Papers by Brendan O'Connor
Coordinates from Context: Using LLMs to Ground Complex Location References (2026.eacl-long)
Copied to clipboard
| Challenge: | Existing geocoding tools can only link locations already in a geographic database, which often do not include compositional locations. |
| Approach: | They propose a geocoding strategy that leverages LLMs' geospatial knowledge versus reasoning skills to improve performance for the task. |
| Outcome: | The proposed model improves performance and is comparable to larger models. |
Contextual morphologically-guided tokenization for Latin encoder models (2026.eacl-long)
Copied to clipboard
| Challenge: | Existing tokenization methods focus on information-theoretical goals like high compression and low fertility rather than linguistic goals like morphological alignment. |
| Approach: | They propose to incorporate morphological knowledge into tokenization to improve both morphology and downstream performance. |
| Outcome: | The proposed tokenization improves overall performance on four downstream tasks. |