This study investigates user interaction patterns in e-commerce chatbot sessions and evaluates the impact of AI-based recommendation simulations. Using a dataset of 10 anonymized sessions, the research analyzes session length, query types, feedback, and click behavior to identify factors influencing recommendation acceptance. Results show improved prediction accuracy for multi-intent sessions and highlight the importance of session-aware modeling and feedback integration in chatbot design.
The study confirms that analyzing user interaction patterns can enhance chatbot recommendation relevance. AI-based simulations showed measurable improvements, especially in complex sessions. Future work may expand datasets, incorporate multimodal inputs, and deploy adaptive recommendation strategies in live environments.
References
Includes references from CIKM, EMNLP, Journal of Retailing and Consumer Services, BMC Psychology, and Wikipedia covering session-aware recommendation systems and chatbot adoption in e-commerce.
