Many activities under certain social contexts can have significant business and social impacts. For instance, large amounts of fraudulent activities in social security program may result in huge government customer debt. Given huge volumes of unbalanced activity transactions, activity mining is devised to discover impact-targeted activity patterns. This seminar will discuss the methodologies and challenges of activity mining and its applications in the domain of social security. Moreover, using the governmental social security datasets as an example, Prof. Zhang will illustrate the issues and the solutions in mining high impact activities from rare, dispersed and imbalanced data set through exceptional behaviour analysis.