Sentiment Analysis using Machine learning algorithm using speech signal

• Published in: 2022 Second International Conference on Advanced Technologies in Intelligent Control, Environment, Computing & Communication Engineering (ICATIECE) pp. 1 - 5
• Main Authors: Deepa, R., Yuvaraj, S., Preethi, T., Patil, Pradnya, N, Ilayaraja
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16.12.2022
• Subjects: Analytical models; Blogs; classification; machine learning; Machine learning algorithms; Semantics; Sentiment analysis; Social networking (online); Speech signal; Waste materials
• DOI: 10.1109/ICATIECE56365.2022.10046898

With the growth of online social networks, people now have a new forum for sharing their thoughts and perspectives with family, friends, and other users on various issues and topics. Users can express their thoughts and emotions in various forms like images, text, memes, postings, and audio/video messages, among which text is the most popular way to communicate on social media. In this study, we collected, tested, and analyzed the data from the most popular social media, Twitter. The primary goal of this work is to identify and assess the emotions and thoughts expressed by users in their text-based Twitter tweets. The Bag-Of-Words model, while the most popular technique for sentiment analysis, has two critical disadvantages, such as applying a manual lexicon for establishing word analysis. The second drawback is it analyses sentiments with high error because it refuses to acknowledge the language grammar impacts of the words and ignores semantics. In this research, we provide a unique approach for assessing online sentiments in a single domain and a solution for addressing crucial challenges in sentiment analysis that improves sentiment analysis accuracy. Using the improved bag-of-words model, word weight is taken to determine the polarity and score instead of term frequency. The proposed method automatically categorizes the keywords and characteristics related to scientific subject areas. This work provides an effective solution for typical sentiment analysis issues. The proposed model is enhanced to achieve maximum sentiment analysis precision.