Main Article Content
Abstract
Sentiment mining is considered as the trending topic of research in determining the potential of the product. Different methodologies are employed through which the information is propagated to produce positive or negative mood about a person or an event. In particular, the sentiment analysis domain includes the benefits of opinion mining for estimating the peoples’ action in reactive to the action taken by the government. In this paper, a Gray Wolf Optimization (GWO)-based Enhanced Long Short-Term Memory (LSTM) Neural Network Scheme (GWO-ELSTM-NNS) for opinion mining is proposed for effective prediction of the semantics and decisive conclusions determined from the process. The text and emojis of the text related to the tweets are mined separately for successive aggregation that aids in the indispensable generation of opinion polarity interpreted in the tweet. In this scheme, the historical data pertaining to the emojis associated with opinions are used for training the LSTM in order to minimize its prediction error in the opinion mining process.