Sentiment analysis of product reviews using weighted distance-based whale optimisation assisted deep belief network

This paper proposes a new sentiment analysis of product review based on intelligent techniques. The proposed model involves six stages: pre-processing, keyword extraction and its sentiment categorisation, semantic word extraction, semantic similarity checking, feature extraction and classification. Initially, the MongoDB documented tweets are subjected to pre-processing steps like stop word removal, stemming, and blank space removal. Further, the keywords are extracted from the pre-processed tweets. With respect to extracted keywords, the existing semantic words are extracted after categorising the sentiment of keywords. To the next, the semantic similarity score with the keywords is measured. The upcoming stage is the feature extraction, which uses two holoentropy measures like joint holoentropy, and cross holoentropy. The classification of the extracted features is done using a deep learning classifier named DBN, in which the optimised activation function is done by WD-WOA. Finally, the enhancement of the proposed model over the conventional models is evaluated through an effective comparative analysis in terms of positive and negative performance measures.