JP2014106943A - Group extraction device, group extraction method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To dynamically extract a friend group in consideration of time series or to extract a group keyword.SOLUTION: A steady feature of a friend set is calculated from a behavior history. Based on steady feature quantity and the behavior history within a specific period, a feature which does not appear regularly but appears only within the specific period is calculated. A friend relationship in the friend set is weighted. A group of friends who have a similar interest or concern within the specific period is extracted from the friend relationship which is weighted.

Description

  In the present invention, on social media, an account group having similar interests or interests within a specific period is extracted from an account set of others connected to the user, and a keyword representing interests / interests for each group is extracted. The present invention relates to a group extraction device, a group extraction method, and a program.

  There is known a technology for providing a community formation system that can form a community that includes users who have not had contacts until now as members. The outline of the processing procedure of this prior art is as follows: (1) A user set that is frequently communicated on a certain topic is extracted as a cluster, and a cluster pair candidate that forms a basis for forming a new cluster is selected from a plurality of clusters. Next, (2) common topic attributes that characterize the community are detected from user information belonging to the cluster pair (mainly profile attributes such as affiliation and hobbies). (3) A target cluster pair is selectively merged with a topic attribute to form and present a community (corresponding to a group in the present invention) (see, for example, Patent Document 1).

  In addition, a technique is known in which accounts connected on social media are divided into a plurality of groups, and keywords representing the characteristics of each group are extracted from the content posted in the group. The outline of this prior art processing procedure is as follows: (1) A set of accounts with close friendships is extracted as a clique (corresponding to a group in the present invention) based on the density of friendships. Next, (2) increase the importance of words from accounts that have a strong influence within each creek (accounts with many friends in the group, or accounts that seem to be news sources) A word having a high importance is extracted and presented as a keyword (see, for example, Non-Patent Document 1).

JP 2010-286868 A Maike Erdmann, Tomoya Takeyoshi, Gen Hattori and Chihiro Ono, "Extraction and Annotation of Personal Cliques from Social Networks," The 12th IEEE / IPSJ International Symposium on Application and the Internet (SAINT2012), July, 2012.

  However, in the techniques described in Patent Literature 1 and Non-Patent Literature 1 described above, the regular topics from the past to the past and topics within a specific period (for example, the following is described taking the most recent week as an example) Since interest such as differences and changes in interest (change in time series) are not considered, a group based on interest and interest within a specific period is not formed preferentially. In addition, since the keywords that characterize the group are based on profile attributes, there is a problem that interests within a specific period are not correctly expressed as keywords.

  Therefore, the present invention has been made in view of the above-described problems, and includes a group extraction apparatus, a group extraction method, and a program capable of performing dynamic friend group extraction or group keyword extraction considering time series. The purpose is to provide.

  The present invention proposes the following matters in order to solve the above problems. In addition, in order to make an understanding easy, although the code | symbol corresponding to embodiment of this invention is attached | subjected and demonstrated, it is not limited to this.

  (1) The present invention is a group extraction apparatus that extracts a group of similar interests or interests within a specific period from a friend set on social media and a friend's action history, and the steady state of the friend set from the action history. On the basis of the global feature calculation unit (for example, equivalent to the global feature calculation unit 202 in FIG. 1), the steady feature amount, and the action history within a specific period. First, a weight calculation unit (for example, corresponding to the weight calculation unit 203 in FIG. 1) that calculates features that appear only within a specific period and weights the relationship between friends in the friend set, and the weighted friend A group extracting unit (for example, equivalent to the group extracting unit 203 in FIG. 1) that extracts a group of friends having similar interests or similar interests within a specific period from the relationship. It has proposed the extraction device.

  According to the present invention, the global feature calculation unit calculates the steady feature of the friend set from the action history, and the weight calculation unit calculates the steady feature based on the steady feature value and the action history within a specific period. Characteristic that does not appear in a specific manner and appears only within a specific period, weights the relationship between friends in the friend set, and the group extraction unit determines whether the interest within the specific period or the Extract a group of friends with similar interests. Therefore, by providing the group extraction result, it is possible to give an overview of the interests / interests within a specific period and the friends who have the interests / interests for friends connected on social media.

  (2) The present invention relates to a group keyword extraction unit (for example, the group keyword in FIG. 1) that extracts a keyword that characterizes a group from the content posted by a friend within a specific period for each group. A group extraction apparatus characterized by having an extraction unit 205) is proposed.

  According to the present invention, the group keyword extraction unit extracts, for each group, a keyword that characterizes the group from the posted contents of friends within a specific period. In other words, for each group, keywords that characterize the group are extracted from the content posted by friends within a specific period, so that accurate keywords based on interests and interests within a specific period can be extracted.

  (3) In the group extraction device of (2), the present invention presents the group keyword to the user, accepts designation of a group that matches the user's interests or interests from the user, and identifies the friends belonging to the designated group A group extraction apparatus is proposed that includes a group filtering unit (for example, equivalent to the group filtering unit 206 in FIG. 9) that acquires and outputs only the posted contents within a specific period.

  According to this invention, the group filtering unit presents a group keyword to the user, receives designation of a group that matches the user's interest or interest from the user, and posts within a specific period of friends belonging to the designated group Get only the contents and output. Therefore, by providing a timeline filter function based on the group extraction result, the user can efficiently follow the posting contents based on a certain interest and interest.

  (4) The present invention relates to the group extraction device of (2), wherein among the group keywords, a word representing an interest or interest registered as profile (attribute) information by a user, or a word extracted from a user's posted content A group extraction apparatus is proposed that includes a keyword matching unit (for example, equivalent to the keyword matching unit 207 in FIG. 12) for obtaining a matching group keyword for each friend.

  According to the present invention, the keyword matching unit adds a group keyword that matches a word representing an interest or interest registered as profile (attribute) information by a user or a word extracted from the posted content of the user among the group keywords. Ask every time. Therefore, by performing keyword matching, enumerating matched friends and keywords, and presenting these lists, it is possible to support the user's destination setting.

  (5) The present invention relates to the group extraction device of (1) to (4), wherein the action history is a post content or a communication history of a friend on social media, and the global feature calculation unit The word group that appears regularly and its appearance frequency, and the combination and frequency of friends that regularly communicate, are extracted, and the weight calculation unit has a specific period compared to the word group that appears regularly The weight of the friend who includes the words whose appearance frequency has improved in the posting content is set higher, and the weight of the friend who has improved the communication frequency within a specific period is higher than the regular communication frequency. A group extraction device characterized by setting is proposed.

  According to the present invention, the action history is a post content or communication history of a friend on social media, and the global feature calculation unit is configured to constantly and regularly express a group of words that appear in the post content and their appearance frequency. The combination of friends who communicate and their frequency are extracted, and the weight calculation unit compares the words whose appearance frequency has improved within a specific period compared to the words that appear regularly. The weight is set high, and the weight between friends whose communication frequency is improved within a specific period is set higher than the regular communication frequency. By performing such processing, it becomes easy to extract a group that transmits a similar topic and that is likely to activate communication.

  (6) In the group extraction device according to any one of (1) to (4), the weight calculation unit further sets a high weight between friends having high similarity of profile attributes. A device is proposed.

  According to this invention, the weight calculation unit further sets a high weight between friends whose profile attribute similarity is high. Thereby, it becomes easy to extract a group to which a friend who belongs to a similar or identical work place or a similar or identical school belongs.

  (7) The present invention is a group extraction method in a group extraction device for extracting a group of similar interests or interests within a specific period from a friend set on social media and a friend's action history, and from the action history Based on the first step (for example, equivalent to step S100 in FIG. 2) for calculating the steady feature of the friend set, the steady feature amount, and the action history within a specific period, it appears regularly. 2nd step (for example, corresponding to step S200 in FIG. 2) of calculating the features that appear within a specific period and weighting the relationship between friends in the friend set, and the weighted friend relationship And a third step (for example, corresponding to step S300 in FIG. 2) for extracting a group of friends having similar interests or similar interests within a specific period. It has proposed a group extraction method to.

  According to this invention, the stationary feature of the friend set is calculated from the behavior history, and based on the stationary feature amount and the behavior history within a specific period, it does not appear regularly, but within a specific period. The feature which appears only is calculated, the relationship between friends in a friend set is weighted, and the group of friends with similar interest or interest within a specific period is extracted from the weighted friend relationships. Therefore, by providing the group extraction result, it is possible to give an overview of the interests / interests within a specific period and the friends who have the interests / interests for friends connected on social media.

  (8) In the group extraction method of (7), the present invention provides a fourth step (for example, step S400 in FIG. 2) for extracting a keyword characterizing a group from the posted contents of friends within a specific period for each group. A group extraction method characterized by comprising

  Further, according to the present invention, for each group, a keyword that characterizes the group is extracted from the content posted by friends within a specific period. In other words, for each group, keywords that characterize the group are extracted from the content posted by friends within a specific period, so that accurate keywords based on interests and interests within a specific period can be extracted.

  (9) In the group extraction method of (8), the present invention presents the group keyword to the user in the fourth step, accepts designation of a group that matches the user's interest or interest from the user, and is designated A group extraction method is proposed in which only the contents of postings within a specific period of friends belonging to a group are acquired and output.

  According to this invention, the group keyword is presented to the user, the user's interest or the designation of the group that matches the interest is accepted from the user, and only the posting contents within a specific period of friends belonging to the designated group are acquired. ,Output. Therefore, by providing a timeline filter function based on the group extraction result, the user can efficiently follow the posting contents based on a certain interest and interest.

  (10) The present invention relates to the group extraction method of (8), in the fourth step, among the group keywords, a word representing interest or interest that the user registers as profile (attribute) information, or a user's post A group extraction method is proposed in which a group keyword matching a word extracted from content is obtained for each friend.

  According to this invention, among group keywords, a group keyword that matches an interest or interest word registered as profile (attribute) information by the user or a word extracted from the posted content of the user is obtained for each friend. Therefore, by performing keyword matching, enumerating matched friends and keywords, and presenting these lists, it is possible to support the user's destination setting.

  (11) The present invention is a program for causing a computer to execute a group extraction method in a group extraction apparatus that extracts groups having similar interests or interests within a specific period from a friend set on social media and a friend's action history. A first step (for example, corresponding to step S100 in FIG. 2) for calculating a stationary feature of a friend set from the behavior history, the stationary feature amount, and a behavior history within a specific period. Based on the second step of calculating features that do not appear regularly but appear within a specific period, and weight the relationship between friends in the friend set (for example, corresponding to step S200 in FIG. 2) And a third step of extracting a group of friends having similar interests or similar interests within a specific period from the weighted friendship (for example, the step of FIG. 2). And Tsu equivalent to up S300), has proposed a program to be executed by the computer.

  According to this invention, the stationary feature of the friend set is calculated from the behavior history, and based on the stationary feature amount and the behavior history within a specific period, it does not appear regularly, but within a specific period. The feature which appears only is calculated, the relationship between friends in a friend set is weighted, and the group of friends with similar interest or interest within a specific period is extracted from the weighted friend relationships. Therefore, by providing the group extraction result, it is possible to give an overview of the interests / interests within a specific period and the friends who have the interests / interests for friends connected on social media.

  (12) The present invention relates to the fourth step of extracting a keyword characterizing a group from the content posted by friends within a specific period for each group in the program of (11) (for example, corresponding to step S400 of FIG. 2). ) Has been proposed.

  Further, according to the present invention, for each group, a keyword that characterizes the group is extracted from the content posted by friends within a specific period. In other words, for each group, keywords that characterize the group are extracted from the content posted by friends within a specific period, so that accurate keywords based on interests and interests within a specific period can be extracted.

  (13) In the program of (12), the present invention presents the group keyword to the user in the fourth step, receives designation of a group that matches the user's interest or interest from the user, and designates the group We propose a program that acquires and outputs only the content of posts within a specific period of friends belonging to.

  According to this invention, the group keyword is presented to the user, the user's interest or the designation of the group that matches the interest is accepted from the user, and only the posting contents within a specific period of friends belonging to the designated group are acquired. ,Output. Therefore, by providing a timeline filter function based on the group extraction result, the user can efficiently follow the posting contents based on a certain interest and interest.

  (14) The present invention relates to the program of (12), from the word representing the interest or interest that the user registers as profile (attribute) information among the group keywords in the fourth step, or the content posted by the user. We have proposed a program characterized by finding a group keyword that matches the extracted word for each friend.

  According to this invention, among group keywords, a group keyword that matches an interest or interest word registered as profile (attribute) information by the user or a word extracted from the posted content of the user is obtained for each friend. Therefore, by performing keyword matching, enumerating matched friends and keywords, and presenting these lists, it is possible to support the user's destination setting.

  According to the present invention, by providing a group extraction result, it is possible to give an overview of interests and interests within a specific period and friends who have interests and interests in friends connected on social media. There is. Also, by providing a timeline filter function based on the group extraction result, there is an effect that the user can efficiently follow the posting contents based on a certain interest and interest.

  Further, the presentation of the group extraction result (presentation of the group configuration account and the keyword) makes it possible to provide a new awareness to a known friend and to expect the activation of communication. In addition, when a user shares or transmits a certain content (for example, Web news), sharing and transmission with only the group member as a destination can be supported, so that setting of a destination of the user can be supported. There is an effect.

It is a figure which shows the structure of the group extraction apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the process of the group extraction apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the process of weight calculation which concerns on the 1st Embodiment of this invention. It is a figure which shows the calculation process of the similarity of the content of contribution concerning the 1st Embodiment of this invention. It is a figure which shows the calculation process of the communication frequency which concerns on the 1st Embodiment of this invention. It is a figure which shows the calculation process of the attribute similarity based on the 1st Embodiment of this invention. It is a figure which shows the outline | summary of this invention. It is a figure which shows the output content of this invention. It is a figure which shows the structure of the group extraction apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the process of the group extraction apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the relationship of the input / output which concerns on the 2nd Embodiment of this invention. It is a figure which shows the structure of the group extraction apparatus which concerns on the 3rd Embodiment of this invention. It is a figure which shows the process of the group extraction apparatus which concerns on the 3rd Embodiment of this invention. It is a figure which shows the structure of the group extraction apparatus which concerns on the 4th Embodiment of this invention. It is a figure which shows the process of the group extraction apparatus which concerns on the 4th Embodiment of this invention.

<Outline of the present invention>
Prior to detailed description of the embodiment, the outline of the present invention will be described with reference to FIG.
First, a group extraction target period (for example, the last one week) and the past to the group extraction period before (see FIG. 7) are divided into social media actions (posts) included in the latter period. And communication), a topic word that appears regularly and a feature representing a friend who regularly communicates are extracted.

  Next, in order to make it possible to extract friends with similar interests within a specific period as members of the same group, the following two types (between the global characteristics and the characteristics within a specific period are compared between friends: Alternatively, three types of index values are calculated, and the weights of friendships are determined using these index values.

  As parameters for determining the weight of friendship, the similarity of posted contents: S (A, B), the communication frequency: F (A, B), or the similarity of profile attributes: P (A, B) can be considered.

  Posting content similarity: S (A, B) is the similarity between A and B regarding the posting content. If the posted content within a specific period is similar, the weight of the friendship is increased. At that time, when words that do not appear regularly but have a high appearance frequency within a specific period appear together in the posted contents of Mr. A and Mr. B, the weight of the friendship is particularly increased. This makes it easier for friends who are referring to the same topic to be included in the same group.

Communication frequency: F (A, B) is a communication frequency between Mr. A and Mr. B. When the communication frequency within a specific period is high, the weight of the friendship is increased. This makes it easier for friends who communicate under a certain topic (= show interest) to be included in the same group. For example, if Mr. B is responding to Mr. A's post “I was impressed with the game of Japan Futsal National Team”, Mr. B thinks that Mr. B is also interested in “Futsal” or “Japan National Team”.

Profile attribute similarity: P (A, B) is the similarity between A and B for profile attributes. If the similarity of profile attributes is high, the weight of friendship is increased. If you want to add attributes such as work or school to a friend group, you can use the similarity of profile attributes, and the degree of emphasis on weighting is the similarity of posted content: S (A, B), communication frequency: F ( A or B) or higher than the similarity of posted content: S (A, B), communication frequency: F (A, B) (that is, for α, β, γ described later, γ> = α, β).

  For example, “I want to divide even the same interests and interests such as soccer into a group of colleagues only and a group of classmates as much as possible”. However, in an environment where profile attributes can be acquired, filtering can be performed using only the profile attribute. Therefore, in the present invention, the position is optional.

From the above values, the “friendship weight” W (A, B) between Mr. A and Mr. B is determined by equation (1).
W (A, B) = α × S (A, B) + β × F (A, B) + γ × P (A, B) (1)
Here, α, β, and γ (all 0 or more) represent the degree of influence on the weight of each index value, and can be freely set when the present invention is provided as a service or application.

Specifically, the following three types of patterns and effects on extracted groups are conceivable.
The degree of influence of α is high: Emphasis is placed on the similarity of the posted contents, and it is easy to extract groups that transmit similar topics.
High degree of influence of β: Emphasis is placed on the occurrence of communication, and it is easy to extract groups that facilitate communication.
High degree of influence of γ: Emphasizing the similarity of profiles, it becomes easy to extract a group to which friends who are similar or at the same work place or school belong.

  Next, from the weighted friendship (social graph), the number of friendships in the group or a group with a high total weight is obtained. And the keyword which characterizes a group is extracted from the contribution content of the friend who belongs to each group. At this time, topic words that do not appear regularly but appear within a specific period are preferentially selected.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Note that the constituent elements in the present embodiment can be appropriately replaced with existing constituent elements and the like, and various variations including combinations with other existing constituent elements are possible. Therefore, the description of the present embodiment does not limit the contents of the invention described in the claims.

<First Embodiment>
An embodiment of the present invention will be described with reference to FIGS.

<Configuration of group extraction device>
As shown in FIG. 1, the group extraction apparatus according to the present embodiment includes social media, an API 100, a data collection unit 201, a global feature calculation unit 202, a weight calculation unit 203, a group extraction unit 204, and a group. The keyword extraction unit 205, the profile DB 211, the posted content DB 212, the communication history DB 213, the friend relationship DB 214, and the display unit 300 are configured.

  Social media is a service based on relationships between accounts (eg, friendships). The API 100 is an interface provided by various social media and capable of acquiring data on the social media from the outside. The display unit 300 presents output information to the user.

  The data collection unit 201 collects necessary data from various social media through the API 100. The global feature calculation unit 202. Calculate global features from collected and accumulated data. Here, the global feature refers to a feature that is not subject to group extraction in time series and is calculated based on data accumulated from the past (period and data corresponding to “past” in FIG. 7). . Group extraction covers data within a specific time period (eg, last week).

  The weight calculation unit 203 calculates a weight for each friend relationship on the social graph from the global feature and data within a specific period. The group extraction unit 204 extracts a dynamic friend group from the social graph and the weight of friendship. For each group, the group keyword extraction unit 205 extracts a keyword that characterizes the group from the content posted by friends in the group.

  The profile DB 211 is a database that stores profile (attribute) information registered by a user's friend in social media. The posted content DB 212 is a database that stores content posted by user's friends on social media. The communication history DB 213 is a database that stores a history of how a user's friend reacted on social media. Here, the reaction indicates, for example, a comment input action in the case of Facebook and a pressing action of a “Like” button. The friend relationship DB 214 is a database that stores a social graph (a network representing who and who are friends).

<Process of group extraction device>
Processing of the group extraction device according to the present embodiment will be described with reference to FIGS.

FIG. 1 shows a large flow of processing of the group extraction apparatus.
According to this figure, first, the global feature calculation unit 202 calculates a global feature amount from various data accumulated in the DB group (step S100 in FIG. 2). The data to be calculated is data obtained by excluding data within a specific period to be group extraction from all data accumulated from the past (period and data corresponding to “past” in FIG. 7). The weight of the friendship is determined in step S200 of FIG. 2 by comparing the feature amount calculated here with data within a specific period.

  Here, in step S100, the calculated global features are A) a global feature for each word, B) a global feature for each friend, and A) for the global feature for each word, the number of posts df (i ) That is, the total number of posts in which a certain word appears in the past data is calculated for each word.

  On the other hand, in B) the global feature regarding each friend, the number of posts P (A) of a friend A of the user, that is, the total number of posts of a friend on the past data is calculated for each friend. In addition, the total number of posts G (A) of common friends with the user and a friend A, the total number C (A) of communication of A with respect to the posts of the common friends with the user and a friend A, and others (for example, The number of times C (A, B) of Mr. A's communication with respect to the post by the common friend B) is obtained. Here, the common friend refers to a common friend between the user and Mr. A, for example, when paying attention to the friend A of the user. At this time, the total number of communications performed by Mr. A with respect to the posting of the common friend is calculated and obtained. (Between Mr. A and Mr. B, Mr. A and Mr. C, and so on)

  In step S200, the weight calculation unit 203 determines the weight of the friendship from the comparison between the global feature calculated in step S100 and the data within a specific period.

Here, the processing in step S200 will be described in detail with reference to FIG.
First, in step S210, friendship is extracted. Specifically, for each friend of the user, an explicit friend relationship on social media is acquired from the friend relationship DB, and a pair of friends who are friends is generated. For example, if there are user friends A, B, C, and D, the following pairs that are friends are listed. (Hereafter, the following list is called List 1)
1) A and B 2) A and C 3) B and C

In step S220, the similarity of the posted content is calculated. Specific contents will be described with reference to FIG.
First, in step S221, a target friend list is created. That is, a list of friends to be processed in step S220 is generated from the friend relationship (a pair that is a friend) acquired in step S210. For example, A, B, and C are extracted from list 1 and a list (hereinafter, this list is referred to as list 2) is generated.

  In step S222, if there is a friend who has not performed the processing from step S223 to step S226, the process proceeds to step S223. If not, the process proceeds to step S227. In step S223, the post content of the unprocessed friend is extracted. That is, the posting content (text) within a specific period of the target friend is extracted from the posting content DB.

In step S224, morphological analysis is performed. That is, morphological analysis is applied to the posted content, and words appearing in the posted content are acquired. In step S225, the importance of the word is calculated. That is, for each word, the importance w (i, A) of a certain word i in the posting of the friend A is calculated by comparing with the global feature. The calculation formula is as follows.
w (i, A) = df (i, A) × idf (i) (2)
Here, df (i, A) represents the number of posts including the word i among the posted contents of the friend A within a specific period, and idf (i) is expressed by the following formula using the above-described df (i). Inverse Document Frequency calculated in (1).
idf (i) = log {| D | / df (i)} (3)
Here, | D | is the total number of posts of the user's friend in the past data. When there is a possibility that df (i) = 0, it is read as df (i) = 1, or df (i) is replaced with df (i) +1.

  In step S226, a feature vector is generated. Specifically, a feature vector is generated for each friend from the word i appearing in the posted content of the friend A in a specific period and the importance w (i, A) of the word. In step S227, if there is a friendship (pair) for which the processing from step S228 to step S229 is not performed, the process proceeds to step S228. If it does not exist, this processing flow ends.

  In step S228, feature vectors of two unprocessed friends are acquired. That is, for an unprocessed friend relationship (pair), a feature vector (generated from step S223 to step S226) of each target friend is acquired.

  In step S229, the similarity between feature vectors is calculated. Specifically, the similarity between the feature vectors is calculated and set as the similarity S (A, B) of the posted content. For example, the cosine similarity corresponding to the distance between vectors is calculated. This degree of similarity corresponds to the degree of similarity regarding the posted content between two friends.

In step S230, the communication frequency is calculated. The details will be described with reference to FIG.
First, in step S231, if there is a friendship (pair) for which the processing from step S232 to step S233 is not performed, the process proceeds to step S232. If it does not exist, this processing flow ends.

In step S232, the communication history / global feature of two unprocessed friends is acquired. Specifically, suppose that two unprocessed friends are Mr. A and Mr. B. From the past data, the global features P (A), P (B), C (A), C (B), C (A, B), C (B, A), G (A), G ( B) is acquired. Next, a numerical value R (A, B) representing the rarity of communication from Mr. A to Mr. B in the past data is calculated by the following equation.
R (A, B) = (C (A) / G (A)) × (C (A, B) / P (B)) (4)

In step S233, the communication frequency between two friends is calculated. Here, the communication frequency F (A, B) between two friends is calculated by the following equation.
F (A, B) =
(iR (A, B) × R ′ (A, B) + iR (B, A) × R ′ (B, A)) / 2 (5)
Here, iR (A, B) represents the reciprocal of Equation (4), and R ′ (A, B) is a value obtained by calculating R (A, B) using only a friend's post within a specific period. Represent. In addition, when the expressions (4) and (5) are applied, if there is a possibility that the denominator becomes 0, the denominator is replaced with 1, or 1 is added by default.

  In step S241, the attribute similarity is calculated. Specific contents will be described with reference to FIG.

  In step S241, if there is a friendship (pair) for which the processing from step S242 to step S244 is not performed, the process proceeds to step S242. On the other hand, if it does not exist, this processing flow ends.

  In step S242, attribute information of two unprocessed friends is acquired. That is, if two unprocessed friends are assumed to be Mr. A and Mr. B, their profile and attribute information are acquired from the profile DB.

  In step S243, an attribute vector for each friend is generated. Specifically, the profile attributes of Mr. A and Mr. B are vectorized according to the obtainable profile and attribute items. For example, if gender, birthplace, and office can be acquired, the attribute vector is “male, XX prefecture, xx company”.

  In step S244, the similarity between attribute vectors is calculated. For example, the similarity is calculated between the attribute vectors of Mr. A and Mr. B, and the profile attribute similarity P (A, B) is obtained. At this time, the Simpson coefficient may be calculated to obtain the profile attribute similarity P (A, B).

  In step S250, the friendship weight is output. That is, S (A, B), F (A, B), and P (A, B) calculated in steps S220 to S240 are substituted into equation (1), and the friendship between A and B The weight W (A, B) of

  In step S300, a group is extracted. The group extraction unit 204 weights the friendship on the social graph with the value output in step S200, and dynamically extracts a group from the weighted graph. Here, extracting a group having a high total number of edges and weights in the group corresponds to extracting a group having similar interests and interests within a specific period. For example, the method described in the document “A. Clauset, M. E. J. Newman and C. Moore,“ Finding community structure in large large networks, ”Phys. Rev. E70, 066111 (2004).” Is applied. By doing so, the above groups can be extracted.

  In step S400, group keywords are extracted. The group keyword extraction unit 205 extracts, for each group extracted in step S300, a keyword that characterizes the group from the contents posted within a specific period of friends in the group. For example, when Mr. A belongs to a group to be calculated, the word importance level w (i, A) is totaled for each word included in the posted content of Mr. A in a specific period, and the value is high. Extract words as group keywords.

  That is, as shown in FIG. 8, accounts having similar affiliations and interests are grouped from the structure of the social graph, and keywords characterizing the group are extracted from the posted contents of friends belonging to the group. Therefore, by automatically generating a friend list, the posted content can be confirmed for each account with similar interests.

  As described above, according to the present embodiment, by providing a group extraction result, an overview of the interests / interests within a specific period and the friends who have the interests / interests for friends connected on social media. can do. Moreover, since the keyword which characterizes a group is extracted from the contribution content of the friend in a specific period for every group, the exact keyword based on the interest and interest in a specific period can be extracted.

<Second Embodiment>
The embodiment of the present invention will be described with reference to FIGS.

<Configuration of group extraction device>
As shown in FIG. 9, the group extraction apparatus according to the present embodiment includes social media, an API 100, a data collection unit 201, a global feature calculation unit 202, a weight calculation unit 203, a group extraction unit 204, and a group. The keyword extraction unit 205, the group filter unit 206, the profile DB 211, the posted content DB 212, the communication history DB 213, the friend relationship DB 214, and the display unit 300 are configured. In addition, about the component which attaches | subjects the same code | symbol as 1st Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The group filter unit 206 acquires and presents only the posting contents of friends belonging to the group designated by the user. In addition, this embodiment provides the function which filters the social stream (flow of a friend's update information) which a user can browse based on the group acquired in 1st Embodiment. A specific example from the group designation to the filtering application result is shown in FIG. 11 as a screen image.

<Process of group extraction device>
The process of the group extraction device according to this embodiment will be described with reference to FIG.

  First, the processing content from step S100 to step S400 is the same as in the first embodiment.

  In step S500, group designation is accepted. In other words, in response to a user's filtering request, a group keyword list is presented for each group extracted in step S400, and the user can select a keyword to select a group. Here, when there are a plurality of group keywords, a predetermined number (for example, one) is presented, and the importance used at the time of group keyword extraction is used as the selection criterion.

  In step S600, a group filter is applied. Specifically, only the posted contents of friends belonging to the group designated by the user are acquired and presented.

  As described above, according to the present embodiment, the function of filtering the social stream (flow of friend update information) that can be browsed by the user is provided based on the group acquired in the first embodiment. By selecting a group keyword extracted and presented, the user can select a group that matches his / her interest and view only the posted content of friends belonging to the specified group. .

<Third Embodiment>
The embodiment of the present invention will be described with reference to FIGS.

<Configuration of group extraction device>
As shown in FIG. 12, the group extraction apparatus according to this embodiment includes social media, an API 100, a data collection unit 201, a global feature calculation unit 202, a weight calculation unit 203, a group extraction unit 204, and a group. The keyword extraction unit 205, the keyword matching unit 207, the profile DB 211, the posted content DB 212, the communication history DB 213, the friend relationship DB 214, and the display unit 300 are configured. In addition, about the component which attaches | subjects the same code | symbol as 1st Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The keyword matching unit 207 acquires a keyword representing the user's interest / interest from the user's profile information or the posted content of the user. In this embodiment, from the group and the group keyword acquired in the first embodiment, the interest is labeled with respect to the friend within a specific period, and the friend who is likely to match the user's interest Provides a function to present the reason (group keyword).

<Process of group extraction device>
The process of the group extraction apparatus according to the present embodiment will be described with reference to FIG.

  First, the processing content from step S100 to step S400 is the same as in the first embodiment.

  In step S700, a user keyword is acquired. Specifically, a keyword representing the user's interest / interest is acquired from the user's profile information or the user's posted content. If the user has already registered specific keywords (for example, proper nouns such as artist names) for categories such as “hobby”, “interest”, and “interest”, the corresponding keywords are listed. In addition, when the user's posted content can be acquired, the posted content is converted into the posted content by processing similar to the processing (step S224 to step S225 in FIG. 4) when the importance of the word is calculated with respect to the posted content of the friend. Assign importance to each word included, and use words with high importance as keywords.

  In step S800, keyword matching is performed. Specifically, a keyword representing the interest of the friend is assigned to each friend who belongs to the group. For example, when a post extracted within a specific period includes a word extracted as a group keyword, the word is assigned to the friend as a keyword. Next, keyword matching is performed between the user and each friend, and the matched friend and keyword are listed.

  In step S900, a friend and keyword pair is presented. That is, in step S800, a list of matched friends and keywords is presented.

  As described above, according to the present embodiment, the interest and interest of a user within a specific period are labeled from the group and the group keyword acquired in the first embodiment, and the user's interest and interest are expressed. By providing a function that presents friends who are likely to match and the reason (group keyword), it is possible to easily know friends that can be expected to activate communication with the users themselves while their interests and interests change. It becomes possible.

<Fourth Embodiment>
The embodiment of the present invention will be described with reference to FIGS. 14 and 15.

  In the present embodiment, the group extraction means and the group keyword extraction means shown in the first embodiment are stored in the accumulated data while shifting the period for group extraction (for example, one week ago, two weeks ago,...). To obtain the group extraction result in each period. Further, by evaluating the burst property (exciting degree) of each group, a group related to a special event (for example, a soccer World Cup or a friend's anniversary) generated and shared between friends is extracted and listed. In other words, in the first to third embodiments, groups are extracted from a specific period with a focus on improving common topics and communication frequency. In this embodiment, individuals or the general public are further extracted. In the case where there is a group related to a topic that is particularly exciting in the above, it is extracted as a special event-related group.

<Configuration of group extraction device>
As shown in FIG. 14, the group extraction apparatus according to the present embodiment includes social media, an API 100, a data collection unit 201, a global feature calculation unit 202, a weight calculation unit 203, a group extraction unit 204, and a group. The keyword extraction unit 205, the group evaluation unit 208, the profile DB 211, the posted content DB 212, the communication history DB 213, the friend relationship DB 214, and the display unit 300 are configured. In addition, about the component which attaches | subjects the same code | symbol as 1st Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The group evaluation unit 208 determines burstiness (swell) from the extracted data of each group, and particularly determines a group that is rising as an event-linked group.

<Process of group extraction device>
The process of the group extraction device according to the present embodiment will be described with reference to FIG.

  First, the processing content from step S100 to step S400 is the same as in the first embodiment.

  In step S1100, if there is a period in which the group extraction process has not been performed, the process proceeds to step S1200, and if not, the process flow ends.

  In step S1200, data is divided. Specifically, the data related to the period to be the group extraction target and the data related to the period to be the global feature calculation target are divided. The division unit may be arbitrary (for example, one week), and may be specified by the service or application provider or the user.

  In step S1300, the keyword group is evaluated. Specifically, burstiness (swelling) is determined from the extracted data of each group, and a particularly exciting group is determined as an event-linked group. Here, threshold processing is applied to the determination of the climax, for example, “if the climax index value of a certain group is X times or more compared to the average value of the climax index calculated from all groups, it is determined as an event-related group ”Is performed. Note that, as the climax index, for example, any one of “total number of posts in the group” and “total number of communication times in the group” or a combination of a plurality of the determinations is determined.

  As described above, according to the present embodiment, the group extraction means shown in the first embodiment, while shifting the period for group extraction (for example, one week ago, two weeks ago,...) And applying the group keyword extraction means to the accumulated data, obtaining group extraction results for each period, and evaluating the burstiness (excitement level) of each group, so that it is generated and shared between friends By extracting and enumerating groups related to events (for example, soccer World Cup or friends' anniversary), it is possible to overview the groups related to individual (friends) or special events that occur in general. Become.

  Note that the group extraction apparatus of the present invention is realized by recording the processing of the group extraction apparatus on a recording medium readable by the computer system, causing the group extraction apparatus to read and execute the program recorded on the recording medium. Can do. The computer system here includes an OS and hardware such as peripheral devices.

  Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW (World Wide Web) system is used. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient.

  The embodiments of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to the embodiments, and includes designs and the like that do not depart from the gist of the present invention.

100; API
201; data collection unit 202; global feature calculation unit 203; weight calculation unit 204; group extraction unit 205; group keyword extraction unit 206; group filter unit 207; keyword matching unit 208: group evaluation unit 211;
212; Post content DB
213: Communication history DB
214; Friendship DB
300: Display section

Claims (14)

A group extraction device that extracts a group of similar interests or interests in a specific period from a friend collection on social media and a friend's action history,
A global feature calculation unit for calculating a steady feature of a friend set from the action history;
A weight that calculates a feature that does not appear regularly but only appears in a specific period based on the steady feature value and an action history in a specific period, and weights the relationship between friends in the friend set A calculation unit;
A group extractor for extracting a group of friends having similar interests or interests within a specific period from the weighted friendships;
A group extracting apparatus comprising:   The group extraction device according to claim 1, further comprising a group keyword extraction unit that extracts a keyword that characterizes a group from the content posted by a friend within a specific period for each group.   Group filtering that presents the group keyword to the user, accepts designation of the group that matches the user's interest or interest from the user, and obtains and outputs only the posting contents within a specific period of friends belonging to the designated group The group extraction apparatus according to claim 2, further comprising a unit.   Among the group keywords, a keyword matching unit is provided for each friend to find a group keyword that matches a word representing an interest or interest registered as profile (attribute) information by the user or a word extracted from the posted content of the user. The group extraction apparatus according to claim 2, wherein: The action history is a posted content of a friend on social media or a communication history,
The global feature calculation unit extracts a group of words that regularly appear in the posted content and its appearance frequency, and a combination and frequency of friends that regularly communicate.
The weight calculation unit sets a high weight between friends including both words whose appearance frequency is improved within a specific period as compared to the regularly appearing word group in the posted content, and further steady communication. The group extraction device according to any one of claims 1 to 4, wherein a weight between friends whose communication frequency is improved within a specific period is set higher than a frequency.   5. The group extraction apparatus according to claim 1, wherein the weight calculation unit further sets a high weight between friends having high similarity of profile attributes. A group extraction method in a group extraction device for extracting a group of similar interests or interests within a specific period from a friend set on social media and a friend's action history,
A first step of calculating a stationary feature of a friend set from the behavior history;
Based on the steady feature value and the action history within a specific period, a feature that does not appear regularly but appears within a specific period is calculated, and the relationship between friends in the friend set is weighted A second step;
A third step of extracting from the weighted friendships a group of friends with similar interests or interests within a specific time period;
A group extraction method comprising:   The group extraction method according to claim 7, further comprising a fourth step of extracting a keyword characterizing the group from the content posted by a friend within a specific period for each group.   In the fourth step, the group keyword is presented to the user, designation of the group that matches the user's interest or interest is accepted from the user, and only the content posted within a specific period of friends belonging to the designated group is received. The group extraction method according to claim 8, wherein the group extraction method is obtained and output.   In the fourth step, among the group keywords, a group keyword that matches an interest or interest expressing word registered as profile (attribute) information by the user or a word extracted from the posted content of the user is obtained for each friend. The group extraction method according to claim 8, wherein: A program for causing a computer to execute a group extraction method in a group extraction apparatus that extracts a group of similar interests or interests within a specific period from a friend set on social media and a friend's action history,
A first step of calculating a stationary feature of a friend set from the behavior history;
Based on the steady feature value and the action history within a specific period, a feature that does not appear regularly but appears within a specific period is calculated, and the relationship between friends in the friend set is weighted A second step;
A third step of extracting from the weighted friendships a group of friends with similar interests or interests within a specific time period;
A program that causes a computer to execute.   The program according to claim 11, further comprising a fourth step of extracting a keyword characterizing the group from the content posted by a friend within a specific period for each group.   In the fourth step, the group keyword is presented to the user, designation of the group that matches the user's interest or interest is accepted from the user, and only the content posted within a specific period of friends belonging to the designated group is received. The program according to claim 12, wherein the program is acquired and output. In the fourth step, among the group keywords, a group keyword that matches an interest or interest expressing word registered as profile (attribute) information by the user or a word extracted from the posted content of the user is obtained for each friend. The program according to claim 12, wherein: