Research Of Spam Call Detection Based On Human Social Graph

With the rapid development of mobile communication technology,more and more people began to use mobile phones.However,when people enjoy the convenience of mobile communication to bring life,they also encountered new problems.More and more groups and individuals began to initiate voice spam to mobile phone users to achieve their interests,and spam has brought a lot of trouble to people’s lives.Spam call recognition and filtering research have been proposed to solve this problem.At present,the common methods of spam call detection mainly include identification method based on communication behavior,recognition method based on reputation model and recognition method based on speech recognition,and they have achieved good results in their respective fields.However,there are many shortcomings in these methods,such as the recognition method based on communication behavior can’t adapt to the dynamic changes of communication behavior of users,the recognition method based on speech recognition has the problem of violating users’ privacy and the identification method based on reputation model only consider the users’ overall credibility,it did not consider the users’ personalized issues.To solve these problems,this paper analyzes users’ call record data,and proposes a spam-call detection algorithm based on the user’s social graph.First we preprocesses the original call record data and build user-level social graph based on user relationship,and we propose a method of calculating trust degree based on call duration and number of calls,then we analyze the characteristics of the graph and users’ social behaviors.We also propose two kinds of spam detection mechanisms in this paper: one is for the indirect connection node,we calculate the local trust between the two nodes based on the shortest path algorithm Dijkstra and local trust algorithm Tidal Trust,then we identify whether the calling number is spam based on local trust;the other is for the isolated node,we calculate the node’s global trust in the social graph based on Page Rank algorithm,then we identify whether the calling number is spam based on global trust.Experiments show that the proposed spam-call detection algorithm has a good recognition effect.