Since more and more customers get interested in e-commerce, marketing strategy on the virtual community becomes a major issue. It is important to help companies to determine which potential customers to market to. In this paper, we study the trust relationships between customers on the virtual community and detect influential target groups of interest. We analyze Epinions, a large and popular online product review web site, using social network analysis method. We crawled data from this real-world community and discovered some properties of the so called "Web of Trust". As trust relationship is asymmetry, traditional clustering algorithms are not applicable to this scenario. Therefore, we proposed an algorithm to detect small target marketing groups according to the properties we had found. Our results show the effectiveness and utility of the method.