Abstract: The present paper summarizes the one-to-many matching model of bipartite model and corresponding methodology formalized in the process of extending a content-based load sharing algorithm, LARD. Such a one-to-many matching is referred to as k-perfect matching, and a charity algorithm is given as the solution. Some experimental results for the matching based on charity algorithm are then presented, with open issues and potential applicability in the context of hyper graph left as discussions.