Fusion of deep recurrent neural network models and fuzzy decision support system for tweet sentiment analysis and classification

With the advent of social media and networking, Tweet Sentiment Analysis (TSA) has become a significant methodology to extract useful information from the “X” platform users. The extraction of useful information from the tweets enables us to develop a decision support process from which necessary opinion mining can be carried out. These opinions help improve business models, product reviews, customer satisfaction, and thereby improve services and quality of any system or product. In this research paper, an innovative fusion of a deep recurrent neural network (DRNN) and a fuzzy decision support system (FDSS) is done to evaluate customer satisfaction based on tweets. The proposed ensemble Bi-directional Long Short-Term Memory (Bi-LSTM) and Gated Recurrent Units (GRU), which are deep recurrent neural network models, were employed to attain the polarity of the tweet sentiments (positive, negative or neutral). The innovative fuzzy decision support system determines the contentment level of customers based on the tweets. FDSS with ensemble Bi-LSTM-GRU (eBi-LSTM-GRU) handles uncertainties and imprecision in tweet sentiments, enhancing performance. The developed fusion model attained an accuracy of 84.33%, 96.92% and 93.81% for the sentiment 140 dataset, the T4SA dataset and the airline Twitter dataset, respectively. The proposed fusion eBi-LSTM-GRU-FDSS model outperforms previous baseline approaches.