Detection of viral messages in twitter using context-based sentiment analysis framework

The prevalence of social media sites like Twitter has made it simpler for individuals and organizations to disseminate incorrect facts or misinformation that can sway public opinion and behavior. It is crucial to create a trustworthy system that can recognize the sentiment of tweets in their context, analyze that sentiment, and pinpoint those that are spreading quickly and could be potentially damaging or deceptive. Hence, to identify viral tweets on Twitter, we suggest a new remora-optimized twofold gated attention neural network (RO-TGANN). This research’s word representation also creates weighted word vectors by including sentiment data in the term frequency-inverse document frequency (TF-IDF) algorithm. The resulting vectors are entered into RO-TGANN to better represent the comment vectors and efficiently collect context information. Multi-layer perceptron (MLP) classification is also employed to determine the sentiment pattern of the message. The proposed technique is contrasted with the current sentiment analytical techniques under comparable circumstances. According to the empirical results, the suggested analytical approach for sentiment classification has a greater accuracy, f-score, precision, and recall. The creation of such a technique can aid in the drive to encourage ethical social networking usage and limit the transmission of dangerous posts on social networking sites.