Research On Trust Degree Prediction Of User In Social Networks Based On Domain Topics And Torpology Features

With the rapid development of computer technology and the popularization of social media,social networks have become one of the main ways for people to get to know each other,communicate,and share information.However,in the process of network socialization,people do not need direct face-to-face communication,and they are often strangers.Therefore,it is particularly important to research and predict the user trust degree in social networks.In the field of user trust degree prediction research in social networks,most of the existing trust prediction methods only use user semantic information or topological features in networks,and the constructed of user trust information is often subjective,unstable and difficult to acquire.In addition,these methods are usually only applicable to small-scale networks or have lower efficiency on large-scale networks.To solve these problems,this paper presents a comprehensive prediction algorithm of user trust degree in social network.That is a trust inference algorithm based on topic similarity and trust propagation(referred as to TI-TS-TP).This algorithm comprehensively considers the information of user domain topics and node topology features in networks,and defines two indicators to measure the user trust degree,which are the user topic weighted similarity and node trust propagation ability.Then,it adopts a kind of hierarchical comprehensively strategy based on considering the weighted topic similarity first and then the trust propagation ability(referred as to TS-TP).So that the strong trust path sets between users are extracted from the original social networks by selecting the trusted top-k neighbors and executing the breadth first search based on limited depth of L and trust threshold of (referred as to -).Based on the strong trust path sets,the trust value between users can be calculated with four kinds of trust integration strategies.Finally,to verify the validity of TI-TS-TP algorithm,many experiments are tested in the public and real-word social network data set.The experimental results show that our algorithm has more accuracy of trust prediction than the typical trust prediction algorithm,and it has the same level of time complexity.In addition,our algorithm has lower time complexity than the traversal trust prediction algorithm.At the same time,it just has a slight decrease of trust prediction accuracy.The study of user trust degree prediction in social network contributes to the user to judge whether users or services are trustworthy in the process of network socialization,which helps user to make a decision about whether to interact or interact with whom.In addition,this paper adopted a hierarchical synthesis strategy TS-TP and proposed a trust inferring algorithm TI-TS-TP.So that has an important reference significance for the research of social network analysis,network security and other academic fields.