<![CDATA[Loneliness has increasingly emerged as a global mental health concern, particularly among vulnerable populations such as the elderly, students in remote learning environments, and individuals experiencing social isolation in urban societies. Advances in affective and positive computing offer promising solutions for addressing these challenges by creating empathetic digital companions capable of responding to human emotions in real time. This study aims to evaluate the effectiveness of sentiment-aware chatbots as digital companions in reducing loneliness and enhancing emotional well-being through positive computing principles. A quantitative experimental approach was employed, integrating sentiment analysis algorithms with Natural Language Processing (NLP) to detect emotional cues from user input and generate empathetic responses. The chatbot system was tested with 150 participants over a six-week period using standardized psychometric instruments, including the UCLA Loneliness Scale and the WHO-5 Well-Being Index. Statistical analysis using paired-sample t-tests, ANOVA, and Structural Equation Modeling (SEM) revealed significant improvements in loneliness reduction and psychological well-being among participants interacting with the sentiment-aware chatbot. Furthermore, perceived empathy and user satisfaction were found to mediate these effects, highlighting the emotional quality of human AI interaction as a crucial determinant of positive outcomes. Findings provide empirical evidence that sentiment-aware chatbots can function as effective digital companions, reducing loneliness and fostering psychological resilience. By integrating affective and positive computing principles, this study contributes to the advancement of compassionate AI systems designed to promote human well-being and support broader societal goals.]]>
