Improve spreading activation algorithm using link assessment between actors from a mobile phone company network based on SMS traffic

Marketing strategies and relationship management of customers are increasingly important today, so investments for these aspects of the business are growing exponentially. To carry out the above, it is necessary to take a look inside the stored knowledge of any enterprise that could visualize the commercial behavior and preferences of their customers. Telecommunications companies deal with a special type of information that is related to the connections that exist between customers. Such information can be used to build a network to examine how customers are related with each other. In this paper, we build a social network based on the analysis of terabytes of Call Detail Record (CDR) data from a telecommunication company to identify and to select the most significant variables that express the link between the actors. As a next step we define the degree of customer relationships using a weighting function based on business rules. Finally, we apply the spreading activation-based technique to predict potential churners.