Key points are not available for this paper at this time.
The growing demand for chatbots in online customer service has led to an increased need for effective evaluation methods of service encounter satisfaction (CSES). Conventional post-interaction surveys, often incomplete and biased, pose challenges in accurately measuring CSES. This study explores the Incribo synthetic sample chatbot dataset, obtained from Kaggle, comprising more than 10,000 distinct messages. The dataset encompasses diverse conversational interactions, with each message annotated with User ID, User Utterance, Bot Response, User Feedback, Conversation Outcome, User Profile, Platform, Language, User Emotion/Sentiment, Location, and User Segment. The research aims to explore the sentiments methods, assess the influence of the mentioned variables, and evaluate the correlations between sentiment and conversational outcomes. Employing a hybrid approach, the study reveals widespread user participation, varied linguistic preferences, and equitable conversation outcomes. The correlation analysis underscores a notable negative association between neutral user sentiments and feedback, providing valuable insights into interactions within online customer service.
Rawat et al. (Mon,) studied this question.