic vector. be able to.

  In the recommendation information generating apparatus according to the present invention, the calculation unit may be configured to weight the multiplication value with a predetermined coefficient based on a browsing history size of a similar user.

  According to this invention, the influence on the parameter of the new user characteristic vector corresponding to the size of the browsing history is changed by weighting the multiplication value with a predetermined coefficient based on the size of the browsing history of the similar user. be able to. Therefore, a more appropriate new user characteristic vector can be generated.

  In the recommendation information generating apparatus according to the present invention, the vector acquisition unit receives a selection unit selected from a plurality of selection target items by a user operation, and the selection received by the reception unit. A storage unit that stores a satisfaction characteristic vector of the target item and a dissatisfaction characteristic vector of the selection target item that has not been received by the receiving unit, and a satisfaction characteristic vector of the selected selection target item that is stored by the storage unit is selected. User characteristic vector generation means that uses an orthogonal vector orthogonal to the separation plane for separating the unsatisfactory characteristic vector of the selection target item that has not been used as a user characteristic vector may be included.

  Further, in the recommendation information generating apparatus of the present invention, the vector acquisition unit receives a selection target item selected from a plurality of selection target items and a selection target item not selected by a user operation. And user characteristic vector generation means for generating a user characteristic vector based on a dissatisfied characteristic vector of a selection target item that has not been received by the reception means.

  According to the present invention, it is possible to generate appropriate recommendation information with reference to the user characteristic vectors of other similar users having a user characteristic vector close to one user. That is, it is possible to obtain a preference that is not in the user's hobby preference but may become a favorite. Furthermore, recommendation information that matches the characteristics of similar users can be provided without using the browsing history of similar users.

It is a system configuration figure showing the system configuration of a recommendation information distribution system. 2 is a block diagram showing a functional configuration of an information distribution server 200. FIG. 2 is a hardware configuration diagram of an information distribution server 200. FIG. FIG. 10 is an explanatory diagram illustrating an example of a menu screen displayed on the mobile phone 100. It is a conceptual diagram which shows the processing concept when producing | generating a separation plane. It is a conceptual diagram which shows the concept when calculating a user characteristic vector based on a separation plane. It is explanatory drawing which shows the memory content of the content management table 205a. It is explanatory drawing which shows the memory content of the content characteristic vector management table 207b. 2 is a block diagram showing a functional configuration of a mobile phone 100. FIG. FIG. 11 is a sequence diagram showing processing performed by mobile phone 100 and information distribution server 200 when content is distributed in response to a request from mobile phone 100. It is a flowchart which shows the calculation process of a user characteristic vector. It is a sequence diagram when distributing recommendation information. It is a flowchart which shows the production | generation process of recommendation information. It is explanatory drawing which shows the structural example of the menu screen in the log | history event a and the log | history event b. It is a figure which shows the positional relationship in the feature space of the characteristic vector of the content selected in the menu screen. It is a figure showing the characteristic vector of content in history space a and b in relative space. It is explanatory drawing when performing a process using relative space when a separation plane cannot be generated. It is a conceptual diagram which shows the process which calculates | requires the gravity center of a characteristic vector in consideration of browsing frequency. It is explanatory drawing which shows a process when producing | generating a separation plane by the NN method in the time-sequential order accessed. It is explanatory drawing which shows the parameter of the characteristic vector in the characteristic space regarding operation.

Embodiments of the present invention will be described with reference to the accompanying drawings. Where possible, the same parts are denoted by the same reference numerals, and redundant description is omitted.
FIG. 1 is a system configuration diagram illustrating a system configuration of a recommended information distribution system including a mobile phone 100 and an information distribution server 200 according to the present embodiment. In response to a request from the mobile phone 100, the information distribution server 200 functions as a recommendation information generation device. The information distribution server 200 is mobile based on respective user characteristic vectors of the user of the mobile phone 100 and users whose characteristics are similar. A new user characteristic vector of the user of the telephone 100 is generated, and content extracted based on the new user characteristic vector is distributed to the mobile phone 100. The information distribution server 200 acquires content from other content providers and distributes the acquired content in response to an access request from the mobile phone 100a to the mobile phone 100c held by each user. History is collected at this time. Further, the mobile phone 100 may be configured to transmit and acquire the access history.

  The configuration of such an information distribution server 200 will be described with reference to FIG. FIG. 2 is a block diagram showing a functional configuration of the information distribution server 200. The information distribution server 200 includes a content request receiving unit 201 (accepting unit), a recommendation request receiving unit 202, a history class separating unit 203, a user characteristic vector calculating unit 204 (user characteristic vector generating unit, vector acquiring unit), and distribution information storage. Unit 205, history storage unit 207 (storage unit), content transmission unit 208, similar user calculation unit 209 (similarity calculation unit), new user characteristic vector calculation unit 209a (selection unit, calculation unit), and recommendation information generation unit 210 ( Generating means).

  FIG. 3 is a hardware configuration diagram of the information distribution server 200. As shown in FIG. 3, the information distribution server 200 shown in FIG. 2 physically includes a CPU 11, a RAM 12 and a ROM 13 that are main storage devices, an input device 14 such as a keyboard and a mouse that are input devices, a display, and the like. The computer system includes an output device 15, a communication module 16 that is a data transmission / reception device such as a network card, and an auxiliary storage device 17 such as a hard disk. Each function described in FIG. 2 has the input device 14, the output device 15, and the communication module 16 under the control of the CPU 11 by reading predetermined computer software on the hardware such as the CPU 11 and the RAM 12 shown in FIG. 3. This is realized by reading and writing data in the RAM 12 and the auxiliary storage device 17. Hereinafter, each functional block will be described based on the functional blocks shown in FIG.

  The content request receiving unit 201 is a part that acquires content from a content provider on the network in response to a content request from the mobile phone 100 (to 100c). The access history can be collected by receiving the URL. The collected access history (such as URL) is output to the history storage unit 207. The content request receiving unit 201 includes, as an access history, content (URL) that is actually accessed in a certain menu screen (event) and content that is displayed on the menu screen but not actually accessed. Collect separately.

  Further, the content request receiving unit 201 may receive not only the access request but also access history information collected in the mobile phone 100 (˜100c) and output it to the history storage unit 207. Similarly, in this case, the cellular phone 100 (˜100c) is configured to distinguish the content selected on a certain menu screen from the content not selected and transmit it to the information distribution server 200 as access history information.

  The recommendation request receiving unit 202 is a part that receives a distribution request for recommendation information from the mobile phone 100 together with the user ID of the user of the mobile phone 100. When the recommendation request receiving unit 202 receives the distribution request, the recommendation request receiving unit 202 notifies the similar user calculation unit 209 to that effect, and instructs to extract similar users. As a modification, the fact that there is a recommendation request is output to the user characteristic vector calculation unit 204, the current user characteristic vector is calculated, the result is notified to the similar user calculation unit 209, and similar users are extracted. It may be.

  The history class separation unit 203 divides content into a satisfaction class (satisfaction characteristic vector) and a dissatisfaction class based on the content characteristic vector included in the browsing history information stored in the history storage unit 207 (browsing history table 207c). This is a part for generating a separation plane for separation into (unsatisfied characteristic vectors). Here, the satisfaction class refers to a class including only the actually selected content among a plurality of contents displayed in a selectable manner on a menu screen for accessing certain content. A class that contains only content that was not selected on the menu screen. The history class separation unit 203 uses the SVM method or the NN method to generate a separation plane that can separate the characteristic vector of each content into a satisfaction class and a dissatisfaction class, thereby separating these classes. it can. Here, the SVM (Support Vector Machine) method is a method in which a plane that maximizes the minimum distance between nodes of each class (between characteristic vectors) is a separation plane, and the NN method is the center of gravity of each class. In other words, a plane that recommends a straight line connecting the centroid of the satisfaction class and the centroid of the dissatisfaction class calculated based on the characteristic vector of each content is used as the separation plane. In the NN method, it is not always necessary to obtain a separation plane, and a straight line connected from the center of gravity of the unsatisfactory class to the center of gravity of the satisfactory class may be used as the user characteristic vector.

  Here, a specific method for generating the separation plane will be described. FIG. 4 is an explanatory diagram illustrating an example of a menu screen displayed on the mobile phone 100. As shown in FIG. 4, Node1 to Node5 are displayed. This Node indicates an item to be selected, for example, an item distinguished for each category on a menu screen on the web, or a title of a web page. For example, Node 1 is content related to automobiles, and Node 2 is content related to finance. The user can browse the corresponding content by selecting any one of Node1 to Node5.

  FIG. 5 is a conceptual diagram showing a processing concept when generating a separation plane. FIG. 5 shows the positional relationship between the characteristic vectors of each content in the feature space in the content selected by the user and the content not selected in the menu screen in FIG. As shown in FIG. 5, the parts indicated by circles are satisfactory contents (Node3, Node4), and the parts indicated by triangles are unsatisfactory contents (Node1, Node2, Node5), and are rectangular. The parts indicated by are unknown unevaluated contents (Node8, Node7). A separation plane is generated so as to separate the satisfactory content and the unsatisfied content. Here, the unevaluated content is not taken into consideration when the separation process is performed. In the figure, the feature space is represented by a two-dimensional plane for convenience. However, there are vector axes corresponding to the parameters constituting the vector, and the feature space is usually composed of vectors of 20 dimensions or more. Yes.

  The user characteristic vector calculation unit 204 is a part that calculates an orthogonal vector orthogonal to the separation plane generated by the history class separation unit 203 and sets this as a user characteristic vector. A specific example is shown in FIG. FIG. 6 is a conceptual diagram illustrating a concept when a user characteristic vector is calculated based on a separation plane. As shown in FIG. 6, a user characteristic vector w that is an orthogonal vector orthogonal to the separation plane is calculated by the user characteristic vector calculation unit 204. The user characteristic vector calculation unit 204 calculates the browsing history every time it is collected, and updates the user characteristic vector stored in the user characteristic vector management table 207a.

  The distribution information storage unit 205 is a part that stores content registered in advance by an operator or content requested by the content request reception unit 201 received from the mobile phones 100 to 100c that are users. A table 205a is stored. FIG. 7 is an explanatory diagram showing the stored contents of the content management table 205a. As shown in FIG. 7, the content management table 205a stores a user ID, a content ID, a category, a title, and a recommended text in association with each other. The user ID is identification information for specifying a user, and is unique information of the mobile phones 100 to 100c. The content ID is identification information for uniquely identifying the content, and the category indicates the information type. For example, information related to CD, information related to DVD, information related to books, information related to TV programs, etc. It is information for indicating. The title is heading information to be displayed as recommendation information. When the recommendation information is distributed, this title portion is displayed. The recommendation sentence is a main part of the recommendation information and is useful information for the user. The information stored in the content management table 205a may be input in advance by an operator, or may be information that is automatically searched and extracted from a content provider and narrowed down by a predetermined filter.

In order to extract a similar user u ′ similar to the user u that issued the recommendation request according to the recommendation request received by the recommendation request receiving unit 202, the similar user calculation unit 209 includes one user characteristic vector w _u This is a part for extracting user characteristic vectors w _{u ′} of other users and calculating their similarity. Specifically, the similar user calculation unit 209 calculates the inner product of one user characteristic vector w _u stored in the user characteristic vector management table 207a and the user characteristic vector w _{u ′ of} another user. Calculate similarity. Specifically, the similarity is calculated according to an equation as shown in equation (1).

  There are many other users, and the similar user calculation unit 209 sets the upper predetermined number (for example, S persons) among the plurality of other users having a high degree of similarity as similar users. The user characteristic vectors of one user and other users are normalized. Here, one user is a user of the mobile phone 100, and a plurality of similar users are users of the mobile phones 100a to 100c.

The new user characteristic vector calculation unit 209a calculates the similarity between each of the one user u and the plurality of similar users u ′ and the characteristic parameters W _{u ′, i of} each feature i constituting the user characteristic vector of the similar user u ′. Is a part for calculating a new user characteristic vector w ′ _u of one user u. Specifically, the user characteristic vector w _{u ′} of the mobile phone 100 that is the similar user u ′ and the user u of the mobile phone 100 that is the one user obtained as shown in the above formula (1) The similarity of the characteristic vector is calculated, and the top S similar users having the highest similarity are obtained.

Then, the new user characteristic vector calculation unit 209a calculates the user characteristic vector w of one user u based on the similarity sim (w _u , w _{u ′} ) between the S users of the similar user u ′ and the one user u. The characteristic parameter W _{u, i} of the feature i constituting _u is calculated as in equation (2). That is, the new user characteristic vector calculation unit 209a determines the similarity sim (w _u , w _{u ′} ) between one user u and the similar user u ′ and the characteristic parameter W _{u ′, i} of the characteristic i of the similar user u ′. And a cumulative value for similar users corresponding to the top S is calculated. Then, a cumulative value for the top S persons in the overall value of the similarity sim (w _u , w _{u ′} ) is calculated and divided. Thereby, the characteristic parameter W ′ _{u, i} in the feature i of one user u can be calculated.
This is performed for the characteristic parameters, and a new user characteristic vector w ′ _u can be derived by normalizing a vector composed of the characteristic parameters. In general, a characteristic vector is composed of several thousand to several tens of thousands of characteristic parameters. It is repeated for the number of characteristic parameters, and a new user characteristic vector w ′ _u of one user _u is generated using these characteristic parameters. can do.

  The recommendation information generation unit 210 extracts and evaluates content based on the new user characteristic vector calculated by the new user characteristic vector calculation unit 209a among the contents stored in the distribution information storage unit 205, and evaluates the evaluation. This is a part that generates recommendation information in which titles of contents are arranged in descending order.

  The history storage unit 207 stores each piece of history information, and stores a user characteristic vector management table 207a, a content characteristic vector management table 207b, and a browsing history table 207c.

  The user characteristic vector management table 207a is a part that stores the user characteristic vector calculated by the user characteristic vector calculation unit 204 in association with the user ID for specifying the user.

  The content characteristic vector management table 207b associates content characteristic vectors stored in advance in the content management table 205a by the operator with information (for example, content ID, URL, etc.) for specifying the content. The part to remember. For example, the content characteristic vector management table 207b stores information shown in FIG. As shown in FIG. 8, a plurality of parameters constituting the characteristic vector are stored in association with the content ID. Although eight parameters are shown in FIG. 8, the characteristic vector is usually composed of more parameters. In addition to the defined parameters, keywords extracted according to morphological analysis for each content may be added as parameters. For each parameter constituting the characteristic vector, a numerical value is described between 0 and 1, indicating which characteristic vector has a feature.

  The browsing history table 207c is a part for storing the destination or collected access history information received by the content request receiving unit 201. For example, the user ID, the content ID (or the URL of the content) and the content (viewed) This is a part for storing the characteristic vectors of the ones that are not viewed and the ones that are not browsed in association with each other.

  The content transmitting unit 208 is a part that transmits the content acquired in response to the content request received by the content request receiving unit 201. For example, the content transmission unit 208 transmits the content acquired in response to the content request received by the content request reception unit 201 and is stored in the distribution information storage unit 205 in response to the recommendation request reception unit 202. The recommendation information generated according to the evaluation result by the recommendation information generation unit 210 based on the content or the content described in the browsing history stored in the browsing history table 207c is transmitted.

  The information distribution server 200 configured as described above, in response to a request from the mobile phone 100, recommend information including content extracted based on the user characteristic vector of each user of the mobile phones 100a to 100c, which are similar users. Can be delivered. Therefore, the browsing history is determined by determining the degree of similarity of the user based on the user characteristic vector, generating a new user characteristic vector based on the user characteristic vector of the similar user, and distributing the content as recommendation information based on the new user characteristic vector. Even if there are few similar users, it is possible to extract content that matches the characteristics of the similar users.

  In the present embodiment, the user characteristic vector is calculated using the history class separating unit 203 and the user characteristic vector calculating unit 204 and stored in the user characteristic vector management table 207a. However, the present invention is not limited to this. Instead, the user characteristic vector may be stored in the user characteristic vector management table 207a as a fixed value. For example, the fixed value may be manually generated based on a result of a questionnaire survey or the like. In this case, the user characteristic vector stored in the user characteristic vector management table 207a may be processed so that it is always updated by replacing the generated new user characteristic vector.

  Alternatively, the user characteristic vector may be sequentially calculated based only on the characteristic vector of the unsatisfied content (unsatisfied characteristic vector) and stored in the user characteristic vector management table 207a. For example, without using the history class separation unit 203, the user characteristic vector calculation unit 204 may generate a user characteristic vector based only on unsatisfied content that has not been selected and store it in the user characteristic vector management table 207a. Good. In this case, the user characteristic vector calculation unit 204 calculates the center of gravity based only on the unsatisfactory content, stores this as a user characteristic vector, stores it in the user characteristic vector management table 207a, and the similar user calculation unit 209 determines based on this Therefore, similar users are determined.

  Next, the mobile phone 100 will be described. FIG. 9 is a block diagram showing a functional configuration of the mobile phone 100. As shown in FIG. 9, the mobile phone 100 includes a browser 101, a history transmission unit 102, a recommendation request transmission unit 103, and a distribution information display unit 104. The cellular phone 100 is configured by hardware such as a CPU, RAM, and ROM, and the CPU is configured to operate according to a stored program. Specifically, it has the same configuration as that of the information distribution server 200 and is realized by the hardware configuration shown in FIG. Each configuration will be described below.

  The browser 101 is an application part for connecting to the Internet, and accesses a site arranged on the Internet and allows a user to view content held on the site. The browser 101 is configured to hold an access history.

  The history transmission unit 102 is a part that transmits an access history (information indicating selected content and unselected content on the same menu screen) of content accessed using the browser 101. The history transmitting unit 102 is not an essential configuration when it has a function of collecting history information every time it is accessed on the information distribution server 200 side.

  The recommendation request transmission unit 103 is a part that transmits a request for recommendation information to the information distribution server 200 together with a user ID when a user operates an operation unit (not shown).

  The distribution information display unit 104 is a part that displays the menu screen and content (web page) acquired by the browser 101.

  The mobile phone 100 configured as described above enables access to the Internet using the browser 101 and holds the access history. The history transmission unit 102 accesses the request from the information distribution server 200 or at a predetermined timing. History can be sent.

  Next, operations of the mobile phone 100 and the information distribution server 200 will be described. FIG. 10 is a sequence diagram illustrating processing performed by the mobile phone 100 and the information distribution server 200 when content is distributed in response to a request from the mobile phone 100.

  As shown in FIG. 10, in response to a request from the browser 101 in the mobile phone 100, a menu screen is distributed from the information distribution server 200 by the content transmission unit 208 (S101). The user of the mobile phone 100 selects any one content (corresponding to the above-mentioned Node) from the menu screen, and when the browser 101 in the mobile phone 100 accepts this (S102), a URL for the content connection request and destination Is transmitted to the information distribution server 200 (S103).

  In the information distribution server 200, the URL is received by the content request receiving unit 201 and stored as an access history in the browsing history table 207c. Based on the access history, the user characteristic vector is calculated and stored (S104). Then, content based on the destination requested in S103 is acquired by the content transmission unit 208 from a content provider (not shown) and distributed (S105). Note that the order of the user characteristic vector calculation process and the content distribution process may be reversed.

  Here, the process will be described in detail for the user characteristic vector calculation in S104. FIG. 11 is a flowchart showing the calculation process of the user characteristic vector. As shown in FIG. 11, first, access history information indicated by an access request or access history received by the content request receiving unit 201 is stored in the browsing history table 207c as a newly added access history (S201). .

  Then, based on the access history information stored in the browsing history table 207c, the history class separation unit 203 extracts characteristic vectors of content belonging to each satisfaction class or dissatisfaction class (S202). The history class separation unit 203 generates a separation plane for separating the satisfaction class and the dissatisfaction class based on these characteristic vectors (S202). Next, an orthogonal vector orthogonal to the generated separation plane is calculated by the user characteristic vector calculation unit 204 to obtain a user characteristic vector (S204). The user characteristic vector obtained here is held in the user characteristic vector management table 207a and is used when evaluating the content stored in the distribution information storage unit 205 at the time of distribution of recommendation information.

  Next, processing when content extracted based on a user characteristic vector of a similar user in the present embodiment is distributed as recommendation information will be described. FIG. 12 is a sequence diagram when distributing recommendation information. As shown in FIG. 12, a recommendation request including a user ID is transmitted by the recommendation request transmitting unit 103 in the mobile phone 100, and the recommendation request is received by the recommendation request receiving unit 202 in the information distribution server 200 (S301). In the information distribution server 200, the new user characteristic vector calculation unit 209a generates a new user characteristic vector based on the user characteristic vector of the similar user, and the recommendation information generation unit 210 generates content based on the new user characteristic vector. An extraction process is performed (S302). Here, the recommendation information generation unit 210 calculates the degree of association with the content, and generates recommendation information in which the titles of the content are arranged in order of the degree of association. And the recommendation information produced | generated in this way is delivered by the content transmission part 208 (S303).

  Here, a more detailed process for generating the recommendation information in S302 will be described. FIG. 13 is a flowchart showing the recommendation information generation process. As shown in FIG. 13, the similar user calculation unit 209 acquires a corresponding user characteristic vector from the user characteristic vector management table 207a based on the user ID transmitted together with the recommendation request (S401). Next, the similarity user calculation unit 209 calculates the degree of similarity between the user who issued the recommendation request and another user. That is, the user characteristic vector of the user and the user characteristic vector of another user are extracted from the user characteristic vector management table 207a, and the inner product thereof is calculated. Inner products with all other users are calculated (S402).

  Next, the new user characteristic vector calculation unit 209a obtains the top S similar users with a high degree of similarity (the above inner product value is large), and acquires the characteristic vector of the similar user (S403). Then, a new user characteristic vector, which is a new characteristic vector of the user, is generated based on the multiplication value of the characteristic parameter constituting the user characteristic vector of the S similar users and the similarity (S404). Then, the recommendation information generation unit 210 extracts contents based on the generated new user characteristic vector, rearranges them in descending order of relevance, and transmits them as recommendation information to the user (mobile phone 100) (S406). This transmission process corresponds to S303 in FIG.

  As described above, the top S similar users are extracted according to the user characteristic vector, a new user characteristic vector is generated based on the similar user, content based on the new user characteristic vector is extracted, and the recommendation is recommended. It can be provided as information.

  Next, a modified example of the above-described embodiment will be described. In the above equation (2), a multiplication value obtained by multiplying the characteristic parameter of the user u ′ and the similarity between the user u and the user u ′ is calculated, and this is used to calculate a multiplication value of S users of the user u ′. Thus, a parameter based on the total value is used as one parameter of the new user characteristic vector.

  However, among the S users u ′, the size of the browsing history (the number of times of browsing) that is the basis of the user characteristic vector of each user u ′ is also different, and it may not be appropriate to handle it uniformly. It is possible.

Therefore, in the information distribution server 200 of the present modification, a new user characteristic vector may be generated by applying the following equation (3) instead of equation (2).

Here, α _{u ′} is a coefficient determined based on, for example, the size (number of browsing) of the browsing history of the user u ′. For example, it is a ratio in the size of the browsing history of the top S users u ′. The larger the browsing history, the greater the influence on the user u ′.

Thus, a more appropriate new user characteristic vector can be generated by weighting according to the number of browsing times.

<Generation of separation plane in relative space>
The above generation method of the separation plane is performed according to the characteristic vector of the content in the absolute space. However, as a method for improving the accuracy, a method using a relative space for which a relative positional relationship is obtained can be considered. Hereinafter, a method for generating the separation plane using the relative space will be described. Note that the processing configuration at this time is the same as that described above, and only the processing content in the history class separation unit 203 is different.

  FIG. 14 shows a configuration example of the menu screen in the history event a and the history event b, and FIG. 14A is a diagram showing the menu screen in the history event a. This menu screen is composed of Node1 to Node3. Yes. FIG. 14B is a diagram showing a menu screen in the history event b, and this menu screen is composed of Node 6 to Node 9. FIG. 14A shows that Node 3 has been selected, and FIG. 14B shows that Node 9 has been selected.

  FIG. 15 is a diagram showing the positional relationship in the feature space of the characteristic vector of the content selected on each menu screen shown in FIG. FIG. 15A shows the feature space (absolute space) of the content selected in the history event a, and FIG. 15B shows the feature space (absolute space) of the content selected in the history event b. . As described above, the selected content is defined as satisfactory content, and the content not selected is defined as unsatisfactory content, and can be classified into a satisfaction class and a dissatisfaction class, respectively.

  FIG. 16 is a diagram showing the characteristic vectors of the contents in the same feature space for the respective history events a and b. The characteristic vectors of the satisfactory contents when the characteristic vector of the unsatisfactory contents is taken as the origin. It is a figure showing the relative space which shows a position. For example, for relative comparison between Node 2 and Node 3 and relative comparison between Node 1 and Node 3, a feature space representing the position of Node 3 when Node 1 and Node 2 are set as the origin is used as a relative space. More specifically, the relative space can be formed by subtracting the node 1 characteristic vector from the node 3 characteristic vector. Similarly, for other Nodes, a relative space can be formed by performing a process of subtracting the characteristic vector of the target content.

  In this relative space, as shown in FIG. 16, the separation plane for separating the satisfaction class and the unsatisfied class based on the content vector of the satisfactory content and the characteristic vector of the unsatisfactory content as shown in FIG. 203. Then, the user characteristic vector calculation unit 204 calculates a user characteristic vector w that is an orthogonal vector orthogonal to the separation plane.

  As described above, even when the historical events are different, the user characteristic vector can be generated using the relative space, and the user characteristic vector with higher discriminability can be generated.

  In this way, a separation plane can be generated using the relative space, and thereby a user characteristic vector can be generated. 15 to 17, a separation plane can be generated even if handled as an absolute space, but a feature space composed of characteristic vectors that cannot generate a separation plane is also conceivable. In that case, the separation plane can be generated by using the relative space described above. The method will be described below.

  FIG. 17A shows a characteristic vector of content in an absolute space. In FIG. 17A, it is assumed that Node 1 and Node 4 are contents existing in the same event (menu screen), and Node 2 and Node 3 are contents existing in the same event. In addition, the characteristic vector (satisfaction class) of the content in which the circular one is selected, and the characteristic vector (unsatisfied class) of the content in which the triangular one is not selected

  As is clear from FIG. 17A, in the absolute space, a separation plane cannot be generated for the content of the characteristic vector having such a positional relationship. This is because the separation plane is a plane for separating the satisfaction class and the dissatisfaction class, and in the example of FIG. Therefore, it is necessary to convert to a relative space representing a relative positional relationship so that a separation plane can be generated.

  FIG. 17B is a diagram showing the characteristic vector of the content in the relative space. In the example of FIG. 17B, Node 2 is moved in parallel so that Node 1 and Node 2 are at the same position. Node3 corresponding to is translated. By converting to relative space in this way, it is possible to generate a separation plane v that can separate Node 1 (Node 2) that is unsatisfactory content from Node 3 and Node 4 that are satisfactory content.

  As described above, even in the case where the separation plane cannot be generated in the absolute space, the separation plane can be generated by converting to the relative space, and thus the user characteristic vector can be calculated.

<Method of generating separation plane considering browsing frequency>
Next, processing of the history class separation unit 203 when performing weighting processing using the content browsing frequency and generating a separation plane using the weighted characteristic vector will be described. Here, generation of a separation plane using the NN method will be described.

FIG. 18 is a conceptual diagram illustrating processing for obtaining the center of gravity of a characteristic vector in consideration of the browsing frequency, and is a conceptual diagram illustrating a feature space that represents the position of the characteristic vector of the contents c _{1 to} c ₃ . The browsing frequency of the content c ₁ is 3 times, the browsing frequency of the content c ₂ is 1 time, and the browsing frequency of the content c ₃ is 10 times. The browsing history table 207c preferably stores the browsing history of the browsed content and the non-viewing history of the content that has not been browsed, and it is necessary to store each frequency so that it can be held or calculated. It is.

なお、ｃ_ｉは閲覧されたコンテンツの特性ベクトル、ｎ_ｉは閲覧頻度である。 In this case, the centroids of the contents c _{1 to} c ₃ are calculated by the following equation (4).

Note that c _i is the characteristic vector of the browsed content, and n _i is the browsing frequency.

  By performing this calculation, the center of gravity is determined. In addition, for the content that is not shown but is not browsed, the center of gravity is obtained using the characteristic vector of the content that has not been browsed using Equation (4). Then, a straight line connecting the centroid of the browsed content to the centroid of the browsed content can be obtained as the user characteristic vector. This straight line has the same direction as the straight line orthogonal to the separation plane, and the separation plane may be calculated to obtain an orthogonal vector orthogonal to the separation plane. However, in the NN method, the separation is not necessarily performed. A plane is not required.

  Thereby, the user characteristic vector can be obtained using the browsing frequency of the content, and the user characteristic vector close to the user's preference can be obtained as compared with the case where the browsing frequency is not used. Therefore, it is possible to evaluate the content according to the user's preference.

<Method for generating separation plane in consideration of selection order>
Next, processing of the history class separation unit 203 when generating a separation plane in consideration of the order of the browsed contents will be described. In general, it can be considered that the most recently accessed content is more important than the previously accessed content or closer to the user's preference. As this time difference becomes larger, the user's preference changes, and it becomes more prominent that content accessed more recently is important. Here, processing when generating a separation plane in consideration of the time difference (time-series component) will be described.

  FIG. 19 is an explanatory diagram showing a process when the history class separation unit 203 generates a separation plane by using the NN method in the accessed time-series order, and FIG. FIG. 19B is an explanatory diagram showing the selection history of the content selected by the user, and FIG. 19C is a characteristic vector of each content in the feature space. FIG. 19D is a conceptual diagram for explaining the generation of the separation plane considering the selection order. The selection history is stored in the browsing history table 207c.

  As shown in FIG. 19A, a menu screen is configured from Node A to Node F. Here, as shown in FIG. 19B, the contents are selected in the order of Node A, Node B, and Node C. Therefore, NodeA, NodeB, and NodeC are contents classified into the satisfaction class, and NodeD, NodeE, and NodeF are contents classified into the dissatisfaction class.

  Next, with reference to FIG. 19C and FIG. 19D, a concept indicating processing when generating a separation plane will be described. As shown in FIG. 19C, NodeA to NodeF are positioned on the feature space according to their characteristic vectors. The calculation is performed so that the influence of the characteristic vector of the most recently selected content is increased. As shown in FIG. 19D, in order to obtain the center of gravity in the satisfaction class, first, the midpoint w1 between Node A and Node B is calculated. Then, a midpoint w2 between the midpoint w1 and NodeC is calculated.

  On the other hand, in the dissatisfied class, the center of gravity w3 is calculated based on NodeD, NodeE, and NodeF. Since there is no concept of selection order in the dissatisfied class, the centroid w3 based on the three nodes is calculated as in the normal processing. Then, a separation plane v in which a line connecting the middle point w2 and the center of gravity w3 is a perpendicular is generated. In the present embodiment, an orthogonal vector orthogonal to the separation plane v is used as a user characteristic vector, which is the same as a straight line connected from the center of gravity w3 toward the middle point w2. For the convenience of explanation in the embodiment, the separation plane is obtained. However, when the NN method is used, obtaining the separation plane is not essential.

  In this way, the midpoint between the characteristic vector and the characteristic vector is calculated, the midpoint between the midpoint and the characteristic vector is calculated, and the midpoint is calculated in time series in order (applying the NN method). Thus, the weight becomes smaller as the previous access history, and the center of gravity can be calculated in consideration of the time series order.

<Apply to operation menu>
By the way, in each of the above-described methods, the menu screen transmitted from the information distribution server 200 that is a website has been described as an example of the menu screen. However, the present invention is not limited to this, and for example, for the operation menu of the mobile phone 100 Then, the recommendation operation information may be displayed as the recommendation information.

  For example, a characteristic vector is assigned to each operation item in the operation menu, and a user characteristic vector related to a user operation is generated according to the user's operation history, and S similar users are extracted based on the user characteristic vector. The new user characteristic vector of the user may be generated based on the S similar users, and recommendation operation information based on the new user characteristic vector may be generated. In this case, the operation history is transmitted to the information distribution server 200 at a predetermined timing and stored. In addition, when a user wants to see an operation history of a similar user, the user can perform an operation so as to make a recommendation operation information request to that effect.

  FIG. 20 is an explanatory diagram illustrating the parameters of the characteristic vector in the feature space related to the operation. As shown in FIG. 20, “Chaku-moji”, “Videophone”, “One Seg”, “Screen Memo”, and “Text Memo” are defined as the terminal functions of the mobile phone 100. In addition, “telephone”, “Internet”, “text”, “memo”, and “television” are defined as parameters constituting the characteristic vector. Thus, a characteristic vector is defined for the function of the mobile phone 100. For example, in “Chaku-moji”, 1 is assigned to the parameters of “telephone” and “text”, indicating that the function is related to telephone and text. The distribution information storage unit 205 stores operation history information operated by a similar user, and is notified when the user performs an operation for displaying an operation menu. Then, the new user characteristic vector calculation unit 209a extracts the top S similar users based on the user characteristic vectors of the similar users, the user characteristic vector of the S similar users, and the similarity between the similar users and the users Based on the above, a new user characteristic vector of the user is generated. Then, the recommendation information generation unit 210 generates recommendation operation information based on the new user characteristic vector, and the content transmission unit 208 can transmit the recommendation operation information and display it on the mobile phone 100.

  As described above, by applying the evaluation method of the present embodiment to the operation menu and providing the recommended operation information to the user, it is possible to notify the user of an operation similar or related to the operation frequently used by the user.

  Next, operational effects of the information distribution server 200 of the present embodiment will be described. In this information distribution server 200, the user characteristic vector calculation unit 204 calculates the user characteristic vector of each user, and the similar user calculation unit 209 calculates the similarity between users based on the calculated user characteristic vector of each user. calculate. Then, the similar user calculation unit 209 selects S similar users similar to one user based on the calculated similarity, and the new user characteristic vector calculation unit 209a selects the user characteristics of the S similar users. A new user characteristic vector is generated based on the similarity between the vector and the S similar users and the target user. The recommendation information generation unit 210 can generate recommendation information for one user based on the generated new user characteristic vector.

  As a result, by extracting content using the new user characteristic vector, it is possible to generate recommendation information that matches the characteristics of the similar user without using the history information of the similar user as compared with the conventional method. be able to.

  Further, in the information distribution server 200 of the present embodiment, the new user characteristic vector calculation unit 209a multiplies each similarity of the plurality of similar users by each of the parameters constituting the user characteristic vector of the similar user. A value is calculated, and a new user characteristic vector of one user is calculated based on the total value of the multiplication values calculated for each similar user. For example, the parameter of the new user characteristic vector can be calculated according to the formula (2) or (3), and the new user characteristic vector can be calculated based on the parameter. This makes it possible to process the parameters constituting the user characteristic vector according to the degree of similarity with a similar user, and to generate a new user characteristic vector for one user while moderately affecting the degree of similarity of the similar user Can do. Therefore, appropriate recommendation information can be provided to one user.

  In the above formula (2) or (3), the process of dividing is performed by the total value of the absolute values of the similarity with similar users. However, the present invention is not limited to this. Also good.

  Further, by multiplying the multiplication value by a predetermined coefficient based on the degree of similarity with respect to the similar user or the frequency in the browsing history of the similar user, the influence can be appropriately lent.

  In addition, the content request receiving unit 201 receives selection of the content that is one selection target item from the menu screen that is a plurality of selection target items by a user operation, and the content characteristic vector management table 207b is selected. The characteristic vector of content and the characteristic vector of unselected content are stored. Then, the history class separation unit 203 generates a separation plane that is formed flat for separating the stored characteristic vector of the selected content and the characteristic vector of the unselected content, and stores the user characteristic The vector calculation unit 204 can generate a user characteristic vector based on the orthogonal vector orthogonal to the generated separation plane. Thereby, an accurate user characteristic vector can be generated.

DESCRIPTION OF SYMBOLS 100 ... Mobile phone, 100a ... Mobile phone, 101 ... Browser, 102 ... History transmission part, 103 ... Recommendation request transmission part, 104 ... Distribution information display part, 200 ... Information distribution server, 201 ... Content request reception part, 202 ... Recommendation Request receiving unit, 203 ... history class separation unit, 204 ... user characteristic vector calculation unit, 205 ... distribution information storage unit, 205a ... content management table, 207 ... history storage unit, 207a ... user characteristic vector management table, 207b ... content characteristic Vector management table, 207c ... browsing history table, 208 ... content transmission unit, 209 ... similar user calculation unit, 209a ... new user characteristic vector calculation unit, 210 ... recommendation information generation unit.

Claims (7)

Vector acquisition means for acquiring a user characteristic vector of each user;
Similarity calculation means for calculating similarity between users based on the user characteristic vector of each user acquired by the vector acquisition means;
Selection means for selecting a plurality of similar users similar to one user based on the similarity calculated by the similarity calculation means;
Calculation means for calculating a new user characteristic vector of the one user based on respective similarities with a plurality of similar users selected by the selection means and user characteristic vectors of the plurality of similar users;
Generating means for generating recommendation information for the one user based on the new user characteristic vector calculated by the calculating means;
A recommendation information generating apparatus comprising: The selecting means selects a plurality of top predetermined number of similar users,
The calculation means calculates a multiplication value obtained by multiplying the similarity of each of the plurality of similar users by each of the parameters constituting the user characteristic vector of the similar user, and calculates the multiplication value calculated for each similar user. The recommendation information generating apparatus according to claim 1, wherein a new user characteristic vector of the one user is calculated based on a total value. The said calculation means calculates the new user characteristic vector of one user based on the parameter which divided the calculated total value by the total value of the absolute value of the similarity with a similar user. The recommended information generating apparatus described.   The recommendation information generation apparatus according to claim 2, wherein the calculation unit weights the multiplication value by a predetermined coefficient based on a browsing history size of a similar user. The vector acquisition means includes
Accepting means for accepting selection of one selection target item from a plurality of selection target items by a user operation;
Storage means for storing a satisfaction characteristic vector of the selection target item received by the reception means and a dissatisfaction characteristic vector of the selection target item not received by the reception means;
A user having a user characteristic vector as an orthogonal vector orthogonal to a separation plane for separating a satisfaction characteristic vector of a selected selection target item stored in the storage means and an unsatisfactory characteristic vector of an unselected selection target item Characteristic vector generation means;
The recommendation information generation device according to claim 1, wherein the recommendation information generation device is configured as follows. The vector acquisition means includes
A receiving unit that receives a selection target item selected from a plurality of selection target items and a selection target item that has not been selected by a user operation;
User characteristic vector generating means for generating a user characteristic vector based on a dissatisfied characteristic vector of a selection target item not received by the receiving means;
The recommendation information generation device according to claim 1, wherein the recommendation information generation device is configured as follows. A vector acquisition step of acquiring a user characteristic vector of each user;
A similarity calculation step of calculating a similarity between users based on a user characteristic vector of each user acquired by the vector acquisition step;
A selection step of selecting a plurality of similar users similar to one user based on the similarity calculated in the similarity calculation step;
A calculation step of calculating a new user characteristic vector of the one user based on respective similarities with the plurality of similar users selected by the selection step and user characteristic vectors of the plurality of similar users;
Based on the new user characteristic vector calculated by the calculating step, based on the characteristic vector of the similar user selected by the selecting step, generating step for generating recommendation information for the one user;
A recommendation information generation method comprising:

Images