Emotion Classification on Social Media Comments Using Categorical Feature Extraction Along With the Bidirectional Encoder-based Recurrent Neural Network Classification

Abstract

All across the world, social media is one of the most widely used platforms for information exchange. Comments on relevant information might be made in response to a video or any other piece of information. A remark may include an emotion that may be recognized by an automated recognition system. On Facebook, Twitter, and YouTube comments, we performed studies to determine their emotional categorization. A set of comments is gathered and manually classified using six fundamental emotion labels (happy, sad, angry, surprised, disgust, and fear) and one neutral label, with each emotion label representing a different emotion category. A prominent approach in natural language processing (NLP), deep learning has been used in a wide range of categorization applications. This procedure begins by preprocessing the input data with normalization, followed by categorizing characteristics in feature extraction utilizing the Linguistic and word count analysis (LIWC). Finally, for the categorization stage, the classify features might be supplied. Finally, for categorizing emotions, the Bidirectional Encoder based recurrent neural network classification approach is used. The studies have been carried out with the use of typical social media data that has been acquired from the kaggle data repository. The findings show that the suggested model outperforms all other existing mechanisms in terms of overall performance.