A parameter-free label propagation algorithm using bipartite heterogeneous networks for text classification

Abstract

A bipartite heterogeneous network is one of the simplest ways to represent a textual document collection. In such case, the network consists of two types of vertices, representing documents and terms, and links connecting terms to the documents. Transductive algorithms are usually applied to perform classi cation of networked objects. This type of classi cation is usually applied when few labeled examples are available, which may be worthwhile for practical situations. Nevertheless, for existing transductive algorithms users have to set several parameters that signi cantly affect the classi cation accuracy. In this paper, we propose a parameter-free algorithm for transductive classi cation of textual data, referred to as LPBHN (Label Propagation using Bipartite Heterogeneous Networks). LPBHN uses a bipartite heterogeneous network to perform the classi cátion task. The proposed algorithm presents accuracy equivalente or higher than state-of-the-art algorithms for transductive classi cation in heterogeneous or homogeneous networks.