Text graph- an enhanced graph fusion model for document clustering

Abstract

Text clustering is a well-known method for refining the eminence in information retrieval, which groups a huge number of unordered text documents into the subgroup of associated documents. It is a contemporary test to investigate minimized and meaningful experiences from substantial accumulations of the unstructured content reports. Different clustering techniques are in use to make the clusters in the text document accessible. This paper introduces a new technique of document clustering based on graph model. The collection of documents is denoted as the graphical network in which the node represents a document and an edge represents the similarity between the two documents. This paper intends a TextGraph algorithm based on the graph structure. The unstructured documents contain a vast number of features; it must be reduced before graph construction. The count and semantic-based feature reduction methods are used to select the vital features. Based on this feature, the algorithm constructs the text graph structures. This paper combines these (word count and semantic) text graph structure to generate a fusion graph model. In, fusion model, each document is associated to its k-nearest neighbors with weighted edges. Finally, on the fused TextGraph, the clustering is performed to group the documents. Experimentations are accompanied on real-time text datasets. The outcomes demonstrated that the proposed fusion graph model overpowers the prevailing methods and improves the outcome of text document clustering techniques in rapports with the purity and normalized mutual information.