Social Network Based Web Service Recommendation Algorithm

As a new distributed computing model, Web services has played a more important role in e-commerce, enterprise application integration etc. With the growth of Web services, how to dig the users’interest and help users accurately find the interested services, has become the research focus in the field of Service Computing. According to the shortcomings and insufficiency of the traditional recommendation algorithm, this paper proposes a social network based Web service recommendation algorithm.First of all, according to the shortcomings of Pearson correlation coefficient, this paper proposes three kinds of users’ similarity algorithm. In order to solve the data sparsity, this paper also puts forward a similarity transfer algorithm, through the similarity transfer to update and fill the users’similarity. The experimental results show that the similarity algorithms proposed in this paper have similar NMAE with the Pearson correlation coefficient, and at the same time, the similarity transfer algorithm is effective to reduce the NMAE.After calculating the users’ similarity, this paper establishes the user’s similar network, and puts forward a user similar network partition algorithm. This algorithm divides similar users into the same group, and the user similar network based recommendation algorithm uses the users who are in the same group with the target user as candidate users for recommendation. This algorithm can expand the similar users, and improve the accuracy of recommendation.This paper also uses users’ trust relationship for recommendation, and proposes a user trust network based recommendation algorithm. Finally, this paper combines users’ similar network and trust network for service recommendation, which can make the recommendation results meeting users’ interest, and at the same time also have the credibility. The experimental results also show that social network based service recommendation algorithm has lower NMAE than the collaborative filtering algorithm.