A service framework for Temporal Link Analysis with historical segments integration

Link analysis is a widely applied technique for unveiling the social network structure among individuals, either in a Web setting or mobile environment. However, usability and efficiency for link analysis are hard to be obtained through prevailing frameworks. In this paper, we present IOHS (integration-of-historical-segments), a service framework for temporal link analysis (TLA). With IOHS, a TLA task is summarized into three subtasks, i.e. fitting the linkages of data in chronicle time intervals, assigning different weights to historical linkages of data and entities involved, and integrating the weights of linkages of data in the main analysis task, such as scoring or searching The main contribution of IOHS is factoring out the non-variant part of a TLA task into the framework to give researchers much freedom to concentrate on model and parameter fitting. Another advantage of using IOHS is that the difficulty in implementing the trial-and-error process is greatly reduced so that optimized models could be obtained with much ease. Our experimental results show that IOHS is of high usability and efficiency.