Revo AI: A Lightweight Website Question Answering System for Educational Information Retrieval

This paper presented REVO AI, a lightweight and API-free question answering system for educational websites. The system combines website crawling, HTML content extraction, TF-IDF vectorization, cosine similarity, and extractive question answering into a single deployable application. Experimental testing showed that the chatbot can retrieve useful information with good accuracy and low response time. The project demonstrates that classical information retrieval methods can still provide effective solutions for domain-specific conversational AI systems without requiring expensive cloud-based AI services. Future enhancements may include dense embedding models, ChromaDB integration, hybrid BM25 retrieval [5], OCR support, PDF question answering, voice interaction, generative AI integration, and user authentication systems. These improvements can significantly increase retrieval accuracy and usability. REVO AI can serve as a practical educational assistant for universities, institutions, and student projects focused on information retrieval and intelligent web applications.  

[1]       E. Brill, S. Dumais, and M. Banko, "An Analysis of the AskMSR Question-Answering System," in Proc. EMNLP, 2002. [2]       P. Rajpurkar, J. Zhang, K. Lopyrev, and P. Liang, "SQuAD: 100,000+ Questions for Machine Comprehension of Text," in Proc. EMNLP, 2016. [3]       P. Lewis, E. Perez, A. Piktus, et al., "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks," in Proc. NeurIPS, 2020. [4]       J. Ramos, "Using TF-IDF to Determine Word Relevance in Document Queries," 2003. [5]       S. Robertson and H. Zaragoza, "The Probabilistic Relevance Framework: BM25 and Beyond," Information Retrieval Journal, 2009. [6]       L. Cui, S. Huang, F. Wei, et al., "SuperAgent: A Customer Service Chatbot for E-commerce Websites," in Proc. ACL, 2017. [7]       W. Zhao, M. Peyrard, F. Liu, et al., "MoverScore: Text Generation Evaluation using Contextualized Embeddings," in Proc. EMNLP, 2019. [8]       T. Mikolov, I. Sutskever, K. Chen, G. Corrado, and J. Dean, "Distributed Representations of Words and Phrases and their Compositionality," in Proc. NeurIPS, 2013.   [9]      A. Vaswani, N. Shazeer, N. Parmar, et al., "Attention Is All You Need," in Proc. NeurIPS, 2017.   [10]     J. Devlin, M. Chang, K. Lee, and K. Toutanova, "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding," in Proc. NAACL, 2019.