Text Categorization is an interesting field in the study of Textual Data Mining. It has attracted an increasing popularity with its explosive growth of textual documents. The documents are connected with exclusive multitude categories i.e sports, medical, health, and Olympic Games). Text categorization paves different opportunities for creating multi-label learning approaches that specifically to textual data. Text mining defines the processes of discovering useful knowledge patterns from textual data. This is one of the factors followed in automated text categorization. It is practiced by developing novel machine learning approaches. Anyhow, the ML model generates low expressivity. The ML model established using Train-Test scenario. In case the existing model is found deficient, the Train-Test-Retrain is developed which is time consuming process. In this paper, we proposed ?Pyramidal Cluster Membership Approach (PCMO)?. It works in two models namely, training and testing model. The training model comprised of four phases, Pyramid-Fuzzy Transmutation, Novel k-edge classifier, Cluster to Category mapping and finding the boundaries. These estimated boundaries are applied on new textual data and the categories are assigned. Experimental results on Freebase dataset show that the proposed approach based on pyramidal membership method can achieve better classification accuracy than the traditional approaches especially that includes over-fitting document categories.
展开▼
