ChatIE

Why we included this project

ChatIE shows how far you can get pulling structured facts out of raw text with zero-shot prompting, no training or labeled data required. It turns ChatGPT into a two-stage, multi-turn question-answering flow that handles entity-relation triple extraction, named entity recognition, and event extraction in both English and Chinese. The code is compact enough to read end to end, which helps researchers and engineers adapt the prompting strategy to their own documents instead of adopting a heavyweight pipeline. Since it runs on the OpenAI API, you will want your own key and some tolerance for variable response latency, but the local setup is straightforward. It is a research tool, so treat it as a proven pattern to build on rather than a turnkey production service.