Research On Resource Recommendation Of Knowledge-Sharing Community Based On Folksonomy

With the development of Web technology,knowledge-sharing community,which has the characteristics of bottom-up and self-organization,is emerging gradually.With the accumulation of resources in the knowledge-sharing community,how to help the users in the community find the resources they need is the main problem facing the further development of the community.Folksonomy is the frequently-used resource classification and organization method in the knowledge-sharing community,the tag set reveals the semantic relationship between resources.By combining the semantic relationship between resources and recommendation algorithm,the recommendation efficiency of the algorithm can be effectively improved.Taking resource recommendation in the knowledge-sharing community as the research object,this paper proposes a resource recommendation method based on folksonomy.Firstly,a method for calculating semantic similarity of resources based on folksonomy is proposed.Tags are constructed as a structured tag-tree to solve the problem of semantic ambiguity and lack of semantic structure of tags.Then the semantic similarity of tags based on co-occurrence and based on tagtree structure is integrated to determine the semantic similarity of tags.The resources are classified according to the tagged status,and the semantic similarity between resources is obtained based on the classification status of resources and the semantic similarity between tags.Then a resource recommendation model based on semantic similarity of resources is proposed.The constructed tag-tree is used to perfect the process of resource retrieval so as to improve the recall rate of retrieval.The semantic similarity of resources is used to improve the traditional user-based collaborative filtering algorithm.The user-resource evaluation matrix is filled based on the resource semantic similarity,then the filled user-resource evaluation matrix is used to computing the similarity between users and determine the user's most nearby users,so as to mitigate data sparsity and cold-start problems.Finally,the proposed algorithm is verified on the Movie Lens dataset,and the experiment proves that the proposed algorithm can effectively improve the recommendation efficiency.