Papers by Suhyeon Jo
Reward-based Input Construction for Cross-document Relation Extraction (2024.acl-long)
Copied to clipboard
| Challenge: | Existing methods for relation extraction are limited to a single sentence or document, but cross-document RE has emerged to address relations across multiple documents. |
| Approach: | They propose a sentence selector that extracts sentences based on relational evidence and rewards it with RE prediction scores. |
| Outcome: | The proposed method outperforms heuristic methods for relation extraction in multiple document scenarios. |