Developed Hybrid Self-Organizing Map, Weighed Quantum Particle Swarm Optimization with Improved K-Means Clustering for Document Clustering

Computational Intelligence (CI) is a computational methodology inspired by nature that includes strategies and techniques for addressing complex problems by adapting to changing environments. CI approaches encompass hybrid paradigms such as artificial machine learning, fuzzy systems, swarm intelligence, and evolutionary computation. The aim of clustering is to group a collection of items into subsets or clusters that are internally consistent while being significantly different from one another. Document or text clustering refers to the process of organizing a set of written materials into clusters such that the documents within each cluster share similar characteristics in different clusters are distinct. For document clustering, a hybrid approach combining Self-Organizing Map, Weighted Quantum Particle Swarm Optimization, and Improved K-Means Clustering (SWI) is proposed. The proposed hybrid method consists of three stages. In the first stage, SOM is used to generate prototype matrices. In the second stage, WQPSO is employed to identify the initial centroids for K-Means. Finally, in the third stage, the K-Means clustering algorithm organizes the preliminary vector representations into clusters. This method addresses some of the limitations of existing K-Means, including its sensitivity to initial cluster center placements, its struggles with clustering high-dimensional data, and its tendency to converge to local minima. The performance of the proposed clustering technique is compared with other methods, the word frequency-based hybrid SWI approach for clustering vectors generated from sense-labeled texts.