Jointly Learning Clinical Entities and Relations with Contextual Language Models and Explicit Context
This work addresses the challenge of extracting structured information from clinical documents, which is incremental as it builds on existing methods by emphasizing context.
The paper tackled the problem of jointly learning named entity recognition and relation extraction in clinical text by integrating explicit contextual information into a multi-task learning framework, achieving near state-of-the-art performance with a 49.07 F1 score on end-to-end relation extraction.
We hypothesize that explicit integration of contextual information into an Multi-task Learning framework would emphasize the significance of context for boosting performance in jointly learning Named Entity Recognition (NER) and Relation Extraction (RE). Our work proves this hypothesis by segmenting entities from their surrounding context and by building contextual representations using each independent segment. This relation representation allows for a joint NER/RE system that achieves near state-of-the-art (SOTA) performance on both NER and RE tasks while beating the SOTA RE system at end-to-end NER & RE with a 49.07 F1.