Negation Handling for Sentiment Classification

Sentiment Classification refers to the computational techniques for categorizing whether the sentiments of a text are positive or negative. Sentiment Classification approaches would suffer due to Negation Modifiers. Negation Modifiers, like word “not” modify the meaning of the associated word. Handling Negation Modifier is important as they may modify the Sentiment conveyed by the associated word. In this paper a model for handling Negation Modifiers for Sentiment Classification was proposed and evaluated. Generally a term in a document is recorded in term-document vector, but if the term was negation modifier, it was treated exceptionally. The negation modifier was ignored but the term next to negation modifier was recorded in term-document vector of opposite orientation. The proposed Negation Modifier handling model was evaluated using movie document dataset. Accuracy slightly increased after handling negation modifiers.