The Study And Implement On Friend Recommendation Algorithm Based On Ontology Of Online Social Network

At present,most of the SNS products or APPs all offer the function about friends' recommendation.Traditional recommendation algorithms,such as Content-based Recommendation(CB),Collaborative Filtering-based Recommendation(CF),or Hybrid Recommendation,have been applied on many scene and acquired excellent results,but all of those cannot cover these algorithms' defects.Though there have been some improved plans based on those traditional recommendation algorithms,it also cannot fulfill the requests of high recommendation precision and low computation cost.With users' expect of recommendation system being more personalized and intelligentized,it is necessary to develop recommendation system based on ontology.All the information users published in SNS,to a large extent,can reflect users' characters,interests or other features.As a result,to find valid knowledge from the information,then understand users' true intent,and recommend potential friends,is a reasonable way to construct recommendation system.Based on ontology,this paper explains that how to make computer understand nature language,and explain how to find users' hidden attributes so that computer can describe users more comprehensively.Ultimately this paper design a recommendation algorithm to make recommendation system being more personalized and intelligentized.And the paper's main concepts are below:(1)With the method of OOP software engineering pre-analysis,this paper studys the knowledge on SNS,and set up SNS ontology model,then optimize the constructed model.(2)With the aid of existing semantic ontology WordNet,this paper supplements the SNS ontology.(3)This paper designed a friend recommendation strategy based on SNS ontology,and introduced a concept of entropy value method to match the weight of each calculation factor,thus improving the effect of matching concept algorithm.(4)This paper implement the recommendation system based on SNS ontology,through which the user can can transform key words of the sample data to the concepts in ontology,and then get hidden features.Based all the works above,the system gives the result set of friend matching.