An end-to-end chatbot that scrapes recent arXiv papers, preprocesses and embeds them using HuggingFace models, and allows question-answering via LangChain. Built for an Information Retrieval course project.
- Scrape papers from arXiv by topic
- Preprocess abstracts + metadata
- Embed using
sentence-transformers - Query papers via chatbot (Streamlit or CLI)
- 100% local and free — no OpenAI key required
- Clone repo & enter directory
git clone https://github.com/vedsub/ir_project.git cd ir_project