Optimising Sentiment Classification using Preprocessing Techniques

Sentiment Classification refers to the computational techniques for classifying whether the sentiments of text are positive or negative. Sentiment Classification being a specialized domain of text mining is expected to benefit after preprocessing. In this paper we propose various models with selective combinations of preprocessing techniques and Sentiment Classifiers, to optimize Sentiment Classification. Unlike traditional preprocessing technique where punctuation symbols are discarded, we proposed a set of rules to handle words with apostrophe and then remove punctuation symbols. Sentiment Classifiers that were proposed in our previous research articles are based on term weighting techniques. We evaluated Sentiment Classification models by comparing them with state of art techniques using the movie sentence and movie document dataset. Accuracy increased from unprocessed dataset to preprocessed data. Our Classifiers handled stopwords thus had hardly any impact of stopwords removal in preprocessing unlike traditional Sentiment Classifiers. Our classifiers also displayed accuracy better than traditional classifier and another surveyed classifier based on term weighting technique. Keywords— Sentiment Classification; Pre-processing; Term Weighting; Term Frequency; Term Presence; Document Vectors