Web Knowledge Recommendation Based On User Tag Network

In the era of Web2.0, Internet has become an available repository with tremendous knowledge, while "knowledge overload","knowledge confusing" and some other problems are troubling users. Personalized knowledge recommendation is one important solution to these questions. Among Web2.0sites, social tagging systems enable users to tag knowledge resources which are coordinated with user’s personal interests. By mining and refining user generated tagging information, Web knowledge recommendation can be carried out. This paper tries to investigate such questions by using user tag network, aiming at a proper approach for Web knowledge recommendation.Firstly, this paper describes some basic theories of social tagging system and personalized knowledge requirement, and concludes basic concepts of Web knowledge recommendation system as well as its common components. Then, in this paper after modeling user tagging behaviors, the underlying construction theories of user tag network are interpreted. Applying network structural analyzing theories such as those in social network, complex network and network theory, this paper comes up with individual, sub-regional and over-all analyzing index systems. The progress of analyzing the structure of user tag network is essentially for organizing user tags, from which user’s personalized interests and requirements can be disclosed by tracking user’s tagging behaviors. Then, based on the former analyzing index systems, this paper gives a new user interest modeling method based on user tag network, with full detailed theories and steps. Furthermore, a framework of Web knowledge recommendation system is introduced. Two similarity methods are given to calculate the similarity between knowledge resources and user interests, which leads to recommendation resources for users.In the end, using real dataset, this paper shows and analyzes user interest modeling based on user tag network, as well as the precision of Web knowledge recommendation systems based on this modeling approach. This shows in favor of the validation and maneuverability of this method in this paper.