Building a WhatsApp Chatbot with Node.js: File Processing and AI Response Generation

This paper presents the development and implementation of an AI-powered chatbot system tailored for educational institutions. The system utilizes a diverse dataset of conversations specific to the educational domain, enabling accurate understanding and response generation for frequently asked questions, inquiries, and specific queries related to the institute’s programs, admissions, courses, and facilities. The dataset is pre-processed to ensure clean and formatted text, incorporating techniques such as tokenization, stemming, and stop-word removal. The underlying model of the chatbot is based on the GPT-2 transformer model, known for its exceptional language generation capabilities and contextual understanding. Additionally, the Logic Learning Machine (LLM) algorithm is employed to generate intelligible rules, enhancing the chatbot’s decision-making process. The implementation involves the integration of various libraries and technologies, including ‘whatsapp-web.js’ for establishing a WhatsApp session, ‘openai’ for AI-driven completions, and ‘twilio’ for response delivery via messaging service. The system demonstrates its effectiveness in accurately understanding and responding to user queries, contributing to improved communication and information dissemination within the educational domain. Overall, the chatbot system presents a valuable tool for educational institutions, enhancing user experience, and streamlining communication processes.