Multi-Class Sentiment Analysis of Hindi Textual Data
Madhurim Gupta, Aryaman Singh Kushwaha, Anushree Sinha, Prakhar Singh, Rajesh Kumar · 2023
Sentiment analysis plays a crucial role in natural language processing, especially when it comes to grasping and examining the emotional aspects within textual content. Categorizing text into multiple sentiment categories is a complex task due to the intricate and subjective nature of human emotions and expressions. In this paper, we present an extensive study on multi-class sentiment analysis of Hindi textual data, where the dataset is categorized into five different sentiments. To attain high accuracy in sentiment analysis, three different methods, namely Long Short Term Memory (LSTM), Random Forest (RF) and Support Vector Machine (SVM) have been employed. This is accomplished by fine-tuning hyperparameters and preprocessing the data, which significantly improves the accuracy of these models. The models have been trained and evaluated using the Hindi text dataset to determine the sentiment conveyed in the text. The experimental findings reveal that the SVM model outperforms the others in terms of accuracy. This study underscores the effectiveness of optimizing hyperparameters and improving data preprocessing to achieve superior accuracy. The insights gained from this research can be valuable for the development of sentiment analysis systems for Hindi text data, with applications ranging from social media analysis and customer feedback evaluation to market research.