Self-Constructing Feature Clustering for Text Classification: An Automated Approach

Abstract

Text classification is a pivotal aspect of natural language processing, requiring advanced techniques for feature extraction and representation. This paper presents a novel approach to feature clustering in text classification, employing a self-constructing algorithm enriched with statistical membership functions to address the challenge of efficient text classification.  The proposed method efficiently reduces the dimensionality of the feature vector by grouping words into clusters. Each cluster is represented by a single feature, automatically generated through a process that considers the equality or dissimilarity of words. The clustering is driven by membership functions incorporating statistical mean and deviation, ensuring robust and representative feature grouping. The automatic creation of clusters enhances adaptability to diverse textual datasets. The integration of self-constructing feature clustering with statistical membership functions contributes to a scalable and adaptive solution for text classification tasks. Experimental results demonstrate the effectiveness of the proposed method, showcasing its ability to enhance text classification performance through efficient feature representation.