An Efficient Pretopological Approach for Document Clustering

In this paper we propose a new document clustering approach that does not require distance metric for aggregating multi variables in order to measure the similarity between documents. Based on calculated coherence (or pseudoclosure) function, closure set of pretopology concepts, we suggest a new proposition for cluster exploring when the connection between data is represented by one/many equivalent relations. Our approach also shows the data structure aggregating multi-criteria by viewing subsequent levels of pseudoclosure function which could represent data expansions. Furthermore, noisy data could be automatically detected by using our work as all elementary closures that have very small size of cardinality signifies their limit connections to the others. We also compare our approach with the popular K-Means and Fast Pair Nearest Neighbor with a given document collection for evaluating the efficiency and performance.