| Detecting the significance of relationships between people is a complicated and difficult task. This is especially true when these relationships are intermixed with hundreds of different mundane and ordinary interactions that are simply background noise. The identification and study of these relationships is called Social Network Analysis (SNA). SNA has many uses in academic, business, regulatory, and law enforcement applications. Sociologists have used SNA to identify and explain human and group behaviors. Economists have used SNA to streamline businesses and gain market advantage. Regulatory and law enforcement agencies use SNA to ensure various rules are being followed and to help identify key members of organized criminal activity.; Unfortunately SNA of complex and unknown relationships is very difficult and can only be performed off-line. Meaning, the data is collected over a period of time and then can be analyzed using various graph theoretic techniques. These techniques often remove all of the dynamic attributes of the relationships and often only discover the dominant and overt activity. In addition all temporal attributes of the relationships are lost in SNA since the data is typically flattened; removing any temporal dimension. By incorporating dynamic attributes such as temporal ordering, local functional role analysis, and aggregate network activity levels, we feel Dynamic Social Network Analysis (DSNA) can be solved as a process detection problem.; Using the Process Query System (PQS) technology as the basis, a novel approach will be generated for on-line DSNA. Due to the complex nature of DSNA, several modifications will be made to allow PQS to both map the social network as well as identify the primitive processes. In addition PQS will be modified to use abstract models that are not specific to the data stream, but of DSNA processes in general. This tool will provide various means for temporally correlating group activity, basic functional role identification as well as generating basic statistics of the primitive processes that are occurring. Allowing for a real-time approach to DSNA of large networks, that can supplement traditional analysis techniques. |
