An Efficient Document Clustering Method Using Improved Bacterial Colony Optimization

This paper presents an efficient document clustering method based on an improved Bacterial Colony Optimization (BCO) algorithm. The proposed approach integrates a centroid opposition- based learning (COBL) strategy for effective population initialization and incorporates Quantum Computing-inspired position updates to enhance the search capability of BCO. To strengthen clustering performance, multiple feature extraction techniques—Bag-of-Words (BoW), Term Frequency-Inverse Document Frequency (TF-IDF), Word2Vec, and GloVe—are utilized in a hybrid manner to capture both statistical and semantic document representations. The framework is rigorously evaluated on six benchmark datasets: Reuters-21578, Classic4, WebKB, SMS, BBC, and DMOZ, and its performance is compared against traditional and evolutionary clustering. Experimental results demonstrate that the proposed COL-QBCO framework significantly outperforms the baseline methods in terms of clustering accuracy, normalized mutual information (NMI), F-measure, and convergence rate, highlighting its robustness and effectiveness in handling diverse and high-dimensional text datasets.