I have always been fascinated by the invisible web of references between the books I read. A while ago I fine-tuned a Roberta model to detect citations in free text, and it worked, but it was brittle and limited. This time I went all in: a full pipeline that uses LLMs to extract every citation from the raw text of a book, then resolves each one against Goodreads and Wikipedia to find the actual book or author being referenced. The result is a D3.js visualization where you can explore citation networks across centuries, from the Epic of Gilgamesh all the way to David Foster Wallace.
The pipeline has four stages (enrich, extract, clean, resolve) and it processes my entire Calibre library in parallel. It even handles tricky stuff like author name variants (Dostoevsky vs Dostoyevsky vs Dostoevski) and can fall back to the LLM’s own knowledge when local databases come up empty. The frontend arranges authors chronologically so you can literally see the flow of ideas through time. You can click any author to see a radial view of who they cite and who cites them. All the code is on GitHub and you can play with the live version here.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.