An Improved Text Feature Selection for Clustering Using Binary Grey Wolf Optimizer

摘要

Text Feature Selection (FS) is a significant step in text clustering (TC). Machine learning applications eliminate unnecessary features in order to enhance learning effectiveness. This work proposes a binary grey wolf optimizer (BGWO) algorithm to tackle the text FS problem. This method introduces a new implementation of the GWO algorithm by selecting informative features from the text. These informative features are evaluated using the clustering technique (i.e., k-means) so that time complexity is reduced, and the clustering algorithm's efficiency is improved. The performance of BGWO is examined on six published datasets, including Tr41, Tr12, Wap, Classic4,20Newsgroups, and CSTR. The results showed that the BGWO output outperformed the rest of the compared algorithms such as GA and BPSO based on the measurements of the evaluation. The experiments also showed that the BGWO method could achieve an average purity of 46.29%, F-measure of 42.23%.