Polarity Discrimination of Evaluation Words in The Product's Review

The customer's review plays an important role in deciding the purchase behaviour for online shopping. This paper intends to extract useful information hidden in customer's reviews using the opinion mining technology. In the reviews, the polarity of the evaluation words reflects the consumer's attitude towards the product, so the discrimination of the polarity of the words plays a crucial role in the sentiment analysis. In this paper, we first make a polarity discrimination based on the sentiment lexicons. For the evaluation words that do not exist in the sentiment lexicon, the similarity between evaluation words and emotional seed words is calculated using the linear superposition of Word2Vec and HowNet (LSWH), and then the polarity of the evaluation word is determined. To evaluate the performance of the proposed method, we comparing with the previous two polarity discriminate methods, the result show that the proposed method has the high accuracy.