Detecting Human Emotion by Text Classification

Nowadays, it's fairly usual to share moments on social media. By communicating thoughts, ideas, and enjoyable experiences over text, we can express our feelings without needing a lot of words. To investigate people's opinions, sentiments, and emotions, for instance, businesses may target YouTube as an abundant source of data. A greater comprehension of an author's emotions is often possible through emotion analysis. Analyzing expressions as positive, negative, or neutral has been the focus of almost all projects evaluating Telugu social media. We'll categorize the expressions in this essay into groups based on the emotions of happiness, fury, fear, disgust, and melancholy. Different approaches have been used in the case of other languages to automatically recognize textual emotions, however few of them were based on deep learning. Now let's talk about the system we utilized to classify the feelings stated in Telugu YouTube comments. For tasks requiring phrase classification, our model includes an XLMRoBERTa and Multilingual BERT that was specifically trained on our dataset using trained word vectors. We contrasted the outcomes of our method with those of other machine learning techniques. The architecture of our deep learning technique is a word-based, end-to-end network, phrase, and document vectorization procedures. The proposed deep learning strategy was tested using the Telugu YouTube comments dataset, and the results were promising when compared to more traditional machine learning methods