Abstract | Telecommunication service providers could analyze the call logs of individual users to provide personalized services like recommending Friends and Family (FnF) numbers, that could be registered for having lower tariff during communication. However, suggesting such numbers is not simple as the decision making process can often be misled by factors like overall talk-time or call frequency. This paper proposes a technique to assist mobile phone users for effectively identifying FnF numbers using the underlying social network of a call graph. It resembles to a star network, having the user at the center, where each of the edges are weighted using three call log attributes: call duration, call location and call period. The weighted call graph is then used to calculate user's social closeness based on which FnF numbers are suggested. The experiment was conducted on a set of real life data and it is seen that the proposed technique can effectively suggest FnF numbers with an F-measure value of 0.67. |
---|