Discovering associations between news and contents in social network sites with the D-Miner service framework

Very often, correlation analysis of behavioral patterns between social network sites and the society suggests that people's behaviors in social network sites are independent from external influences. Recently, some research works have demonstrated that the assumptions are not always true. The work presented in this paper shows an approach to identify the possible associations between social network sites and the society. It utilized the D-Miner service framework in which different social network analysis tools can be plugged-in and used. The framework is supported by multi-agents, which include crawlers for different social network sites, schedulers to dispatch user requests, and analysis engines with different analytical algorithms. Two new agents have been developed for the association identification. A crawler agent is to collect incidents in the society and an association agent is to identify which social media messages are correlated to corresponding incidents. These identified associations can be applied to the evaluation of correlation analysis such as tracing the information propagation between social network sites and the society; and indentifying whether the correlations of behavioral patterns between social network sites and the society have been dominated by those incidents or not. The new agents have been tested with satisfactory results in identifying the number of connections which support the association between social network sites and the society.