Reasoning with Latent Structure Refinement for Document-Level Relation Extraction
This addresses the challenge of integrating information across sentences for relation extraction in documents, which is incremental but improves performance on specific benchmarks.
The paper tackles the problem of document-level relation extraction by proposing a model that automatically induces a latent document-level graph and uses a refinement strategy for multi-hop reasoning, achieving an F1 score of 59.05 on DocRED and setting new state-of-the-art results on CDR and GDA datasets.
Document-level relation extraction requires integrating information within and across multiple sentences of a document and capturing complex interactions between inter-sentence entities. However, effective aggregation of relevant information in the document remains a challenging research question. Existing approaches construct static document-level graphs based on syntactic trees, co-references or heuristics from the unstructured text to model the dependencies. Unlike previous methods that may not be able to capture rich non-local interactions for inference, we propose a novel model that empowers the relational reasoning across sentences by automatically inducing the latent document-level graph. We further develop a refinement strategy, which enables the model to incrementally aggregate relevant information for multi-hop reasoning. Specifically, our model achieves an F1 score of 59.05 on a large-scale document-level dataset (DocRED), significantly improving over the previous results, and also yields new state-of-the-art results on the CDR and GDA dataset. Furthermore, extensive analyses show that the model is able to discover more accurate inter-sentence relations.