Research On User Interest Identification In Folksonomy Based On Formal Concept Analysis

Along with the prevailing of user label use behavior, the system function of Folksonomy has been given higher requirements. As one of the important research fields in Folksonomy, how to meet the needs of personalized information of users’receive extensive attention of academia. As a conceptual clustering technique, Formal Concept Analysis can not only identify but also visualize concept by concept lattice in data set, provide formal tool and theoretical foundation for the analysis of Folksonomy. The research based on Formal Concept Analysis has been attention by part of information science and computer science researchers. However, the research of user interest based on Formal Concept Analysis in Folksonomy mainly focused on identifying friends relationship with common interest and calculating user interest by tags used frequency at the present stage. Without considering the structural of user interest, it is more rare to mine the potential interests of user by user group. Based on this situation, a series of user interest identification in Folksonomy based on Formal Concept Analysis problems has been researched in this paper.Based on the learning theories of Formal Concept Analysis and analyzing structure of concept lattice, this research proposes a method based on Formal Concept Analysis, which use for user interest identification in Folksonomy, including: ①Proposed a method used for calculating user interest, based on analyzing of single user’s R-T concept lattice. It considers not only the frequency of tag concept, but also the hierarchical structure of the concepts, witch more conforms to the cognitive structure of the user. ②Proposed the strategy based on concept lattice structure of user interest for interest recommending, in detail. It considers not only the degree of user interest, but also the hierarchical structure of the concept, thus the intension of recommend interest concept is more moderate. ③Based on analyzing of users’U-T concept lattice, this paper puts forward on the idea of mining user’s potential interest by association rules of users interest. ④Discussed the problem of probability factors may lead to association rules. Proposed a improved method to recalculate confidence, which is the basis of association rules judgment, to eliminate the influence of probability factor on association rules. ⑤Proposed the methods to identify user’s potential interest through the user group of the association rules and form the comprehensive interest.In this paper, tags,users and resources data from Douban film is collected and preproccessed. The proposed methods that identification of user interest and mining association rules has been analyzed and validated through some methods, such as case studies, mathematical statistics, questionnaire survey and so on. The results show that the degree of user interest concept identified by single user’s concept lattice is higher the others, in general; And according to investigation of many users, display that it does have relations between user interest as association rules mining by the method based on U-T users’ concept lattice, to a certain extent; And the modified confidence does eliminate the influence of probability factors on association rules, ensures reliability of association rules between interest concept. The empirical results show that the methods been proposed in this paper is effective, which offers a new way for user interest identification.and recommendation.