Personalized Information Recommendation Based On Semantic Community In Tagging System

In the Web2.0 environment,the internet has become the world's largest database of information resources.While a large amount of information brings many conveniences to people,it also inevitably leads to problems such as "information flooding" and"information trekking." Personalized information recommendation can actively provide Web users with appropriate information according to the information needs of Web users,and is one of the important means to solve these problems.In the tagging system,the"user-resource-label" relationship network provides valuable basic data for personalized information recommendation.However,the existing related research mainly analyzes the relationship between users from the structural characteristics of the relation network,ignores the rich semantic information in the label system,and restricts the quality of the personalized information push model based on the social network.The study of this article mainly includes three aspects:Firstly,build a concept space based on tags.As a kind of user-generated metadata,the tag embodies the user's cognitive structure features and contains rich semantic information.After clustering users by tags,the interest structures of each user belonging to the same category have similar possibilities,and the probability of generating interactive relationships between them will be relatively high.Therefore,this paper firstly forms tag themes by tag clustering,and then builds a concept space model based on tags by constructing spatial vectors,association matrix,similarity matrix,etc.,to achieve the semantic level analysis of label topics,and to lay the foundation for subsequent clustering of users using tagsSecondly,a semantic community discovery algorithm for tag concept space fusion is proposed.Tag-based user clustering brings together Web users with similar cognitive structures.This tagging system can be used to construct hidden social networks.This paper proposes a semantic community discovery algorithm in the concept space of the fusion tag for the hidden social networks within each tag topic.The tag semantics(conceptual space model)of the internal nodes of the network are introduced into the community discovery process,making the semantic similarity of the users within the community as large as possible,the semantic similarity of users outside the community is as small as possible.In the specific implementation process,the similarity between users is first calculated according to the frequency of the use of the tag by the internal user of the hidden social network and the TF-IDF value corresponding to the user use tag.and the degree of similarity is used to measure the degree of affinity between users,thus forming a network of users.Then,based on the SemTagP algorithm,the semantic information of the internal nodes of the community is fully considered in the process of community discovery,and the concept space model is integrated with the community discovery to realize the division of the semantic community.Thirdly,based on the semantic community to achieve personalized information recommendation.The division of the semantic community is based on the hidden social network corresponding to each tag topic,so,the personalized information recommendation in this article is only for the community structure formed within each hidden social network.From the basic characteristics of the community structure,it can be concluded that the interests of Web users who belong to the same community have greater similarities,and the interests of Web users in different communities have greater differences.Therefore,in the real-time personalized information recommendation,the collaborative information filtering idea is used to implement personalized information recommendation within the community.Personalized information recommendation between clubs considers the possibility of strong ties between two nodes belonging to different communities,and utilizes nodes that play the role of "information bridge" to achieve personalized information recommendation between societies.