KR101073599B1 - System and method for recommending personal media - Google Patents

Abstract

A system and method for recommending personal media is disclosed. A method of recommending personal media in a server connected to a plurality of personal media via a network, the method comprising: generating a plurality of virtual relationship groups including personal media having a connection or content similarity with the recommendation requester personal media; The plurality of virtual relationship groups each comprise a plurality of different personal media; Selecting at least one virtual relationship group among the generated plurality of virtual relationship groups as an evaluation target group; And recommending the personal media to the recommendation requestor by performing evaluation on the plurality of personal media included in the selected evaluation target population. According to the present invention, the problem of scarcity caused by concentrating the connection relationship on a specific personal media by selecting a target person by creating a virtual relationship group using content attributes of the personal media as well as the connection relationship information between the personal media to select an evaluation target Can be solved.

Description

System and Method for Recommending Personal Media}

The present invention relates to a system and method for recommending personal media, and more particularly, to a system and method for accurately recommending personal media with high user interest.

Personal media has grown exponentially as the opportunity for web users has expanded in the era of Web 2.0. Personal media comes in many forms, including blogs, Facebook or Twitter, and UCC. The personal media sector has grown rapidly, creating intricate connections and vast personal media information.

Due to the vast amount of personal media information, it is difficult for a user to obtain the information he / she wants. Therefore, there is a need for efficient management of personal media. The recent research on the recommendation of personal media focuses on the efficient use and management of personal media by providing users with appropriate information for each individual. The analysis of accumulated personal media information, which is the core of personal media recommendation, makes it possible to judge the propensity and relationship of personal media users and to predict and present appropriate information to users.

The main issue in personal media recommendation is the selection of evaluation targets. In order to select an evaluation target, an evaluation target is conventionally selected based on a personal media 103 directly connected to a user's personal media 101 and an indirectly connected personal media 103 as shown in FIG. 1. However, in actual connection relations between personal media, as shown in FIG. Referring to FIG. 2, a node number means a personal media and a comment ratio means a connection relationship between personal media. In other words, in personal media, the connection between personal media is formed through tools provided in personal media such as comments, trackbacks, and bookmarks. This connection relationship is concentrated only on specific popular personal media, so the formation rate of the connection relation is rare. appear.

When the evaluation target is selected for the personal media directly or indirectly connected to the user's personal media due to such scarcity, there is a problem that it is difficult to ensure the association between the requestor's personal media and the evaluation target.

In order to solve the problems of the prior art as described above, the present invention is to propose a system and method for recommending personal media that can increase the user's satisfaction.

According to a preferred embodiment of the present invention to achieve the above object, a method for recommending personal media in a server connected to a plurality of personal media via a network, the connection relationship or content similarity with the recommendation requester personal media Creating a plurality of virtual relationship groups comprising personal media having said plurality of virtual relationship groups each comprising a plurality of different personal media; Selecting at least one virtual relationship group among the generated plurality of virtual relationship groups as an evaluation target group; And recommending the personal media to the recommendation requestor by performing evaluation on a plurality of personal media included in the selected evaluation target population.

The generating of the virtual relationship group may include: generating first media content information using a frequency of occurrence of each word of the content posted on the recommendation requester personal media; Generating second media content information by using a frequency of occurrence of each word of content posted on other personal media other than the recommendation requester personal media; And generating n virtual relationship groups (where n is one or more natural numbers) including a plurality of different personal media using content similarity between the first media content information and the second media content information. Characterized in that.

The selecting of the evaluation target group may include determining content similarity between the first media content information and second media content information of the other personal media included in the n virtual relationship groups, respectively; Obtaining a content similarity mean value for each virtual relationship group using the content similarity; And selecting a subject to be evaluated using the content similarity average value, wherein the content similarity has different connection weights according to a connection relationship between the recommendation requester personal media and the other personal media.

Subsequently, recommending the personal media to the recommendation requester performs an evaluation on each of the plurality of other personal media using the content similarity between the recommendation requester personal media and a plurality of other personal media included in the evaluation target group. Making; Recommending at least one personal media to the recommendation requester using the evaluation result.

The performing of the evaluation may include determining an intimacy weight assigned to the intimacy according to the intimacy according to the connection between the recommendation requester personal media and the other personal media and whether the personal media subscribed to a particular Internet service; And further comprising the step of determining the popularity of the other personal media, characterized in that the evaluation of each of the other personal media in consideration of the subscription weight and the popularity further characterized.

Here, the subscription weight is higher than that when the recommendation requester personal media is subscribed to a specific Internet service and the other personal media is subscribed to the specific Internet service, when the other personal media is not subscribed to the specific Internet service. It is characterized by having.

The popularity may be determined by using any one or more of the number of comments, the number of trackbacks, the number of scraps, or the number of favorites for the document posted on the other personal media.

In addition, according to another embodiment of the present invention, media content information generating means for generating media content information of a personal media; Virtual relationship group generating means for generating a plurality of virtual relationship groups including personal media having a connection relationship or content similarity with the recommendation requester personal media by using the media content information, wherein the plurality of virtual relationship groups are different from each other; Contains personal media of; Evaluation target group selection means for selecting one or more virtual relationship groups among the generated plurality of virtual relationship groups as evaluation target groups; And a personal media recommending means for performing a valuation of a plurality of personal media included in the selected evaluation target population and recommending the personal media to the recommendation requester.

The present invention solves the problem of scarcity caused by concentrating the connection relationship on a specific personal media by creating a virtual relationship group using content attributes of the personal media as well as the connection relationship information between the personal media to select an evaluation target. Can be.

1 is a diagram illustrating a connection relationship between personal media.
2 is a diagram illustrating an incidence of connection relationships in personal media.
3 is a block diagram illustrating a personal media recommendation system according to an embodiment of the present invention.
4 illustrates an example of selecting an evaluation target using a connection relationship and selecting an evaluation target using a virtual relationship.
5 is a flowchart illustrating the entire process of the recommendation method of the personal media according to an embodiment of the present invention.
6 is a flowchart illustrating a specific process of creating a virtual relationship group according to an embodiment of the present invention.
7 is a flowchart illustrating a specific process of selecting an evaluation target group according to an embodiment of the present invention.
8 is a flowchart illustrating a specific process of recommending personal media according to an embodiment of the present invention.

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. Like reference numerals are used for like elements in describing each drawing. Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram illustrating a personal media recommendation system according to an embodiment of the present invention.

Referring to FIG. 3, the personal media recommendation system 300 according to an embodiment of the present invention may include media content information generating means 301, virtual relationship group generating means 303, evaluation target group selecting means 305, and individual. Media recommendation means 307 may be included. Hereinafter, the function of each component will be described in detail.

The personal media recommendation system 300 may be provided in a server connected to a plurality of personal media via a network.

The media content information generating means 301 generates content information from the content included in the personal media.

Here, the personal media is a personal area owned by an individual on the web and is expressed in various ways such as blog, Facebook, Twitter, UCC, and the like. Personal media may refer to a site that can freely publish content according to interests on a particular subject, and the content may refer to a document including text, image, image, voice, or flash on a specific subject.

In general, the content may be a text (post) posted by the user, and the following description will be mainly focused on the case where the content is a post.

According to an embodiment of the present invention, the media content information may be generated by using vector weights of various words included in a post. The weight of a word refers to the importance of each word in the content.

The weight of these words can be determined using the TF-IDF weight using the frequency of appearance of words in the document. TF-IDF is a weight used in information retrieval and text mining. It is a statistical value that indicates how important a word is in a particular document when there are a group of documents.

TF (term frequency) is a value that indicates how often a particular word appears in a document. The higher this value, the more important the document may be considered. However, if the word itself is often used in a family of documents, this means that the word is common. This is called DF (document frequency), and the inverse of this value is called inverse document frequency (IDF). TF-IDF is the product of TF and IDF.

The TF-IDF weights in the personal media may appear as many as the number of word sets designated by the user, and media content information may be generated using the weights of the words as vector elements.

Recommendation requestor requesting the recommendation of the personal media generated by the media content information generating means 301 The content similarity between the personal media is measured by comparing the first media content information of the personal media with the second media content information of another personal media. can do.

According to an embodiment of the present invention, a cosine similarity method may be used as a method of measuring content similarity. The cosine similarity method is a method of setting a document as one vector and determining the similarity at an angle formed by two vectors.

Therefore, since the media content information generated by the media content information generating means is expressed as a vector using weights of words, content similarity between personal media can be measured using a cosine-like scheme.

The virtual relationship group generation means 303 generates a virtual relationship group including a plurality of personal media having a connection relationship or content similarity with the recommendation requester personal media for selecting an evaluation target.

Here, the virtual relation group is a set generated using personal media having content similarity as well as connection relation information with the requestor personal media. High association can be guaranteed.

4 illustrates an example of selecting an evaluation target using a connection relationship and selecting an evaluation target using a virtual relationship.

Referring to FIG. 4A, as an example of selecting an evaluation target using an initial connection relationship with the recommendation requester personal media 401, personal media directly or indirectly connected to the recommendation requester personal media 401 are selected as evaluation targets. do.

However, as shown in FIG. 2, the connection relationship that the recommendation requester personal media 401 has is rare, and most connection relationships are concentrated in a few personal media. Therefore, the subjects of the evaluation must be focused on the minority media, and if one connection is selected for evaluation by the minority media and the personal media connected to the minority media, the personal media other than the minority media are the requester personal media. There is a problem that it is difficult to guarantee the association with the (401).

Referring to FIG. 4 (b), an evaluation target is selected for a virtual relationship group using content similarity.

When the virtual relationship group is constructed using the similarity of contents, the relationship area of personal media, which was defined only as direct and indirect connection, changes. The area A 403 denotes a case in which the personal media which have already connected with the recommendation requester personal media 401 maintain the relationship by showing high content similarity. The area B 405 means that the personal media belonging to one virtual relation group is composed only of the direct requester personal media 401. Area C means that indirectly connected personal media lose their original direct connection and are organized into a new virtual relationship group, or that personal media that do not have an original direct or indirect connection are included in the virtual relationship group due to content similarity.

Therefore, when creating a virtual relationship group by using content similarity, personal media based on content similarity are added in addition to the personal media in direct connection or indirect connection, thereby solving the problem of the scarcity of the conventional connection and creating the virtual relationship group. By considering the content similarity of the city recommendation requester personal media 401, personal media highly related to the recommendation requester personal media 401 can be selected as an evaluation target.

The evaluation target group selecting means 305 selects one or more virtual relationship groups among the plurality of virtual relationship groups generated by the virtual relationship group generation means 303 as the evaluation target group.

Selection of the subject to be evaluated is performed by using the content similarity average value using the connection relationship and the content similarity between the plurality of personal media included in the virtual relation group and the recommendation requester personal media 401.

The personal media recommending means 307 recommends to the recommendation requester at least one or more personal media having high content similarity with the recommendation requester personal media 401 among the personal media included in the evaluation target group.

Hereinafter, the personal media recommendation method according to an embodiment of the present invention will be described in detail.

5 is a flowchart illustrating the entire process of the recommendation method of the personal media according to an embodiment of the present invention.

First, in step S501, a virtual relationship group including a plurality of personal media having a connection relationship or content similarity with the recommendation requester personal media is generated.

In operation S501, a plurality of virtual relationship groups may be generated.

Subsequently, in step S503, one or more virtual relationship groups among the plurality of virtual relationship groups are selected as the evaluation target group using the content similarity between the recommendation requester personal media and the virtual relationship group.

In step S505, the evaluation of each of the plurality of other personal media is performed by using content similarity between the recommendation requester personal media and a plurality of other personal media included in the evaluation target group, and the recommendation is performed using the evaluation result. Recommend at least one or more personal media to the claimant.

6 is a flowchart illustrating a specific process of creating a virtual relationship group according to an embodiment of the present invention.

In step S601, the media content information generating means 301 generates the first media content information of the recommendation requester personal media 401. The virtual relation group is composed of a collection of personal media having content information similar to the first media content information.

According to another embodiment of the present invention, the first media content information generation may be generated for the recommendation requester personal media 401 and personal media having a direct and indirect connection with the personal media 401. In this case, the content of the content included in the recommendation requester personal media 401 and the personal media associated with the recommendation requester personal media 401 becomes first media content information for generating a virtual relationship group.

Here, the reason for extending to the personal media having a direct or indirect connection in addition to the recommendation requester personal media 401 to generate basic personal media content information is included in the personal media formed in connection with the recommendation requester personal media. The content may also be of interest to the recommendation requester.

In step S603, the media content information generating means 301 generates second media content information of another personal media other than the recommendation requester personal media 401.

In step 605, the virtual relationship group generation means 303 generates a plurality of virtual relationship groups including a plurality of different personal media using the content similarity between the first media content information and the second media content information.

According to an embodiment of the present invention, a clustering technique using a K-average algorithm may be used to generate a plurality of virtual relationship groups. The K-average algorithm is an algorithm for grouping given data into k clusters. The similarity between k cluster centers randomly determined based on the first media content information and the second media content information of another personal media is calculated using the cosine similarity technique. do.

For example, 100 words having a high weight may be selected from the first media content information, and the center of the first cluster may be determined using the same.

Each personal media belongs to the cluster-centric cluster with the highest similarity. Once the clusters for all personal media have been determined, the cosine similarity average of each cluster is obtained and this average becomes the new cluster center.

The cluster operation is repeated around the new cluster, and the iteration for cluster improvement is terminated when the cluster for the new cluster center is not changed from the previous cluster.

As a result, k virtual relationship groups are created, and each virtual relationship group includes a plurality of different personal media.

7 is a flowchart illustrating a specific process of selecting an evaluation target group according to an embodiment of the present invention.

Step S701 is a step of determining content similarity between the recommendation requester personal media 401 and the other personal media.

Here, the plurality of other personal media constituting the virtual relationship group may have a direct connection, an indirect connection, and an arbitrary connection in a relationship with the recommendation requester personal media 401.

The content similarity average of other personal media having a direct connection with the recommendation requester personal media 401 according to an embodiment of the present invention may be determined as in Equation 1 below.

[ Equation 1 ]

는 직접 연결관계를 가지고 있는 타 개인 미디어에 대한 내용 유사도 평균값, n은 직접 연결관계를 가지고 있는 타 개인 미디어의 개수, CosineSim(d,r)은 n개의 직접 연결관계를 가지고 있는 타 개인 미디어 중 d 번째 타 개인 미디어와 추천 요구자 개인 미디어(401)와의 코사인 유사도,

는 직접 연결관계에 대한 가중치를 각각 의미한다.here,

Is the average value of content similarity with other personal media with direct connection, n is the number of other personal media with direct connection, and CosineSim (d, r) is d among other personal media with direct connection. Cosine similarity between the first and other personal media and the requestor personal media 401,

Denotes the weight for each direct connection.

The weight for the direct link has a higher value than the weight for the indirect or arbitrary link. The fact that the personal media has a direct connection with the recommendation requester personal media 401 is because the likelihood that the personal media includes content satisfying the interest of the recommendation requester is high. Here, when there are a plurality of other personal media that have a connection relationship, the average value of the content similarity they have is obtained.

The content similarity average of other personal media having an indirect connection with the recommendation requester personal media 401 according to an embodiment of the present invention may be determined as in Equation 2 below.

[ Equation 2 ]

는 간접 연결관계를 가지고 있는 타 개인 미디어에 대한 내용 유사도 평균값, n은 간접 연결관계를 가지고 있는 타 개인 미디어의 개수, CosineSim(i,r)은 n개의 간접 연결관계를 가지고 있는 타 개인 미디어 중 i 번째 타 개인 미디어와 추천 요구자 개인 미디어(401)와의 코사인 유사도,

는 간접 연결관계에 대한 가중치를 각각 의미한다.here,

Is the average value of content similarity for other personal media with indirect connection, n is the number of other personal media with indirect connection, and CosineSim (i, r) is i among other personal media with indirect connection. Cosine similarity between the first and other personal media and the requestor personal media 401,

Denotes the weight for each indirect connection.

The weight for the indirect connection is given a value lower than the weight for the direct connection. When there are a plurality of other personal media having indirect connection, the average value of the content similarity is obtained.

The content similarity average of other personal media having a random connection with the recommendation requester personal media 401 according to an embodiment of the present invention may be determined as in Equation 3 below.

[ Equation 3 ]

는 임의 연결관계를 가지고 있는 타 개인 미디어에 대한 내용 유사도 평균값. n은 임의 연결관계를 가지고 있는 타 개인 미디어의 개수, CosineSim(x,r)은 n개의 간접 연결관계를 가지고 있는 타 개인 미디어 중 x 번째 타 개인 미디어와 추천 요구자 개인 미디어(401)와의 코사인 유사도,

는 직접 연결관계에 대한 가중치, CBRate는 연결 개인 미디어 비율을 각각 의미한다.here,

Is the average value of content similarity for other personal media with random connection. n is the number of other personal media having a random connection, CosineSim (x, r) is the cosine similarity between the x-th personal media and the recommendation requester personal media 401 among the n indirect connections,

Is the weight for direct connection and CBRate is the ratio of connected personal media.

Here, the weight for any connection is determined according to the ratio of direct and indirect connections. This may be determined by the ratio of directly or indirectly connected personal media without setting the connection weight to a fixed value because other randomly connected personal media does not have any connection relationship with the recommendation requester personal media 401.

In other words, if there are many personal media with connected relationships included in the virtual relationship group, the proportion of individual media with random connection for the evaluation of the entire virtual relationship group is lowered. Personal media with random connections to the multi-faceted group assessments will be more involved.

The connection relationship between other personal media included in the virtual relationship group with the recommendation requester personal media 401 may be shown in a complex form with Table 1 below.

Direct connection Direct connection + indirect connection Direct connection + random connection Direct connection + indirect connection + random connection Random connection

Therefore, the average value of content similarity for the virtual relationship group may be determined as in Equation 4 below.

[ Equation 4 ]

은 가상 관계 집단에 대한 내용 유사도 평균값, N은 가상 관계 집단에 포함된 타 개인 미디어의 총 개수를 의미한다. here,

Is the average value of content similarity for the virtual relationship group, and N is the total number of other personal media included in the virtual relationship group.

Step S703 calculates an average value of content similarity for each of the plurality of virtual relationship groups.

In step S705, a virtual relationship group having a high content similarity mean value is selected as the evaluation target group using the content similarity mean value for the virtual relationship group.

8 is a flowchart illustrating a specific process of recommending personal media according to an embodiment of the present invention.

The final purpose of the personal media recommendation model is to use the information of the personal media recommendation requester to determine and present personal media highly related to the recommendation requester. It is necessary to evaluate other personal media included in the evaluation target group in order to determine whether it is personal media highly related to the recommendation requester.

Step S801 determines the intimacy and subscription weight of other personal media included in the population to be evaluated.

Intimacy is determined by three values of direct, indirect, and arbitrary relationships according to the connection relationship between the other personal media and the recommender personal media 401. In the case of a direct connection, the highest intimacy is obtained, and in the case of a random relationship, the intimacy is low.

The subscription weight is a value determined by the operation method of the recommendation requester personal media 401. The operation of personal media may be divided into a method of subscribing to a specific Internet service or operating an individual domain as shown in Table 2 below. Specific internet services are personal media services provided by Naver and Facebook.

Sign up for a service Personal media Internal personal media External personal media Individual Domain Personal Media External personal media

A connection relationship with other personal media may be newly defined according to the operation method of the recommendation requester personal media 401. If the referrer operates personal media subscribed to a particular Internet service, the referrer may be required to use personal media belonging to the same service (hereinafter referred to as internal personal media) or personal media not belonging to the same service (hereinafter referred to as external personal media). There will be a connection. On the contrary, if the recommender operates a personal media using an individual domain that is not subscribed to the Internet service, only the connection with the external personal media is established.

That is, the subscription weight is a weight for reflecting that the personal media subscribed to the service is more likely to connect with the internal personal media than the external personal media.

In step S803, the popularity of other personal media included in the evaluation target population is determined.

Popularity is determined based on a plurality of other personal media included in the evaluation target group, and may be determined using one or more of the number of comments, the number of trackbacks, the number of scraps, or the number of favorites.

For example, if the popularity is determined using the number of comments, the popularity can be determined by comparing the maximum number of comments of the group to be evaluated with the number of comments of other personal media.

Subsequently, in step S805, each of the other personal media is evaluated using the similarity between contents of the recommendation requester personal media 401 and the other personal media, the intimacy weighted by the subscription weight, and the popularity.

The process of evaluating each of the other personal media may be expressed as in Equation 5 below.

[ Equation 5 ]

는 i번째 개인 미디어에 대한 평가 지수,

는 i번째 개인 미디어에 대한 가입 가중치,

는 i번째 개인 미디어에 대한 친밀도,

는 i번째 개인 미디어에 대한 친밀도,

는 i번째 개인 미디어와 추천자의 개인 미디어 간의 내용 유사도를 각각 의미한다.
here

Is the rating index for the i-th personal media,

Is the subscription weight for the i th personal media,

Intimacy with the i personal media,

Intimacy with the i personal media,

Means content similarity between the i-th personal media and the recommender's personal media.

Step S807 recommends at least one or more personal media to the recommendation requester using the evaluation index for each of the other personal media.

As described above, the present invention has been described by specific embodiments such as specific components and the like. For those skilled in the art to which the present invention pertains, various modifications and variations are possible. Therefore, the spirit of the present invention should not be limited to the described embodiments, and all of the equivalents or equivalents of the claims as well as the claims to be described later will belong to the scope of the present invention. .

101: Referrer Personal Media 103: Directly Connected Personal Media
105: indirectly connected personal media
401: Referrer personal media 403: A area
405: B area 407: C area

Claims (16)

As a method for recommending personal media in a server connected to a plurality of personal media via a network,
Creating a plurality of virtual relationship groups comprising personal media having a connection relationship or content similarity with the recommendation requester personal media, wherein the plurality of virtual relationship groups each include a plurality of different personal media;
Selecting at least one virtual relationship group among the generated plurality of virtual relationship groups as an evaluation target group; And
Recommending the personal media to the recommendation requestor by performing evaluation on a plurality of personal media included in the selected evaluation target group;
Personal media recommendation method comprising a The method of claim 1,
Creating the virtual relationship group,
Generating first media content information using a frequency of occurrence of each word of the content posted on the recommendation requester personal media;
Generating second media content information by using a frequency of occurrence of each word of content posted on other personal media other than the recommendation requester personal media; And
Generating n virtual relationship groups including a plurality of different personal media, wherein n is one or more natural numbers, using content similarity between the first media content information and the second media content information.
Personal media recommendation method comprising a. The method of claim 2,
The media content information is a personal media recommendation method, characterized in that the frequency of appearance of each word as a vector element. The method of claim 3,
The content similarity between the first media content information and the second media content information is determined using an angle formed by a vector element constituting the media content information. The method of claim 2,
And the first media content information is generated from the recommendation requester personal media and a plurality of personal media directly / indirectly connected to the recommendation requester personal media. The method of claim 2,
Selecting the evaluation target group,
Determining content similarity between the first media content information and second media content information of the other personal media included in the n virtual relationship groups, respectively;
Obtaining a content similarity mean value for each virtual relationship group using the content similarity;
Selecting an evaluation target group using the content similarity mean value
Including,
The content similarity has a different connection weight according to a connection relationship between the recommendation requester personal media and the other personal media. The method of claim 6,
The connection weight is,
When the recommendation requester personal media and the other personal media is directly connected, indirect connection through the directly connected personal media, have a different value depending on the case where there is no direct or indirect connection relationship How to recommend personal media. The method of claim 6,
Recommend the personal media to the recommendation requester,
Performing evaluation on each of the plurality of other personal media using the content similarity between the recommendation requester personal media and a plurality of other personal media included in the evaluation target group;
Recommending at least one personal media to the recommendation requester using the evaluation result
Personal media recommendation method comprising a. The method of claim 8, wherein
Performing the evaluation,
Determining a subscription weight given to the intimacy according to the intimacy according to the connection between the recommendation requester personal media and the other personal media and the personal media subscribed to a specific Internet service; And
Determining the popularity of the other personal media
Further comprising:
Personal media recommendation method characterized in that the evaluation of each of the other personal media in consideration of the intimacy and the popularity. 10. The method of claim 9,
The subscription weight is,
When the recommendation requester personal media is subscribed to a specific internet service and the other personal media is subscribed to the specific internet service, the individual personal media has different values when the other personal media is not subscribed to the specific internet service. How to recommend media. 10. The method of claim 9,
The popularity is
Personal media recommendation method, characterized in that determined using any one or more of the number of comments, the number of trackbacks, the number of scraps or the number of favorites for the documents published on the other personal media. Media content information generating means for generating media content information of the personal media;
Virtual relationship group generating means for generating a plurality of virtual relationship groups including personal media having a connection relationship or content similarity with the recommendation requester personal media by using the media content information, wherein the plurality of virtual relationship groups are different from each other; Contains personal media of;
Evaluation target group selection means for selecting one or more virtual relationship groups among the generated plurality of virtual relationship groups as evaluation target groups; And
Personal media recommending means for recommending the personal media to the recommendation requestor by evaluating a plurality of personal media included in the selected evaluation target group
Personal media recommendation system comprising a. The method of claim 12,
The virtual relationship group generation means,
N virtual relation groups including different personal media using different first media content information of the recommendation requester personal media and second media content information of other personal media other than the recommendation requester personal media, wherein n is one or more Personal media recommendation system. The method of claim 13,
The evaluation target group selection means,
A content similarity average value is calculated for each virtual relationship group by using content similarity between the first media content information and second media content information of the other personal media included in the n virtual relationship groups, and the content similarity average value. Use to select the group to be evaluated,
The content similarity has a different connection weight according to a connection relationship between the recommendation requester personal media and the other personal media. The method of claim 12,
The personal media recommendation means,
Evaluating each of the plurality of other personal media using the content similarity between the recommendation requester personal media and a plurality of other personal media included in the evaluation target group, and recommending the personal media according to the evaluation result. Featured personal media recommendation system. 16. The method of claim 15,
The personal media recommendation means,
Recommend personal media in consideration of the intimacy of the recommendation requester personal media and the other personal media and the popularity of the other personal media,
Personal media recommendation system, characterized in that the subscription weight is given to the intimacy according to whether or not personal media subscribed to a particular Internet service.

Images