Enhanced feature selection with bacterial foraging and rough set analysis for document clustering

Most applications, such as Information Retrieval and Natural Language Processing (NLP), utilize document clustering to improve their analysis. The document consists of various features that are utilized to determine the similar and dissimilar documents. However, the traditional techniques consume high computation difficulties and convergence problems while analyzing high-dimensional data. The research difficulties are addressed with the help of Bacterial Foraging and Rough Set Analysis (BF-RSA). This study uses the TF-IDF features for analyzing similar documents. The extracted features are explored using the Bacterial Foraging Optimization (BFO) approach that uses the exploration and exploitation characteristics to improve the overall clustering quality. The collected documents are analyzed using a roughest approach that generates the discernible matrix which helps to identify similar and dissimilar features. Then bacterial foraging method computes the fitness value according to their behavior to identify the optimal solution. The selected feature set is further analyzed in the roughest approximation condition to minimize the uncertainty and interpretability issues. The effective integration of bacteria foraging and rough set approach maximizes the feature selection accuracy and improves the clustering accuracy (97.05%) with minimum convergence speed (0.063 s).