GEN AI INTELLIGENT TRAVEL CHATBOT

Abstract

This project presents the development of a Gen AI-based Intelligent Travel Chatbot designed to enhance the travel, tourism, and hospitality industry by providing smart, real-time, and personalized travel assistance through natural language conversations. The system utilizes advanced Generative Artificial Intelligence (Gen AI) technologies to understand user queries and deliver accurate, context-aware, and human-like responses related to travel planning and assistance. The proposed chatbot integrates a Vision-Language Model (VLM) pipeline combined with a fine-tuned Large Language Model (LLM) capable of generating rich travel recommendations based on user preferences and travel requirements. The system employs Retrieval-Augmented Generation (RAG) with a FAISS vector database containing more than 250,000 travel-domain embeddings, enabling efficient retrieval of up-to-date information regarding flights, hotels, tourist attractions, restaurants, transportation, visa requirements, weather conditions, and travel schedules. The core conversational model is based on the LLaMA-2-7B architecture fine-tuned using LoRA (Low-Rank Adaptation) on over 180,000 curated travel question-answer pairs. The trained model achieves high performance with a CIDEr score of 1.42 and a BLEU-4 score of 41.2%, demonstrating significant improvements in response quality, contextual understanding, and conversational accuracy compared to baseline chatbot models.