Research On Recommendation Algorithm Of Group And Source Credibility Based On Trust

With the rapid development of information technology, people’s daily life isoccupied gradually by social networks, such as electronic contacts, which try to obtainsome user’s preference information accurately and then recommend them to other users orpotential customers in order to gain benefits. Therefore, it drives the recommendationalgorithm of research. However, the traditional recommendation algorithm not onlydoesn’t take the user’s relationship into account, but also does not consider the users’characteristics in different groups, it does not work well. To address this issue, this papertries to improve the group recommendation and source credibility with the trust theory.First of all, this paper summarizes the traditional thoughts of current recommendationalgorithm and analyzes existing trust model method. Especially giving more attention torecommendations algorithm based on CF and respectively explaining explicit and implicitthe trust model method of as well as the different involves algorithms. To solve existingproblems that group recommendations without considering the user character andrelationship, proposed the group’s recommendations based on the different users’preferences and trust relationship. These methods choose different factors onpolymerization and identify the most reasonable polymerization strategy. According to theproposed algorithm the mean absolute error and the improved accuracy are selected asevaluation standards..Secondly, in view of the problems of traditional CF, proposed trust recommendalgorithm based on theory which expand the dimension of traditional cooperative filteralgorithm by calculating the expertise, credibility and similarity in the source credibility tochoose neighbors set, and set up different significant weight according to the users’ viewin these three aspects to provide more reasonable and personalized recommendation. Inaddition, giving different benchmark system and evaluation standard to measure theeffectiveness of the algorithm.Finally, this paper designs two simulation experiments. The first experiment analysesthe accuracy and validity of various group recommendation algorithm with differentpolymerization algorithm; The second experiment analyses the performance of trust recommendation algorithm based on source credibility and compare it to otherbenchmarks system.