Fuzzy Document Clustering using Weighted Conceptual Model

摘要

Document clustering techniques mostly rely on single term analysis which can not reveal the potential semantic relationship between terms. To better capture the semantic subject of documents, this study proposes weighted conceptual model for document presentation. The new model divides the document concepts into centroid concepts and peripheral concepts due to their semantic relations to subject. The semantic similarity between two documents is calculated by centroid concepts and peripheral concepts respectively. A fuzzy semantic clustering method is put forward bases on the new semantic model. Experimental results show that the method enhances semi-structured document clustering quality significantly and outperforms K-Means and Fuzzy C-Means.