Research Of Semantic Tag Recommendation Model And Algorithm Based On SCOT

Popularity of Web2.0 and Social tagging system making Folksonomy and tag recommendation system drew more and more attention. With the deep research of Semantic Web, researchers try to solve the disadvantage of Folksonomy and build a unified structure and semantic representation for it.In order to solve the problems of multi-system bottlenecks, recommendation quality is not high and can not recommended semantic tags which existed in traditional recommendation method, we propose a semantic tag recommendation model based on SCOT after combine tag ontology and Folksonomy, the key technology and principles of semantic extraction module, ontology integration module, ontology management module and semantic recommendation module of the model are also presented and detailed in this paper.Based on the model, we design a semantic tag recommendation algorithm based on the SCOT as an implement of the Semantic recommendation engine. The algorithm combines the advantages of the traditional tag recommendation methods and SCOT tag ontology, use co-occurrence instance of SCOT tag ontology to process a tag co-occurrence expanding for candidate sets which consist of resource description, user interest and resource tags, and then use tag semantic similarity based SCOT to process tag semantic disambiguation to get recommended sets, so as to effectively improve the recommendation quality and recommend semantic tags for users.Finally, we designed the prototype of the semantic tag recommendation model based on SCOT, and implement the semantic tag recommendation algorithm based on SCOT to verify the feasibility of our recommendation model and the effectiveness of our recommendation algorithm proposed in this paper. Experimental results show that the semantic tag recommendation algorithm based on SCOT which proposed under our model can improve the recommendation accuracy and recommend semantic tags, which further illustrates the semantic tag recommendation model based on SCOT is feasible, it can use Semantic Web technology and SCOT tag ontology to make up the shortage of traditional tag recommendation methods and play an important role under the Semantic Web.