A Mixture Approach for Multi-Label Document Classification

Abstract

Multi-label classification learning concerns the determination of categories in the situation where one pattern may belong to more than one category. In this paper we propose a mixture approach, named FSMLKNN, which combines Fuzzy Similarity Measure (FSM) and Multi-Label K-Nearest Neighbor (MLKNN) for multi-label document classification. One of the problems associated with KNN-like approaches is its demanding computational cost in finding the K nearest neighbors from all training patterns. For FSMLKNN, FSM is used as an efficient clustering approach before MLKNN is applied. For a document pattern, its K nearest neighbors are only calculated from the closest cluster having the highest fuzzy similarity to the document pattern. Experimental results show that our proposed method can maintain a good performance and achieve a high efficiency simultaneously.