1 repo
Techniques for performing sequential extraction passes to improve data recall.
Distinguishing note: Focuses on the iterative improvement of extraction results.
Explore 1 awesome GitHub repository matching data & databases · Multi-Pass Extraction Pipelines. Refine with filters or upvote what's useful.
Langextract is a framework designed to transform unstructured text into structured, machine-readable data using language model orchestration. It provides a high-performance pipeline that processes large volumes of narrative text by utilizing parallel execution and sequential extraction passes. The library is built to handle complex data extraction tasks, including specialized support for clinical information and medical entity relationship recognition. The project distinguishes itself through a plugin-based architecture that supports both local hardware execution and cloud-hosted model endpoi
Performs multiple independent extraction passes over text to improve recall and capture missed entities.