The widespread popularity and vigorous growth of micro-blogging systems provides a fertile source for analyzing social networks and phenomenon. Currently, few data mining tools can deal with unique characteristics of microblogging systems. In this study, we propose an integrate approach for mining user relationships in micro-blogging systems. The approach starts from macroscopic analysis of social networks by grouping users with the method of maximal strongly connected components (MSCC). Following that, a measure of condensation level of groups are calculated to find out the most influential group , and all groups can be ranked according to this measure; then a new algorithm is presented to evaluate the influence of a specific user within a group. The integrated approach is capable to analyze large amount data sets. It is useful for exploring directions of information diffusion and evaluating the scope and the strength of individual user's influence in micro-blogging systems.