Improved Aspect Based Sentiment Analysis with DBN-RNN model

Abstract

Mining the sentiment target included in a sentence or text is the main aim of Aspect Based Sentiment Analysis (ABSA). This task's main challenge is to efficient extraction of a specific sentiment item's sentiment polarity. This work proposes a model namely Improved ABSA with Deep Belief Network-Recurrent Neural Network (DBN-RNN), which includes 3 working phases. Processes like stemming, stop word removal, lemmatization as well and tokenization are conducted in the initial pre-processing phase. Furthermore, in the aspect sentiment extraction phase, improved aspect term extraction (I-ATE) along with cosine similarity and word co-occurrence are used to extract the complex features from the pre-processed data. In the sentiment analysis phase, a hybrid classification model named DBN-RNN is utilized to effectively categorize the sentiments as neutral, positive, and negative polarities. The performance of proposed work is evaluated in terms of different performance measures.