Summary of Gega: Graph Convolutional Networks and Evidence Retrieval Guided Attention For Enhanced Document-level Relation Extraction, by Yanxu Mao et al.
GEGA: Graph Convolutional Networks and Evidence Retrieval Guided Attention for Enhanced Document-level Relation Extraction
by Yanxu Mao, Xiaohui Chen, Peipei Liu, Tiehan Cui, Zuhui Yue, Zheng Li
First submitted to arxiv on: 31 Jul 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed novel model, GEGA, aims to overcome the challenges in document-level relation extraction (DocRE) by leveraging graph neural networks and multi-scale representation aggregation. Currently, logical rules within evidence sentences are utilized to enhance DocRE performance, but this approach is limited when there is no provided evidence sentence. GEGA constructs multiple weight matrices to guide attention allocation to evidence sentences, improving the relevance between evidence and entity pairs. The model also employs evidence retrieval to integrate efficient information for both fully supervised and weakly supervised training processes. Experimental results on three benchmark datasets (DocRED, Re-DocRED, and Revisit-DocRED) demonstrate comprehensive improvements over the existing state-of-the-art model. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GEGA is a new way to understand relationships between things in documents. Right now, researchers are using rules within sentences to get better at this task, but it’s not very good when there are no sentence-level clues. The GEGA model uses special computer techniques called graph neural networks and multi-scale representation aggregation to make sense of long pieces of text. It also retrieves important information from the documents to help train itself. When tested on several datasets, GEGA outperformed other models that tried this task. |
Keywords
» Artificial intelligence » Attention » Supervised
