COMPUTATIONAL RECOGNITION OF EMOTIONS: SENTIMENT ANALYSIS IN HUMAN COMPUTER INTERACTION

Abstract

In today’s digitally-driven world, Human-Computer Interaction (HCI) has evolved far beyond simple input-output mechanisms. As technology becomes more integrated into daily life, there is a growing need for machines to understand and respond to human emotions in a meaningful way. This thesis focuses on the computational recognition of emotions through sentiment analysis—a subfield of Natural Language Processing (NLP)—and explores how it enhances the quality and effectiveness of human-computer interaction. The thesis includes a preface on the impact of emotions on communication and judgment, demonstrating the importance of emotional intelligence in a wide range of systems such as e-learning platforms, virtual assistants, and customer care chatbots. It then covers the fundamental methods of sentiment analysis that would be necessary to classify user input as either good, negative, or neutral, while exploring rule-based methods, machine learning methods, and deep learning methods. One other purpose of these methods is to infer deeper emotional states like joy, anger, fear, or surprise. A large component of the thesis is assessing how well robots are able to identify the emotional tone of a user's text-based Realtime interactions, as well as how the incorporation of text, visual, and audio input-into multimodal sentiment analysis systems-can improve their ability to assess emotional tone. A significant portion of the thesis is devoted to examining current challenges in the field, including sarcasm detection, ambiguity in language, and cultural variances in emotional display. It also highlights directions for the future of research with a focus on context-aware systems that perform multimodal sentiment analysis, and posited potential solutions. Overall, this study shows that sentiment analysis is essential to overcoming the emotional divide between people and computers and opening the door to more intelligent, sympathetic, and realistic human-computer interactions.