Document-Level N-ary Relation Extraction with Multiscale Representation Learning

Proceedings of NAACL-HLT 2019, pages 3693–3704 Minneapolis, Minnesota, June 2 - June 7, 2019. c⃝2019 Association for Computational Linguistics 3693 Document-Level N-ary Relation Extraction with Multiscale Representation Learning Robin Jia1∗ Cliff Wong2 Hoifung Poon2 1 Stanford University, Stanford, California, USA 2 Microsoft Research, Redmond, Washington, USA robinjia@cs.stanford.edu, {cliff.wong,hoifung}@microsoft.com Abstract Most information extraction methods focus on binary relations expressed within single sen- tences. In high-value domains, however, n-ary relations are of great demand (e.g., drug-gene- mutation interactions in precision oncology). Such relations often involve entity mentions that are far apart in the document, yet exist- ing work on cross-sentence relation extraction is generally confined to small text spans (e.g., three consecutive sentences), which severely limits recall. In this paper, we propose a novel multiscale neural architecture for document- level n-ary relation extraction. Our system combines representations learned over various text spans throughout the document and across the subrelation hierarchy. Widening the sys- tem’s purview to the entire document maxi- mizes potential recall. Moreover, by integrat- ing weak signals across the document, multi- scale modeling increases precision, even in the presence of noisy labels from distant super- vision. Experiments on biomedical machine reading show that our approach substantially outperforms previous n-ary relation extraction methods. 1 Introduction Knowledge acquisition is a perennial challenge in AI. In high-value domains, it has acquired new urgency in recent years due to the advent of big data. For example, the dramatic drop in genome sequencing cost has created unprecedented oppor- tunities for tailoring cancer treatment to a tumor’s genetic composition (Bahcall, 2015). Despite this potential, operationalizing personalized medicine is difficult, in part because it requires painstaking curation of precision oncology knowledge from biomedical literature. With tens of millions of pa- pers on PubMed, and thousands more added every ∗Work done as an intern at Microsoft Research. “We next expressed ALK F1174L, ALK F1174L/L1198P, ALK F1174L/G1123S, and ALK F1174L/G1123D in the original SH-SY5Y cell line.” (. . . 15 sentences spanning 3 paragraphs . . . ) “The 2 mutations that were only found in the neurob- lastoma resistance screen (G1123S/D) are located in the glycine-rich loop, which is known to be crucial for ATP and ligand binding and are the first mutations de- scribed that induce resistance to TAE684, but not to PF02341066.” Figure 1: Two examples of drug-gene-mutation re- lations from a biomedical journal paper. The rela- tions are expressed across multiple paragraphs, requir- ing document-level extraction. day,1 we are sorely in need of automated methods to accelerate manual curation. Prior work in machine reading has made great strides in sentence-level binary relation extraction. However, generalizing extraction to n-ary rela- tions poses new challenges. Higher-order relations often involve entity mentions that are far away in the document. Recent work on n-ary relation ex- traction has begun to explore cross-sentence ex- traction (Peng et al., 2017; Wang and Poon, 2018), but the scope is still confined to short text spans (e.g., three consecutive sentences), even though a document may contain hundreds of sentences and tens of thousands …

Content truncated to 500 words