The Improvement in Detecting Social Network Change Based on Statistical Process Control

Social network analysis (SNA) has become an important tool for analyzing covert network, especially of terrorist networks. Detecting changes over time from an SNA perspective has an important significance. It may signal an underlying change ignored, and may even predict significant event and behaviors. It is promising to use abnormality detection method based on Statistical Process Control (SPC) in SNA. In this way before, people simply compute parameters of social network on average. However, in the real world, especially of the terrorist networks that whose degree distribution is likely to obey power-law distribution that which is uneven. So the parameters calculated by a simple averaging method can't precisely representative of the characteristics of the network. This paper presents an internal weighted averaging method to calculate the parameters, and applies this method to analyze the al-Qaida datasets. Compared with the previous results, it is significant that this method has better effects.