Community search on attributed weighted bipartite graphs is an essential problem and its aim is to retrieve high-quality communities containing a query vertex. An important evaluation metric to evaluate the quality of the output community is the attribute cohesiveness of the community search model on attributed weighted bipartite graphs. Nevertheless, the attribute cohesiveness of the existing community search model on attributed weighted bipartite graphs is relatively low, resulting in poor quality of the output communities. To retrieve high-quality communities for a query vertex on attributed weighted bipartite graphs, we propose a novel k-bitruss-Attributed Weighted Community model, where the structure cohesiveness is described by the k-bitruss model, and the attribute cohesiveness and the interaction cohesiveness is maximized. To support fast retrieval of k-bitruss-Attributed Weighted Community, we propose a new three-step approach. Under this three-step approach, a novel indexing technique and efficient algorithms are proposed to retrieve the k-bitruss-Attributed Weighted Community. We conduct comprehensive experiments on five real-world graphs, the experimental results demonstrate that the attribute cohesiveness of the k-bitruss-Attributed Weighted Community model surpasses all baselines and the attribute cohesiveness of the k bitruss-Attributed Weighted Community model exhibits greater stability compared to the existing community search model on attributed weighted bipartite graphs.