Summary of Graphere: Jointly Multiple Event-event Relation Extraction Via Graph-enhanced Event Embeddings, by Haochen Li et al.
GraphERE: Jointly Multiple Event-Event Relation Extraction via Graph-Enhanced Event Embeddings
by Haochen Li, Di Geng
First submitted to arxiv on: 19 Mar 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 GraphERE framework addresses two main challenges in Event-Event Relation Extraction (ERE): using only event trigger embeddings, ignoring event arguments and their structure, and neglecting interconnections between relations. To overcome these limitations, GraphERE enriches event embeddings with argument and structure features from static AMR graphs and IE graphs. The framework jointly extracts multiple event relations using Node Transformers and constructs task-specific dynamic event graphs for each type of relation. A multi-task learning strategy is employed to train the entire framework. Experimental results on the MAVEN-ERE dataset demonstrate significant improvements over existing methods, highlighting the effectiveness of graph-enhanced event embeddings and joint extraction. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to understand connections between events in documents has been proposed. This method, called GraphERE, fixes two problems with current approaches: only considering the start of an event and not accounting for how different events relate to each other. To solve these issues, GraphERE combines information from event triggers, arguments like time and place, and the structure within the event. It also considers how different types of connections between events work together. This new approach has been tested on a dataset and shown to be much better than existing methods. |
Keywords
» Artificial intelligence » Multi task
