Machine learning-based emotional identification of social media text
Aayam Chug, Sirisetti Sudeep, Polineni Arun Kumar, Akhil Sharma, Deekshitha Maganti, Enjula Uchoi · Computational Methods in Science and Technology · 2024
The method of sentiment analysis is employed to ascertain people&s;s thoughts, feelings, and perceptions of a wide range of topics, including people, objects, organizations, services, problems and goods. A subclass of sentiment analysis is emotion detection which forecasts the distinctive feeling as opposed to just expressing neutral, negative, or positive feelings. Many scholars have previously studied speaking and facial movements for the purpose of recognizing emotions in the past. Sentiment indicators like pitch, facial expression, tone tension and so on are absent in text, making the process of identifying emotions in text time-consuming. Using NLP techniques, many ways were previously proposed to extract emotions from text- the ML approach and the methods of using keywords and lexicons. However, keywords as well as lexicon approaches have inherent limitations since they emphasize semantic connections. In this post, we have introduced a hybrid (DL + ML) model to detect emotions in text. Bi-GRU and Convolutional Neural Network approaches were applied. Support Vector Machines are one type of machine learning technique. The three distinct dataset types employed to evaluate the efficiency of the proposed approach which attains an accuracy of 93.33%—are phrases, messages, and dialogues.