2019/12/02 by Trapit Bansal, Bansal, Trapit, Pat Verga +5
Biochemistry, Genetics and Molecular Biology · Computer Science · #Biomedical Text Mining and Ontologies #Computation and Language (cs.CL) #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Machine Learning (cs.LG) #Natural Language Processing Techniques #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1912.01070
openalex publication_date 2019/12/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Understanding the meaning of text often involves reasoning about entities and\ntheir relationships. This requires identifying textual mentions of entities,\nlinking them to a canonical concept, and discerning their relationships. These\ntasks are nearly always viewed as separate components within a pipeline, each\nrequiring a distinct model and training data. While relation extraction can\noften be trained with readily available weak or distant supervision, entity\nlinkers typically require expensive mention-level supervision -- which is not\navailable in many domains. Instead, we propose a model which is trained to\nsimultaneously produce entity linking and relation decisions while requiring no\nmention-level annotations. This approach avoids cascading errors that arise\nfrom pipelined methods and more accurately predicts entity relationships from\ntext. We show that our model outperforms a state-of-the art entity linking and\nrelation extraction pipeline on two biomedical datasets and can drastically\nimprove the overall recall of the system.\n