JP2011170471A - Social graph generation method, social graph generation device and program - Google Patents

Abstract

A social graph can be generated without active operation of users while ensuring the reliability between users.
When a social graph generation device receives evaluation request information from an evaluated user, a tag evaluation value determination unit selects a plurality of evaluation users who request setting of evaluation values related to the attribute information of the evaluated user. To do. Then, when receiving tag evaluation information in which an evaluation value is set from each selected evaluation user, the tag evaluation value determination unit 11 determines whether each evaluation user is a good intention user or a malicious user using the trust DB 32. When the number of good intention users is larger than the number of malicious users, the evaluation value of the user to be evaluated is determined. Subsequently, the social graph generation unit 13 extracts collaborative filtering using the attribute information set by the user to be evaluated and the evaluation value, thereby extracting a user highly related to the user to be evaluated, and social Generate a graph.
[Selection] Figure 1

Description

  The present invention relates to a social graph generation method, a social graph generation device, and a program for generating a social graph used in a social network king service (hereinafter referred to as “SNS”).

  A social graph is a graph showing the relationship between users participating in SNS. For example, as shown in FIG. 7, each user can be represented by a node, and the relationship between users can be represented by an edge (line).

  Conventionally, the social graph 100 is managed by each SNS, and is created by a user who participates in the SNS. Therefore, the relationship between the users of the social graph 100 is limited only within the SNS to which the user belongs, and a relationship with a new user is constructed in cooperation with the SNS other than the SNS to which the user belongs. I couldn't.

  In this regard, in the future SNS, for example, an NPO (Non Profit Organization) that summarizes the social graph 100 is created, and the NPO provides information such as an API (Application Program Interface) and data regarding the information of the social graph 100. Has been proposed (see Non-Patent Document 1). In addition, an open social API that uses a SNS by operating a common application across a plurality of websites has been proposed (see Non-Patent Document 2). According to these, as shown in FIG. 8, the social graph 100 individually managed by a plurality of SNS servers 3 (3A, 3B, 3C) is converted into a common database (social graph generation apparatus 1) common database ( DB). And the social graph production | generation apparatus 1 can manage the social graph 100 centrally because each user terminal 2 comprises API for utilizing this centralized service. Therefore, the user terminal 2 of each user accesses the social graph generation device 1 to establish a relationship even if the user belongs to another SNS without being limited to the SNS to which the user himself belongs. Is possible.

Here, the construction method of the relationship between the users on the social graph 100 in the conventional SNS will be described.
(1) First, a certain user finds a friend user or a user who has a good taste and taste by searching for a friend on the SNS.
(2) The user notifies the friend user or the like that he / she wants to have a friendship on the social graph 100.
(3) The friend user who has received the notification determines approval or non-approval for the notification.
(4) When a friend user or the like approves, the approval is reflected in the social graph 100, and the relationship (friend relationship) between the users on the social graph 100 is constructed.

  In the conventional SNS, the construction of the relationship (friend relationship) between users on the social graph 100 has been performed by the user himself / herself performing a manual operation as described above.

Brad Fitzpatrick, ‘Thoughts on the Social Graph’, [online], [searched February 9, 2010], Internet <URL: http://bradfitz.com/social-graph-problem/> Google, ‘OpenSocial’, [online], [Search February 9, 2010], Internet <URL: http://code.google.com/intl/en/apis/opensocial/>

  However, as described above, when trying to manage the social graph 100 by integrating a plurality of SNSs in a unified manner, there are the following problems.

(1) In a manual operation by a user, the user must actively search from information including not only the SNS to which the user belongs but also other SNSs to find a user who is appropriate for the user. In the centralized social graph generation apparatus 1, enormous number of users are registered. Therefore, for example, when the user searches for a user who is a partner for the purpose of building a friendship or the like, there may be cases where there are many corresponding users, and the user himself / herself may It becomes difficult to select a more appropriate user.
(2) In the manual operation by the user, the reliability of the social graph 100 is ensured by the user performing approval and non-approval individually. On the other hand, when a large number of users including users belonging to SNS other than the SNS to which the user belongs is registered, when determining individual approval or non-approval for the target user, the reliability of the target user is determined. The difficulty of judgment by the user himself increases.

  In view of such a background, the present invention has been made, and the present invention can generate a social graph without actively operating a social graph while ensuring reliability between users. It is an object to provide a method, a social graph generation device, and a program.

  In order to solve the above-mentioned problem, the invention according to claim 1 is directed to a user terminal of a plurality of users participating in each of a plurality of social networking services (SNS), a user terminal of the plurality of users, and a communication network. A social graph generation method for use in a social graph generation system configured to include a social graph generation device that generates a social graph indicating a relationship between the users who are connected and participate in one of the plurality of SNSs. The social graph generation device is (1) an evaluation value that is set as a higher value for each attribute information of the user set by the user, the stronger the relevance between the user and the attribute information is Tag information stored in association with the user, and (2) the attribute information of the user Each evaluation user who is a user who sets a value includes a storage unit for storing trust information for storing whether the evaluation user is a bona fide user or a malicious user, and the user terminal of the user to be evaluated The attribute information of the evaluation user is received, the evaluation request information including the received attribute information is transmitted to the social graph creation device, and the social graph generation device receives the evaluation request information of the user to be evaluated And registering the attribute information included in the evaluation request information as the attribute information of the user to be evaluated in the tag information, and the attribute information of the user to be evaluated among other users than the user to be evaluated A plurality of the evaluation users who request the setting of the evaluation value related to the user are selected, and the user terminal of the selected evaluation user is Transmitting tag evaluation request information including attribute information of the user to be evaluated, and receiving the tag evaluation request information from the social graph creation device by the user terminal of the selected evaluation user; Receiving the evaluation value related to the attribute information of the user to be evaluated indicated in the received tag evaluation request information, and returning the received evaluation value to the social graph generation device as tag evaluation information, The social graph generating device receives the tag evaluation information from each of the user terminals of the plurality of evaluation users, and refers to the trust information for each of the evaluation users who set the evaluation value indicated in the tag evaluation information. To determine whether the user is a good intention user or the malicious user, and the number of the good intention users is Determining an evaluation value of the attribute information by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information when there are more than the number of malicious users; and Storing the determined evaluation value of the attribute information as the evaluation value of the evaluated user in the tag information, the determined evaluation value of the evaluated user, and each of the plurality of evaluation users included in the tag evaluation information Is compared with the evaluation value set, and the evaluation user who has set the evaluation value within a predetermined range is set as the bona fide user on the basis of the determined evaluation value, and the evaluation value outside the predetermined range is set. With the evaluation user as the malicious user, the trust information is updated, the attribute information of each evaluated user indicated in the tag information, and the The collaborative filtering is performed using each evaluation value of the attribute information thus determined, a user highly related to the user to be evaluated is extracted, and the social graph is associated with the user to be evaluated and the extracted user. And a step of generating the social graph.

  The invention according to claim 4 is connected to user terminals of a plurality of users participating in each of a plurality of social networking services (SNS) via a communication network and participates in any of the plurality of SNSs. A social graph generation device that generates a social graph indicating a relationship between users, wherein the social graph generation device (1) for each attribute information of the user set by the user, Tag information that stores an evaluation value that is set higher as the association with attribute information is higher, and (2) evaluation that is a user who sets the evaluation value of the attribute information of the user A storage unit storing trust information for storing whether the evaluation user is a bona fide user or a malicious user for each user; Evaluation request information including attribute information input by the user to be evaluated is received from a user terminal of the user, and the attribute information included in the evaluation request information is registered as attribute information of the user to be evaluated in the tag information. Then, among other users other than the user to be evaluated, a plurality of the evaluation users who request setting of the evaluation value related to the attribute information of the user to be evaluated is selected, and the user terminal of the selected evaluation user is Tag evaluation request information including attribute information of the user to be evaluated is transmitted, and an evaluation value related to the attribute information of the user to be evaluated set by the user is evaluated from each of the selected user terminals of the plurality of evaluation users. For each of the evaluation users who have received the information and set the evaluation value indicated in the tag evaluation information, the trust information If the number of the bona fide users is larger than the number of the bona fide users, the evaluation of the user to be evaluated included in the received plurality of tag evaluation information A tag evaluation value determination unit that determines an evaluation value of the attribute information by obtaining an average value of the values, and stores the determined evaluation value of the attribute information as an evaluation value of the user to be evaluated in the tag information; The evaluation value of the user to be evaluated determined by the tag evaluation value determination unit is compared with the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information, with the determined evaluation value as a reference, The evaluation user who has set an evaluation value within a predetermined range is the goodwill user, and the evaluation user who has set an evaluation value outside the predetermined range is the malicious user. Using the trust value updating unit that updates the last information, the attribute information of each evaluated user indicated in the tag information, and the evaluation value of each of the determined attribute information, collaborative filtering is performed, and the evaluated user And a social graph generation unit that generates the social graph by associating the user to be evaluated and the extracted user with each other.

  By doing in this way, a social graph production | generation apparatus receives the evaluation value which the several evaluation user set about the attribute information which the to-be-evaluated user input from the user terminal of several evaluation users as tag evaluation information, Whether the evaluation user is a good intention user or a malicious user is determined based on the trust information, and when the number of good intention users is larger than the number of malicious users, the average value of the evaluation values included in the tag evaluation information is the evaluation value of the attribute information. And is stored in the tag information. And a social graph production | generation apparatus performs collaborative filtering using the attribute information of the to-be-evaluated user shown by tag information, and the determined evaluation value, extracts a user with strong relevance to an to-be-evaluated user, and a social graph Generate.

  Therefore, according to the present invention, an evaluation user is evaluated using the attribute information set by the user and the evaluation value determined by the social graph generation device while securing the reliability for each evaluation user using the trust information. Users with strong relationships can be extracted and a social graph can be generated.

  In the invention according to claim 2, the storage unit stores, as the trust information, the reliability of the evaluation user as a trust value that indicates a higher value for a reliable user, and the social graph generation device includes: In the step of determining the evaluation value of the attribute information, the trust value of each of the evaluation users who set the evaluation value indicated in the tag evaluation information is acquired with reference to the trust information, and the acquired trust value is When the trust value evaluation user is determined to be the bona fide user when the value is greater than or equal to a predetermined value, on the other hand, when the acquired trust value is smaller than the predetermined value, the trust value evaluation user is determined to be the malicious user, When the number of good-will users is larger than the number of malicious users, the evaluation of the user to be evaluated included in the plurality of received tag evaluation information In the step of determining the evaluation value of the attribute information and the step of updating the trust information by obtaining the average value of the evaluation information, the evaluation value of the determined user to be evaluated, and the plurality of the plurality of tags included in the tag evaluation information The evaluation value set by each evaluation user is compared, the trust value of the evaluation user who has set the evaluation value within a predetermined range is increased with reference to the evaluation value of the determined user to be evaluated, and the predetermined value The social graph generation method according to claim 1, wherein the trust information of the evaluation user who has set the evaluation value out of the range is decreased and the trust information is updated.

  In the invention according to claim 5, the storage unit stores the reliability of the evaluation user as a trust value indicating a higher value as a reliable user as the trust information, and the tag evaluation value determination The section acquires the trust value of each of the evaluation users who set the evaluation value indicated in the tag evaluation information with reference to the trust information, and the trust value when the acquired trust value is a predetermined value or more When the acquired trust value is smaller than a predetermined value, the evaluation user of the trust value is determined as the malicious user, and the number of good intention users is the number of the malicious users. When the number is larger than the number, the evaluation value of the attribute information is obtained by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information The trust value update unit compares the evaluation value of the evaluated user determined by the tag evaluation value determination unit with the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information. , By using the evaluation value of the determined user to be evaluated as a reference, increasing the trust value of the evaluation user who has set an evaluation value within a predetermined range, and setting the evaluation value out of the predetermined range The social graph generation device according to claim 4, wherein the trust information is decreased and the trust information is updated.

  Therefore, according to the present invention, the reliability of each evaluation user is stored in the trust information as a trust value that shows a higher value as a reliable user, and based on this trust value, whether each evaluation user is a good-will user or a malicious user Can be determined. Then, the evaluation value of the determined user to be evaluated and the evaluation value set by each evaluation user are compared, and if the value is within the predetermined range, the trust value is increased, and if the value is outside the predetermined range, the trust value is decreased. Update. By doing in this way, it can be judged whether an evaluation user is a good intention user or a malicious user using the trust value which raised further the reliability regarding evaluation of an evaluation user.

  The invention according to claim 3, when the social graph generation device determines whether the user is a good intention user or the malicious user for each of the evaluation users who set the evaluation value indicated in the tag evaluation information. When the number of malicious users is equal to or greater than the number of malicious users, among the received tag evaluation information, the tag evaluation information received from the malicious users is the number of malicious users. The social graph generation method according to claim 1 or 2, wherein the step of deleting until it becomes smaller is further executed.

  According to a sixth aspect of the present invention, when the tag evaluation value determination unit determines whether the user is a good intention user or the malicious user for each of the evaluation users who set the evaluation value indicated in the tag evaluation information. In addition, when the number of malicious users is greater than or equal to the number of malicious users, the tag evaluation information received from the malicious users among the plurality of received tag evaluation information is the number of malicious users is the good intentions. It deletes until it becomes smaller than the number of users, It was set as the social graph production | generation apparatus of Claim 4 or Claim 5 characterized by the above-mentioned.

  By doing in this way, when the social graph generation device determines whether each evaluation user is a bona fide user or a bona fide user, if the number of bona fide users is equal to or greater than the number of bona fide users, the tag from the bona fide user Delete evaluation information. Thus, the evaluation value of the user to be evaluated can be determined after ensuring the reliability of the evaluation user by increasing the number of good-will users than the number of malicious users.

  The invention according to claim 7 is a program for executing the social graph generation method according to any one of claims 1 to 3 on a general computer.

  According to such a program, the social graph generation method according to any one of claims 1 to 3 can be executed by a general computer.

  According to the present invention, it is possible to provide a social graph generation method, a social graph generation device, and a program capable of generating a social graph without actively operating the user while ensuring reliability between users. Can do.

It is a functional block diagram which shows the structural example of the social graph production | generation apparatus which concerns on this embodiment. It is a figure which shows an example of a data structure of tag DB (tag information) concerning this embodiment. It is a figure which shows an example of a data structure of trust DB (trust information) which concerns on this embodiment. It is a sequence diagram which shows the flow of a process of the social graph production | generation method which concerns on this embodiment. It is a flowchart which shows the flow of the evaluation value determination process which the tag evaluation value determination part which concerns on this embodiment performs. It is a flowchart which shows the flow of the trust value update process which the trust value update part which concerns on this embodiment performs. It is a figure for demonstrating an example of a social graph. It is a figure for demonstrating the system by which a some SNS server is centrally managed by the social graph production | generation apparatus.

  Next, modes for carrying out the present invention (referred to as “embodiments”) will be described in detail with reference to the drawings as appropriate.

<Overview of operation>
First, the operation | movement outline | summary of the social graph production | generation system containing the social graph production | generation apparatus 1 which concerns on this embodiment is demonstrated.

  As shown in FIG. 8, the social graph generation system according to the present embodiment includes a plurality of user terminals 2 participating in each of a plurality of social networking services (SNS) and the social graph generation device 1 in a communication network. Connected through. Each user terminal 2 is assumed to have a common API for transmitting and receiving information to and from the social graph generation device 1.

  First, among the plurality of users, the user terminal 2 (2A) of the user to be evaluated who is a user who requests generation of the social graph 100 (see FIG. 7) sends the evaluation request information including attribute information to the social graph generation device. 1 to send. The social graph generation device 1 selects a plurality of evaluation users who request setting of evaluation values related to the attribute information of the user to be evaluated, and sends tag evaluation request information including the attribute information of the user to be evaluated to the user terminal 2 ( 2B). Next, an evaluation value related to the attribute information of the user to be evaluated is set by each evaluation user, and is returned to the social graph generation device 1 as tag evaluation information. The social graph generation device 1 determines whether the evaluation user who transmitted the tag evaluation information is a bona fide user or a malicious user using a trust DB (trust information) 32 to be described later, and the number of bona fide users is greater than the number of malicious users. If there are many, the evaluation value of the user to be evaluated is determined. Then, by executing collaborative filtering using the attribute information set by the user to be evaluated and the evaluation value of the attribute information that has been determined, the user who is strongly related to the user to be evaluated is extracted, and the social graph 100 is displayed. Generate.

  Here, the attribute information includes various information such as hobbies, preferences, affiliations such as work and school, current position, user profile, etc., and this attribute information is set by the evaluated user. For example, a hobby is a piano or English conversation, and a preference is pasta or coffee. In the affiliation, the company name or school name, and the current position may be a place where the user terminal 2 is installed (Shinjuku, Shibuya, etc.) or a place obtained by GPS (Global Positioning System). In the profile, name, age, gender, annual income, educational background, occupation, qualification, and the like may be set as attribute information by the evaluated user. The user to be evaluated sets these pieces of attribute information in the user terminal 2 and transmits them to the social graph generation device 1 to construct a relationship on the social graph 100 with a user who is deeply related to the attribute information. Request.

  As described above, according to the social graph generation system including the social graph generation device 1 according to the present embodiment, the friend search by the user himself, which is performed by the method of building the relationship between users in the conventional SNS. In addition, it is possible to eliminate the need for manual operations such as approving or not approving friendships on the social graph 100. Further, an evaluation value is provided for the attribute information set by the user to be evaluated, and the setting of the evaluation value of the attribute information is requested to other users (evaluation users) instead of the user himself / herself. Then, the reliability of the evaluation user is registered in the trust DB (trust information) 32, and the evaluation value of the attribute information is determined when the number of good intention users is larger than the number of malicious users among the evaluation users. By doing so, the reliability of the social graph 100 is ensured.

<Configuration of social graph generation device>
Next, the configuration of the social graph generation device 1 according to the present embodiment will be specifically described. FIG. 1 is a functional block diagram illustrating a configuration example of the social graph generation device 1 according to the present embodiment. As shown in FIG. 1, the social graph generation device 1 according to the present embodiment includes a control unit 10, an input / output unit 20, and a storage unit 30.

  Here, the input / output unit 20 is realized by an input / output interface capable of inputting / outputting various information to / from the user terminal 2. The storage unit 30 is storage means realized by a hard disk or the like, and includes a tag DB (tag information) 31, a trust DB (trust information) 32, and a social graph DB 33. Details of each DB (database) will be described later. The storage unit 30 stores address information for identifying each user who participates in each SNS server 3 (3A, 3B, 3C) shown in FIG.

  The control unit 10 controls the entire social graph generation device 1 and includes a tag evaluation value determination unit 11, a trust value update unit 12, a social graph generation unit 13, and a social graph management unit 14. . The function of the control unit 10 is realized by, for example, a CPU (Central Processing Unit) developing and executing a program stored in the storage unit 30 of the social graph generating device 1 in a memory unit (not shown).

  The tag evaluation value determination unit 11 performs (1) registration of attribute information in the tag DB 31, (2) selection of an evaluation user, and (3) evaluation value determination processing described below.

(Registration of attribute information in tag DB)
The tag evaluation value determination unit 11 receives the evaluation request information including the attribute information input by the user to be evaluated from the user terminal 2A of the user to be evaluated through the input / output unit 20, and includes the attribute information included in the evaluation request information. Are registered as attribute information of the user to be evaluated in the tag DB 31 in the storage unit 30.

  FIG. 2 is a diagram illustrating an example of a data configuration of a tag DB (tag information) 31 according to the present embodiment. As shown in FIG. 2, in the tag DB 31, for each user (evaluated user), attribute information set by the evaluated user is registered as a tag. And the evaluation value regarding the attribute information of the to-be-evaluated user determined by the tag evaluation value determination unit 11 is stored. For example, the user A indicates that attribute information is set for “piano”, “English conversation”, and “pasta”. The evaluation value is set as a higher value as the relationship between the user to be evaluated and the attribute information is stronger. For example, the evaluation value is “1” to “10” for the attribute information of a user to be evaluated. Are set in 10 stages. When the evaluation value column is “-”, the attribute information is set in the tag by the tag evaluation value determination unit 11, but the evaluation value determination process is currently being performed, and the evaluation value is still Indicates that it has not been decided.

(Selection of evaluation users)
The tag evaluation value determination unit 11 selects a plurality of evaluation users who request setting of evaluation values related to the attribute information of the user to be evaluated among other users than the user to be evaluated. Then, the tag evaluation value determination unit 11 transmits tag evaluation request information including the attribute information of the user to be evaluated to the user terminal 2B of the selected evaluation user via the input / output unit 20.

  The selection of the evaluation user by the tag evaluation value determination unit 11 may be performed by, for example, the social graph DB 33 in the storage unit 30 if the user to be evaluated already has a user who has established a friendship on the social graph 100. The user may be preferentially selected as an evaluation user, and a user who sets the same or similar attribute information as the attribute information set by the user to be evaluated as his / her attribute information You may select as an evaluation user preferentially using DB31. Moreover, you may select an evaluation user from all the users quite randomly, and you may combine these.

(Evaluation value determination process)
The tag evaluation value determination unit 11 uses the input / output unit 20 to evaluate the evaluation value related to the attribute information of the user to be evaluated set by the evaluation user from each of the user terminals 2B of the plurality of evaluation users selected by the tag evaluation value. Receive as information.
Then, the tag evaluation value determination unit 11 refers to the trust DB 32 for each evaluation user who has set the evaluation value indicated in the tag evaluation information, determines whether the user is a good intention user or a malicious user, and the number of good intention users is that of the malicious user. When the number is larger than the number, the evaluation value of the attribute information is determined by obtaining the average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information. Further, the tag evaluation value determination unit 11 stores the evaluation value of the determined attribute information as the evaluation value of the user to be evaluated in the tag DB 31.

FIG. 3 is a diagram illustrating an example of a data configuration of the trust DB (trust information) 32 according to the present embodiment. As shown in FIG. 3, the trust DB 32 stores the reliability for each user as a trust value that indicates a higher value for a reliable user. This trust value is set, for example, between “−10” and “+10”. When the predetermined value is “0”, the tag evaluation value determining unit 11 sets “0” to “0” or more. A user having a trust value of +10 is determined as a bona fide user, and a user having a trust value from “−10” to “−1” which is smaller than “0” is determined as a malicious user.
In this trust DB 32, instead of providing a trust value for each user, “good intention user” or “malicious user” may be stored in association with each user. The trust DB 32 updating method will be described in the trust value updating unit 12.

The tag evaluation value determination unit 11 refers to the trust DB 32 to determine whether each evaluation user is a bona fide user or a bona fide user. When the number of bona fide users is larger than the number of bona fide users, a plurality of received tags By determining the average value of the evaluation values of the user to be evaluated included in the evaluation information, the evaluation value of the attribute information is determined.
Moreover, when the number of malicious users is equal to or greater than the number of good-will users as a result of determining the evaluation users, the tag evaluation value determination unit 11 evaluates the evaluation users determined as malicious users among the received plurality of tag evaluation information The tag evaluation information received from the user terminal 2B is deleted until the number of malicious users becomes smaller than the number of malicious users. By doing in this way, the number of good-will users can be made larger than the number of malicious users, an evaluation value can be determined, and the reliability of the evaluation value of the attribute information can be ensured. In addition, selection of the tag evaluation information of the malicious user to be deleted may be performed in order of increasing trust value of the malicious user or may be selected randomly from the malicious users.

The timing of the evaluation value determination process by the tag evaluation value determination unit 11 is not limited to when tag evaluation information is returned from all of the plurality of evaluation users who have transmitted the tag evaluation request information. The tag evaluation information may be returned from a number of evaluation users, or the tag evaluation information may be returned from a predetermined proportion of evaluation users among a plurality of evaluation users.
In addition, when the tag evaluation value determination unit 11 performs the evaluation value determination process when the predetermined number or the predetermined ratio is reached, when the number of malicious users is equal to or more than the number of good-will users as described above, The evaluation value determination process may be interrupted, wait until more tag evaluation information is returned from the evaluation user, and the evaluation value determination process may be restarted including the newly received tag evaluation information.

  Returning to FIG. 1, the trust value update unit 12 compares the evaluation value of the evaluated user determined by the tag evaluation value determination unit 11 with the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information, By using the determined evaluation value as a reference, the trust value of the evaluation user who has set the evaluation value within the predetermined range is increased, the trust value of the evaluation user who has set the evaluation value outside the predetermined range is decreased, and the trust DB 32 is stored. Update.

Specifically, for example, the trust value update unit 12 generates a normal distribution using the evaluation values set by each evaluation user, determines a predetermined range, and sets an evaluation value within the predetermined range. The trust value of the evaluated user is incremented, the trust value of the evaluated user who has set the evaluation value outside the predetermined range is decremented, and the trust DB 32 is updated.
In the trust DB 32, when “good intention user” or “malicious user” is registered in association with each evaluation user in the trust DB 32, the trust value update unit 12 sets an evaluation value within a predetermined range. The set evaluation user may be updated as a “good intention user”, and the evaluation user set with an evaluation value that is out of a predetermined range may be updated as a “malicious user”.

  Next, the social graph generation unit 13 executes collaborative filtering using the attribute information of each evaluated user indicated in the tag DB 31 and the evaluation values of the attribute information determined by the tag evaluation value determination unit 11, A user who has a strong relationship with the evaluation user is extracted, and the social graph 100 is generated by associating the user to be evaluated with the extracted user. Then, the generated social graph 100 is stored in the social graph DB 33 via the social graph management unit 14.

  This collaborative filtering is an existing well-known technique that presents information used and evaluated by other users who have similar interests and interests to the user in providing information desired by the user. The social graph generation unit 13 according to the present embodiment executes the emphasis filtering using the attribute information set by the user to be evaluated, the user who sets similar attribute information, and the evaluation value related to the attribute information, Extract users who are strongly related to users.

Next, the social graph management unit 14 manages input / output of information related to the social graph 100 to the social graph DB 33. In addition, as shown in FIG. 7, social graph DB33 is a database which memorize | stores the social graph 100 showing the relationship between the to-be-evaluated user and another user for every user (to-be-evaluated user). In the social graph DB 33, a user whose relationship is recognized in any attribute information set by the user to be evaluated and whose friendship is registered is stored in association with each user to be evaluated.
And the social graph management part 14 will be memorize | stored in social graph DB33, if the social graph call information which is an acquisition request of the social graph 100 is received via the input / output part 20 from user terminal 2A of an evaluated user, for example. The social graph 100 of the user to be evaluated is output and transmitted to the user terminal 2A of the user to be evaluated.

<Operation procedure>
Next, processing of the social graph generation method according to the present embodiment will be described along FIG. 4 with reference to FIGS.
FIG. 4 is a sequence diagram illustrating a process flow of the social graph generation method according to the present embodiment.

  First, attribute information about the user is set by the user to be evaluated (step S101), and evaluation request information including the attribute information is transmitted from the user terminal 2A of the user to be evaluated to the social graph generation device 1 (step S101). S102).

  Next, in the social graph generation device 1, the tag evaluation value determination unit 11 receives the evaluation request information via the input / output unit 20. Then, the tag evaluation value determination unit 11 registers the attribute information included in the received evaluation request information as a tag of the tag DB 31 (see FIG. 2) (step S103).

  Subsequently, the tag evaluation value determination unit 11 selects a plurality of evaluation users who request setting of evaluation values related to the attribute information of the user to be evaluated from other users other than the user to be evaluated (step S104).

  Then, the tag evaluation value determination unit 11 transmits tag evaluation request information including attribute information of the user to be evaluated to the selected user terminals 2B of the plurality of evaluation users (step S105).

  Next, the user terminal 2B of the evaluation user receives the tag evaluation request information from the social graph generation device 1, and the evaluation value related to the attribute information included in the tag evaluation request information is set by the evaluation user (step S106). . And the user terminal 2B of an evaluation user transmits the evaluation value regarding the set attribute information to the social graph production | generation apparatus 1 as tag evaluation information (step S107).

  Then, the tag evaluation value determination part 11 of the social graph production | generation apparatus 1 receives tag evaluation information from the user terminal 2B of each evaluation user. Then, the tag evaluation value determination unit 11 includes a plurality of tag evaluation information received from the user terminal 2B of the evaluation user, and a trust value stored in the trust DB 32 (see FIG. 3) of the evaluation user who transmitted the tag evaluation information. Based on the above, an evaluation value determination process is executed (step S108), and the evaluation value of the attribute information of the user to be evaluated is determined. Details of the evaluation value determination process will be described later with reference to FIG.

  Next, the tag evaluation value determination unit 11 transmits the evaluation value of the evaluated user determined in step S108 to the user terminal 2A of the evaluated user as evaluation value determination information (step S109).

  And the trust value update part 12 of the social graph production | generation apparatus 1 compares the evaluation value which the tag evaluation value determination part 11 determined with the evaluation value which each of the some evaluation user contained in tag evaluation information set. For each evaluation user, a trust value update process for updating the trust value stored in the trust DB 32 is executed (step S110). Details of this trust value update processing will be described later with reference to FIG.

  Subsequently, the social graph generation unit 13 of the social graph generation device 1 executes collaborative filtering processing using the attribute information stored in the tag DB 31 and the evaluation value determined for the attribute information (step S111). Extract users who are strongly related to the evaluation user. Then, the social graph generation unit 13 generates the social graph 100 by associating the evaluated user with the user extracted by the emphasis filtering (step S112), and the generated social graph 100 is stored in the social graph DB 33 (see FIG. 1). Remember.

  On the other hand, in step S109, the user terminal 2A of the user to be evaluated who has received the evaluation value determination information from the social graph generator 1 acquires the social graph call information in order to acquire the social graph 100 generated by the user to be evaluated. And transmitted to the social graph generating device 1 (step S113).

  And the social graph management part 14 of the social graph production | generation apparatus 1 which received social graph call information acquires the social graph 100 of the to-be-evaluated user memorize | stored in social graph DB33, and transmits to the user terminal 2A of to-be-evaluated user. (Step S114).

(Evaluation value determination process)
Next, an evaluation value determination process performed by the tag evaluation value determination unit 11 of the social graph generation device 1 in step S108 of FIG. 4 will be described. FIG. 5 is a flowchart showing a flow of evaluation value determination processing performed by the tag evaluation value determination unit 11 according to the present embodiment.

  First, the tag evaluation value determination unit 11 receives tag evaluation information from the user terminals 2B of a plurality of evaluation users (step S201).

  Next, the tag evaluation value determination unit 11 acquires the trust value of each evaluation user who has set the evaluation value indicated in the received tag evaluation information from the trust DB 32 (see FIG. 3) (step S202). Then, the tag evaluation value determination unit 11 determines that the evaluation user of the trust value is a bona fide user when the acquired trust value is equal to or greater than a predetermined value, and on the other hand, when the acquired trust value is smaller than the predetermined value, The user who evaluates the trust value is determined as a malicious user (step S203).

  Subsequently, the tag evaluation value determination unit 11 determines whether all the evaluation users determined in step S203 are good intention users (step S204). Here, if all the evaluation users are good-will users (step S204 → Yes), the process proceeds to step S207.

  On the other hand, if all the evaluation users are not good intention users (step S204 → No), the tag evaluation value determination unit 11 proceeds to step S205, and determines whether the number of good intention users is larger than the number of malicious users. Here, when the number of good-will users is larger than the number of malicious users (step S205 → Yes), the process proceeds to step S207. On the other hand, when the number of good intention users is equal to or less than the number of malicious users (step S205 → No), the process proceeds to the next step S206.

In step S <b> 206, the tag evaluation value determination unit 11 deletes the tag evaluation information transmitted by the evaluation user determined to be a malicious user until the number of malicious users exceeds the number of malicious users. Then, the process proceeds to next Step S207.
Here, instead of deleting the malicious user's tag evaluation information in step S206, the tag evaluation value determination unit 11 returns to step S201 in order to make the number of goodwill users larger than the number of malicious users. You may make it wait for the new tag evaluation information transmitted from the user terminal 2B of an evaluation user. Further, the process may return to step S104 in FIG. 4 to further select an evaluation user and wait for the tag evaluation information to be transmitted from the user terminal 2B of the selected evaluation user.

  Next, the tag evaluation value determination unit 11 calculates the average value using the evaluation value of the attribute information included in each tag evaluation information transmitted from the user terminal 2B of the evaluation user, and is set by the user to be evaluated An evaluation value of attribute information is determined (step S207).

  Subsequently, the tag evaluation value determination unit 11 stores the evaluation value determined in step S207 in the tag DB 31 (see FIG. 2) (step S208).

(Trust value update process)
Next, the trust value update process performed by the trust value update unit 12 of the social graph generation device 1 in step S110 of FIG. 4 will be described. FIG. 6 is a flowchart showing a flow of trust value update processing performed by the trust value update unit 12 according to the present embodiment.

  First, in step S107 of FIG. 4, the trust value update unit 12 selects one evaluation user from among a plurality of evaluation users that transmitted the tag evaluation information (step S301).

Next, the trust value update unit 12 receives the evaluation value set in the tag evaluation information by the evaluation user selected in step S301, and the evaluation value determined by the tag evaluation value determination unit 11 in step S207 of FIG. Is compared with the evaluation user to perform the trust determination (step S302).
Specifically, the trust value update unit 12 determines an evaluation user who sets an evaluation value within a predetermined range in the tag evaluation information as a good faith user based on the evaluation value determined by the tag evaluation value determination unit 11. An evaluation user who has set an evaluation value outside the predetermined range in the tag evaluation information is determined as a malicious user.

  If it is determined in step S302 that the evaluation user is a bona fide user (step S302 → bona fide), the process proceeds to the next step S303, and the evaluation user's trust value stored in the trust DB 32 (see FIG. 3). And the trust DB 32 is updated. On the other hand, if it is determined that the evaluation user is a malicious user (step S302 → malicious), the process proceeds to the next step S304, the trust value of the evaluation user stored in the trust DB 32 is decremented, and the trust DB 32 Update.

  Next, the trust value update unit 12 determines whether or not all evaluation users have been processed (step S305). If there is an evaluation user that has not yet been processed (step S305 → No), the process returns to step S301 to continue the processing. On the other hand, if all the evaluation users have been processed (step S305 → Yes), the trust value update unit 12 ends the trust value update process.

  As described above, according to the social graph generation method, the social graph generation device, and the program according to the present embodiment, the reliability of each evaluation user is secured by using the trust DB (trust information) 32. Using the attribute information set by the user to be evaluated and the evaluation value determined by the social graph generation device 1, a user having a strong relationship with the evaluation user can be extracted and a social graph can be generated. Moreover, it is not necessary to perform manual operations such as friend search by the user himself / herself and approval / non-approval of the friend relationship on the social graph 100, which has been performed in the conventional method of building the relationship between users in SNS. be able to.

  In the present embodiment, the tag evaluation value determination unit 11 is a user who is both an evaluation user determined as a good intention user and an evaluation user determined as a malicious user when the number of good intention users is greater than the number of malicious users. By obtaining the average value of the evaluation values of the evaluated users included in the tag evaluation information received from the terminal 2, the evaluation value of the attribute information was determined. However, as a modification according to the present embodiment, the tag evaluation value determination unit 11 calculates the average value using only the evaluation value of the tag evaluation information received from the evaluation user determined by the bona fide user, The evaluation value of the attribute information of the user to be evaluated may be determined. However, even in such a case, the trust value update unit 12 is an evaluation that is determined by an average value of only the evaluation values of the evaluation users determined as good intention users so that the evaluation user once determined as a malicious user is not fixed as a malicious user. By comparing the value and the evaluation value set in the tag evaluation information by each evaluation user determined by the malicious user, the trust value of the evaluation user determined by the malicious user can be updated.

DESCRIPTION OF SYMBOLS 1 Social graph production | generation apparatus 2 User terminal 3 SNS server 10 Control part 11 Tag evaluation value determination part 12 Trust value update part 13 Social graph production | generation part 14 Social graph management part 20 Input / output part 30 Storage part 31 Tag DB (tag information)
32 Trust DB (Trust Information)
33 Social Graph DB
100 social graph

Claims (7)

User terminals of a plurality of users participating in each of a plurality of social networking services (SNS), and the users connected to the user terminals of the plurality of users via a communication network and participating in any of the plurality of SNSs A social graph generation method used in a social graph generation system configured to include a social graph generation device that generates a social graph indicating a relationship between each other,
The social graph generation device includes:
(1) A tag that stores, in association with each user, an evaluation value that is set higher as the relevance between the user and the attribute information is stronger for each user attribute information set by the user A storage unit for storing information and (2) trust information for storing whether the evaluation user is a good intention user or a malicious user for each evaluation user who sets the evaluation value of the user attribute information. Has
The user terminal of the evaluated user is
Receiving the attribute information of the user to be evaluated, and executing the step of transmitting evaluation request information including the received attribute information to the social graph creating device;
The social graph generation device includes:
Receiving the evaluation request information of the evaluated user, and registering the attribute information included in the evaluation request information as attribute information of the evaluated user in the tag information;
Among the users other than the user to be evaluated, a plurality of the evaluation users who request setting of the evaluation value related to the attribute information of the user to be evaluated is selected, and the user terminal of the selected evaluation user is selected to receive the evaluation target Sending tag evaluation request information including user attribute information, and
The user terminal of the selected evaluation user is
Receiving the tag evaluation request information from the social graph creation device;
Receiving the evaluation value related to the attribute information of the user to be evaluated indicated in the received tag evaluation request information, and returning the received evaluation value to the social graph generation device as tag evaluation information,
The social graph generation device includes:
Receiving the tag evaluation information from each of the user terminals of the plurality of evaluation users;
For each of the evaluation users who set the evaluation value indicated in the tag evaluation information, the trust information is referred to determine whether the user is the good intention user or the malicious user, and the number of the good intention users is larger than the number of the malicious users. And determining the evaluation value of the attribute information by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information,
Storing the determined evaluation value of the attribute information as the evaluation value of the user to be evaluated in the tag information;
The evaluation value of the determined user to be evaluated and the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information are compared, and the evaluation value within a predetermined range based on the determined evaluation value Updating the trust information with the evaluation user who sets the evaluation user as the bona fide user and the evaluation user who sets the evaluation value out of the predetermined range as the malicious user;
Using the attribute information of each evaluated user shown in the tag information and the evaluation value of each of the determined attribute information, collaborative filtering is performed, and a user highly related to the evaluated user is extracted, Associating the user to be evaluated with the extracted user to generate the social graph;
The social graph generation method characterized by performing. The storage unit stores, as the trust information, the reliability of the evaluation user as a trust value indicating a higher value for a reliable user,
The social graph generation device includes:
In the step of determining an evaluation value of the attribute information,
The trust value of each of the evaluation users who set the evaluation value indicated in the tag evaluation information is acquired with reference to the trust information, and when the acquired trust value is a predetermined value or more, the evaluation user of the trust value If the acquired trust value is smaller than a predetermined value, the evaluation user of the trust value is determined as the malicious user, and the number of the good faith users is larger than the number of the malicious users. And determining the evaluation value of the attribute information by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information,
In the step of updating the trust information,
The evaluation value of the determined user to be evaluated and the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information are compared, and a predetermined range is set based on the evaluation value of the determined user to be evaluated. Updating the trust information by increasing the trust value of the evaluation user who has set the evaluation value within, and decreasing the trust value of the evaluation user who has set the evaluation value outside the predetermined range;
The social graph generation method according to claim 1, wherein: The social graph generation device includes:
For each of the evaluation users who set the evaluation value indicated in the tag evaluation information, when determining whether the user is a malicious user or the malicious user, the number of malicious users is equal to or greater than the number of the malicious users, Deleting the tag evaluation information received from the malicious user among the plurality of received tag evaluation information, until the number of malicious users is smaller than the number of malicious users,
The social graph generation method according to claim 1, further comprising executing the method. A social graph connected to user terminals of a plurality of users participating in each of a plurality of social networking services (SNS) via a communication network, and showing a relationship between the users participating in any of the plurality of SNSs A social graph generation device for generating,
The social graph generation device includes:
(1) A tag that stores, in association with each user, an evaluation value that is set higher as the relevance between the user and the attribute information is stronger for each user attribute information set by the user A storage unit for storing information, and (2) trust information for storing whether the evaluation user is a bona fide user or a bad user for each evaluation user who sets the evaluation value of the user attribute information ,
Evaluation request information including attribute information input by the user to be evaluated is received from the user terminal of the user to be evaluated, and the attribute information included in the evaluation request information is attribute information of the user to be evaluated in the tag information. Register as
Among the users other than the user to be evaluated, a plurality of the evaluation users who request setting of the evaluation value related to the attribute information of the user to be evaluated is selected, and the user terminal of the selected evaluation user is selected to receive the evaluation target Send tag evaluation request information including user attribute information,
From each of the selected user terminals of the plurality of evaluation users, an evaluation value related to the attribute information of the user to be evaluated set by the evaluation user is received as the tag evaluation information,
For each of the evaluation users who set the evaluation value indicated in the tag evaluation information, the trust information is referred to determine whether the user is the good intention user or the malicious user, and the number of the good intention users is larger than the number of the malicious users. In this case, the evaluation value of the attribute information is determined by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of tag evaluation information, and the evaluation value of the attribute information is determined. A tag evaluation value determination unit that stores the evaluation value of the user to be evaluated in the tag information;
The evaluation value of the user to be evaluated determined by the tag evaluation value determination unit is compared with the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information, with the determined evaluation value as a reference, A trust value updating unit configured to update the trust information with the evaluation user who has set an evaluation value within a predetermined range as the bona fide user and with the evaluation user who has set an evaluation value outside the predetermined range as the malicious user; ,
Using the attribute information of each evaluated user shown in the tag information and the evaluation value of each of the determined attribute information, collaborative filtering is performed, and a user highly related to the evaluated user is extracted, A social graph generation unit that associates the user to be evaluated with the extracted user and generates the social graph;
A social graph generation device comprising: The storage unit stores, as the trust information, the reliability of the evaluation user as a trust value indicating a higher value for a reliable user,
The tag evaluation value determination unit
The trust value of each of the evaluation users who set the evaluation value indicated in the tag evaluation information is acquired with reference to the trust information, and when the acquired trust value is a predetermined value or more, the evaluation user of the trust value If the acquired trust value is smaller than a predetermined value, the evaluation user of the trust value is determined as the malicious user, and the number of the good faith users is larger than the number of the malicious users. In this case, an evaluation value of the attribute information is determined by obtaining an average value of the evaluation values of the evaluated users included in the received plurality of the tag evaluation information,
The trust value update unit includes:
The evaluation value of the evaluated user determined by the tag evaluation value determining unit is compared with the evaluation value set by each of the plurality of evaluation users included in the tag evaluation information. The trust value of the evaluation user who has set the evaluation value within a predetermined range is increased, and the trust value of the evaluation user who has set the evaluation value outside the predetermined range is decreased, and the trust Updating information,
The social graph generation device according to claim 4. The tag evaluation value determination unit
For each of the evaluation users who set the evaluation value indicated in the tag evaluation information, when determining whether the user is a malicious user or the malicious user, the number of malicious users is equal to or greater than the number of the malicious users, Deleting the tag evaluation information received from the malicious user among the plurality of received tag evaluation information until the number of malicious users is smaller than the number of good intention users,
The social graph generation device according to claim 4, wherein:   The program for performing the social graph production | generation method of any one of Claim 1 thru | or 3 on a computer.

Images