Improving Performance of Elastic K-Means Clustering using similarity Measures

Abstract

Clustering assign a proper membership to the data. Clustering has numerous applications inmarket research, pattern recognition, data analysis and image processing. Standard K-means clustering uses crisp membership to assign the data to a single cluster only. In real world data, some noise or ambiguity is present, so k-means clustering is not able to handle those data and it generates wrong membership for the clusters. In the proposed system, Elastic k-means clustering is used for creating flexible membership. For that purpose elastic k-means clustering uses vectorization and similarity measures for improving the membership and also handles the problem of the noisy data. Unstructured documents are used for experiment and the experimental resultsshows that, the proposed system gives better accuracy than existing one.