A Similarity Measure for Text Document Using Term Cardinality

With enormous development of digital technology, data is being generated at rapid rate with various application domains. Data has to be extracted or filtered to find useful information. A basic concept for these tasks and applications are the distance measures to effectively determine how similar two objects are. In this paper, a novel similarity measure for clustering text documents is proposed using the cardinality of the terms in the documents. The bench mark algorithm k-medoids is used for clustering task. The results obtained from the proposed distance measure are compared with other standard distance measures like Manhattan, Euclidean distance measure. Dunn Index is used to analyze the cluster validation of the results obtained from the distance measure.