Objective: To extract sentiment or opinion words using repository of lexicons, to calculate overall polarity score for the product, to extract the aspects in the reviews, to devise polarity score for each aspects of the product and to develop a summary of the product aspects(targets) with its polarity score from the reviews.
Methods: There are several methods built up for sentiment analysis and opinion mining. In order to increase the recall, accuracy and precision openNLP parser with naïve bayes classifier is proposed. The opinion lexicons are used to produce summary about the reviews.
Findings: Opinion mining is a challenging Natural Language Processing or text mining problem. The reason behind this is we can’t exactly decide what user says about particular product. Because each one’s writing style would be different. All the reviews expressed in the websites cannot be processed directly. The reviews must be preprocessed in order to eliminate unnecessary characters. Many techniques were proposed for opinion mining but it lacks in accuracy, precision and recall.
Applications/improvements: To enhance the accuracy of opinion mining openNLP parser with naïve bayes classifier is proposed.
Keywords
