Few-Shot Document-Level Relation Extraction
This work addresses the need for more realistic benchmarks in relation extraction for researchers, though it is incremental as it adapts existing methods to new data.
The authors tackled the problem of few-shot document-level relation extraction by creating a new benchmark (FREDo) based on existing datasets, and found it to be a challenging setting with novel characteristics like sampling NOTA instances from support sets.
We present FREDo, a few-shot document-level relation extraction (FSDLRE) benchmark. As opposed to existing benchmarks which are built on sentence-level relation extraction corpora, we argue that document-level corpora provide more realism, particularly regarding none-of-the-above (NOTA) distributions. Therefore, we propose a set of FSDLRE tasks and construct a benchmark based on two existing supervised learning data sets, DocRED and sciERC. We adapt the state-of-the-art sentence-level method MNAV to the document-level and develop it further for improved domain adaptation. We find FSDLRE to be a challenging setting with interesting new characteristics such as the ability to sample NOTA instances from the support set. The data, code, and trained models are available online (https://github.com/nicpopovic/FREDo).