A weighted seeds affinity propagation clustering for efficient document mining

Clustering is widely used in data mining and learning systems. It is not one specific algorithm, but a general task to be solved which can be achieved by various algorithms that differ significantly in their notion of what constitutes a cluster and how to efficiently find them. However the clustering is not easy task especially for the complex datasets like text mining where the information does not depends only on terms frequency. This paper presents an effective approach for dealing with similar problems. The proposed algorithm is a category dependent weighted seeds affinity clustering algorithm. The advantage of the proposed algorithm is that clusters can be easily modified according to the field of interest of the user. The superiority of the proposed algorithm is also validated by the simulation results comparison using Reuters-21578 dataset. Results shows improvement over k-means, Affinity and Seeds Affinity Algorithm.