Denoising Relation Extraction from Document-level Distant Supervision
This work addresses noise in document-level relation extraction for natural language processing, but it is incremental as it builds on existing distant supervision methods.
The paper tackles the challenge of applying distant supervision to document-level relation extraction by proposing a pre-trained model that denoises the data through multiple pre-training tasks, achieving promising results on a large-scale benchmark.
Distant supervision (DS) has been widely used to generate auto-labeled data for sentence-level relation extraction (RE), which improves RE performance. However, the existing success of DS cannot be directly transferred to the more challenging document-level relation extraction (DocRE), since the inherent noise in DS may be even multiplied in document level and significantly harm the performance of RE. To address this challenge, we propose a novel pre-trained model for DocRE, which denoises the document-level DS data via multiple pre-training tasks. Experimental results on the large-scale DocRE benchmark show that our model can capture useful information from noisy DS data and achieve promising results.