A Streamlit app to summarize arXiv.org academic research

Hi all, long time listener / first time caller. Please let me know if this topic belongs somewhere else or requires more detail.

App here: https://arxiv-explorer.streamlit.app

I am introducing the arXiv Explorer, a streamlit app that allows you to catch up on recent arXiv.org activity in any category. We use a combination of NLP tokenization techniques, LDA analyses for topic modeling, and a HuggingFace LLM for single-article summaries. Use cases may include:

  • A researcher who wants to understand the latest trends in their field without reading 500+ academic papers every week
  • An academic who learns about recent developments in adjacent fields
  • An undergraduate looking for research topics in a desired field
  • A lay-person seeking an introduction of the vocabulary used in a specific academic domain
  • Any of the above who want to summarize a specific technical article into 5-15 “conversational” chunks

Please check out the application and GitHub repository and let me know if you have questions or improvements! This is my first time really diving into statefulness, LLMs, and external visualization libraries (pyLDAvis) for Streamlit apps, and I’m very proud of the results.

For avoidance of doubt, I am in no way affiliated with arXiv.org, and I want to thank arXiv for providing use of its open access interoperability.

Thanks everyone!

3 Likes