Dual Bi-LSTM-GRU based stance detection in tweets ordered classes

There has been a tremendous increase in social media text-based opinions and reviews as a result of the quick development of social media. It emphasizes those users on platforms like social media often express their viewpoints, favor, or against for other views or ideas. The existing stance detection methods can determine users’ stances based on the text through various classes and ordered classification. This implies a possible vulnerability in the current stance detection methods, which primarily focus on binary type (favor or against), an enormous range of stances for specific sequences in which they were expressed. To tackle these challenges, we proposed a model that can effectively handle stance detection using ordered classification (SDOC). The SDOC model accurately identified the sequence and arrangement of different positions within the given text. So our proposed model, SDOC, is based on Dual BiLSTM-GRU models for word embedding techniques (word2vec and glove 100, 200, and 300 dimensions) to different epoch sizes. The proposed model performs better accuracy, precision, recall, and F1-Score for specific epoch values (5,10,15).