A semantic approach for text document clustering using frequent itemsets and WordNet

Abstract

Document Clustering is an unsupervised method for classified documents in clusters on the basis of their similarity. Any document get it place in any specific cluster, on the basis of membership score, which calculated through membership function. But many of the traditional clustering algorithms are generally based on only BOW (Bag of Words), which ignores the semantic similarity between document and Cluster. In this research we consider the semantic association between cluster and text document during the calculation of membership score of any document for any specific cluster. Several researchers are working on semantic aspects of document clustering to develop clustering performance. Many external knowledge bases like WordNet, Wikipedia, Lucene etc. are utilized for this purpose. The proposed approach exploits WordNet to improve cluster member ship function. The experimental result shows that clustering quality improved significantly by using proposed framework of semantic approach.

CBER: An Effective Classification Approach Based on Enrichment Representation for Short Text Documents

“…Combining the two approaches solves the limitations of the previous work by adapting the Lesk dictionary algorithm [16]. Experiments show that the CBER with its two approaches overcomes the limitations found in other works [9,18,21].…”

“…Moreover, a reduction approach is proposed to solve the problems of short text in classification [9,14]. This approach reduces the document features and exchanges them with new terms.…”

“…-In reduction methods [9,16], documents are represented only by their topics using Wikipedia or WordNet.…”

“…Wikipedia [7,9,13] and WordNet [3] are two main types of semantic knowledge that are involved in document classification. Semantic knowledge approaches represent text documents as a bag of concepts (BOC).…”

State of the art document clustering algorithms based on semantic similarity

“…However, sometimes when looking for some queries, we also get a lot of unrelated information about our query, with less related information [4], [5]. Therefore, document clustering is being used in all search engines to show the queries' results in an ordered and efficient way [6], [7]. Clustering documents aim to assist humans in searching and understanding information [8].…”