MarbleBot: A Conversational Recommender and Assistance Chatbot for Marble Selection Based on Dialogflow

The marble industry offers a vast selection of styles, presenting a challenge for customers seeking the ideal marble for their homes or businesses. To address this, chatbots have emerged as a convenient means for customers to receive instant assistance and personalized recommendations. This paper introduces a marble recommender and assistance chatbot designed for a Marble and Granite selling company. The recommendation process integrates both content-based filtering and natural language processing (NLP) techniques. Utilizing Dialogflow, we developed and implemented the chatbot on the company's website. Through dialogue-based interactions, a conversational recommender system was established. The system adeptly comprehends various natural language queries and delivers relevant recommendations based on individual customer preferences. To evaluate the system's performance, we measured the accuracy of recommendations and user satisfaction. Our user study findings reveal that the chatbot provided users with a satisfying experience and achieved an impressive accuracy rate of 85%. The chatbot, named MarbleBot, proves to be an indispensable tool in helping users discover the perfect marble that suits their needs. The paper proceeds in a chronological order, beginning with the introduction, followed by a comprehensive review of relevant literature. The research methodology section outlines the tools employed, including Dialogflow methodology, Flask-API, MySQL database, system design, and pre-processing steps. Subsequently, we present the results, encompassing evaluation tools and an analysis of testing outcomes. Finally, the paper concludes with a summary of our findings and potential future directions.