Research map / Image generation / Text-to-image diffusion
Scale Up Event Extraction Learning via Automatic Training Data Generation
Ying Zeng, Yansong Feng, Rong Ma, Zheng Wang, Rui Yan, Chongde Shi, Dongyan Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add a CRF layer to BLSTM (ablation) | BLSTM event extractor · FBWiki automatic annotations | event type F-measure: 74.1 -> 75.7 | Improved |
| Add a CRF layer to BLSTM (ablation) | BLSTM event extractor · FBWiki automatic annotations | key argument detection F-measure: 53.5 -> 55.8 | Improved |
| Add a CRF layer to BLSTM (ablation) | BLSTM event extractor · FBWiki automatic annotations | all argument detection F-measure: 33.7 -> 36.5 | Improved |
| Add ILP post-inference to BLSTM-CRF (ablation) | BLSTM-CRF · FBWiki automatic annotations | event type F-measure: 75.7 -> 78.9 | Improved |
| Add ILP post-inference to BLSTM-CRF (ablation) | BLSTM-CRF · FBWiki automatic annotations | key argument detection F-measure: 55.8 -> 57.3 | Improved |
| Add ILP post-inference to BLSTM-CRF (ablation) | BLSTM-CRF · FBWiki automatic annotations | all argument detection F-measure: 36.5 -> 37.7 | Improved |
| Allow ILP to output multiple optimal label sequences (ablation) | BLSTM-CRF-ILP · FBWiki automatic annotations | event type recall: 74.1 -> 79.9 | Improved |
| Allow ILP to output multiple optimal label sequences (ablation) | BLSTM-CRF-ILP · FBWiki automatic annotations | event type precision: 84.4 -> 85.3 | Hurt |
| Select top-half arguments by importance instead of requiring all arguments (ablation) | Distant-supervision data generation · FBWiki data generation | positive percentage: 98% -> 22% | Hurt |
| Filter examples whose key-argument dependency distances exceed 2 (ablation) | IMP data generation · FBWiki data generation | positive percentage: 22% -> 37% | Improved |
| Add a time-related argument to the selected important arguments (ablation) | IMP data generation · FBWiki data generation | positive percentage: 22% -> 83% | Improved |
| Add a time-related argument and dependency-distance filtering (ablation) | IMP data generation · FBWiki data generation | positive percentage: 22% -> 91% | Improved |
Closest papers by meaning
- Is a Large Language Model a Good Annotator for Event Extraction? (AAAI 2024 · similarity 0.89)
- STAR: Boosting Low-Resource Information Extraction by Structure-to-Text Data Generation with Large Language Models (AAAI 2024 · similarity 0.87)
- A Joint Framework with Heterogeneous-Relation-Aware Graph and Multi-Channel Label Enhancing Strategy for Event Causality Extraction (AAAI 2024 · similarity 0.85)
- Logic Induced High-Order Reasoning Network for Event-Event Relation Extraction (AAAI 2025 · similarity 0.85)
- Beyond Grounding: Extracting Fine-Grained Event Hierarchies across Modalities (AAAI 2024 · similarity 0.85)
- EventSum: A Large-Scale Event-Centric Summarization Dataset for Chinese Multi-News Documents (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
