Comma: A Collaborative Medical Text Annotation Platform

Abstract

High-quality annotated data plays an important role in data-driven or machine-learning-based medical studies. Large amounts of valuable data existed in free text records such as Electronic Health Records. Manual annotation is paramount for intelligent algorithm applications where plentiful structured data could be extracted. With the exponential growth of medical data, large-scale collaborative annotation is required where different annotation tools should be used. We investigated the existing annotation platform and discovered partial common problems that need to be solved, e.g., lack of integrated annotation workflow, absence of reasonable review and review-assisting process, frequent lexicon upload failure, prolonged auto annotation, unified export standards, etc. In response to the above requirements, we developed Comma, a collaborative medical annotation platform that allows English and Chinese annotation. Comma supports functions involving multiple ways with high success rates to upload documents, self-defined structured entity types, customized auto annotation, annotators capability assessment, various collaborative modes, statistics display, self-determined exportation, etc. It could provide a user-friendly interface, diverse functionalities, and fluent interactions for people demanding annotations. Comma is available online at http://comma.phoc.org.cn.

Ähnliche Arbeiten