Hierarchical approach for scientific document classification

Classification is the grouping of information or objects in predefined labeled categories based on similarities. Exponential growth rates of scientific document collection leads to unmanageable manual classification. Feature extraction is the central prerequisite of automatic document classification. TF-IDF (term frequency-inverse document frequency) is commonly used to express the text feature weight. This paper proposes a new feature weighting method by modifying TF-IDF formula.