Fish Schooling Genetic Algorithm for Text Document Clustering Using Pattern Features

Abstract

As digital text content increases on servers in form of blogs, research content, news retrieval of relevant information was hard. Here unstructured content increase the complexity of this retrieval process. Research has proposed several model for its clustering but unsupervised model is highly desired. This paper has proposed an document retrieval algorithm without any structural input of dataset. Here model is unsupervised so prior information of type of document not required. Fish Schooling Genetic Algorithm FSGA was used for the clustering of the documents. Proposed model use pattern feature from the content to evaluate distance between documents. Experiment was done on real dataset where research papers of different fields were taken to  cluster. Comparison of model was on existing methods and it was obtained that use of fish schooling genetic algorithm improve clustering accuracy by 6.92%.