Text Classification by Augmenting Bag of Words (BOW) Representation with Co-occurrence Feature

Abstract

Text classification is the task of assigning predefined categories to free-text documents based on their content. Traditional approaches used unigram based models for text classification. Unigram based models such as Bag Of Words(BOW) models are not considering co-occurrence of set of words in a document level. This paper proposes a way to find co-occurrence feature from anchor text of wikipedia pages, proposes a way to incorporate co-occurrence feature to BOW model. Finally the method is analyzed to know how it performs in task of text classification.