A methodology for text classification based on feature clustering

Recent advances in feature clustering offer a viable alternative for text classification. In this paper, we propose a methodology in which all the terms are clustered based on χ² measures and only part of the clusters are selected to build the feature space. We also present that, higher precision and recall could be acquired when we select feature clusters base on their average χ² measure and take CF-IDF to weight the clusters after clustering in comparatively lower cluster size