Decomposing a large graph into cohesive subgraphs plays a significant role in finding community structures in social network analysis. Among a lot of definitions of cohesive subgraphs, the k-truss is one of the best known cohesive subgraphs with a good trade-off between computational efficiency and clique approximation. In this paper, a quasi-truss decomposition algorithm for bipartite graphs is proposed based on the truss decomposition algorithm for general graphs. The proposed method can be used for analyzing the international business, such as the relationship between clients and sales volume in a certain period, and also analyze the social networking, such as users and hashtags in the twitter community. The twitter community is considered, and useful information can be extracted hierarchically by using the algorithm proposed in this paper.