Context Driven Technique for Document Classification

In this paper we present an innovative hybrid Text Classification (TC) system that bridges the gap between statistical and context based techniques. Our algorithm harnesses contextual information at two stages. First it extracts a cohesive set of keywords for each category by using lexical references, implicit context as derived from LSA and wordvicinity driven semantics. And secondly, each document is represented by a set of context rich features whose values are derived by considering both lexical cohesion as well as the extent of coverage of salient concepts via lexical chaining. After keywords are extracted, a subset of the input documents is apportioned as training set. Its members are assigned categories based on their keyword representation. These labeled documents are used to train binary SVM classifiers, one for each category. The remaining documents are supplied to the trained classifiers in the form of their context-enhanced feature vectors. Each document is finally ascribed its appropriate category by an SVM classifier.