Inference of mobile users' social relationships using Bayesian belief network

Today, mobile phones due to the rapid growth of new technologies and the emergence of a new generation of smart phones have become more than a communication tool that not only resolves the user's communication needs, but also provide them, services for many applications. List of mobile phone subscribers include slow-level simple concepts such as time and place of calls, talk time, etc. The analysis of these low-level concepts and their influences on each other can lead to identification of higher-level concepts such as user social relationships, feelings as well as many other concepts that are not recorded in the list. Data mining process as a powerful knowledge management techniques, by exploring the history list from the interaction of mobile subscribers, which represents intentions and actual behavior of each of them, as well as behavioral characteristics of each one can provide approaches and policies for social network analysis of mobile phone subscribers and ultimately brought intelligent semantic services by deduction and finally adaptation of behavioral and social implications. In this paper, due to uncertainty in the mobile sensors, Bayesian networks have been used to identify the relations of friendship between mobile users. In addition, tried to have a special look at each person and important parameters for creation of Bayesian network to be set independently according to each person's behavioral characteristics and behavioral similarities between the two individuals for every call. The results show that the proposed method has higher accuracy (0.80) than previous methods. After calculating, the probabilities of friends believe based on social network friendship that its analysis can play an important role in identifying influential actors in social networks. Therefore, the proposed framework due to focus on mobile social networks could be used in the real world.