Blended Sentiment Polarity (BSP) for Product Review Analysis

Abstract

Product reviews are very much important for buyers in helping them to make decisions. For judging a review sentence, a lot of opinion mining techniques have been used. Normally, Neural Network (NN) learns the valuable representation automatically without human efforts. The key challenge is to choose the positive or the negative orientation. Nowadays, deep learning techniques have come forward to be a successful way for resolving the sentiment classification problems. The proposed framework consists of three layers. The first layer is the input layer, where the input sentiments are converted to vector using the Word2Vec method. The second layer is an embedding layer, where vector is embedded using the GloVe and the proposed Blended Sentiment Polarity (BSP) techniques. The proposed BSP is comprised of the scores containing TextBlob and VADER methods. The third layer is the classification layer, where the Long Short Term Memory (LSTM) is applied. Therefore, the LSTM with BSP for product review sentiment analysis has been proposed. The proposed framework is evaluated using 1.1M weakly labeled review sentences and 11,754 labeled review sentences from Amazon. Experimental results show that the LSTM with the proposed method provides good results.