|
Towards a Reappraisal of the Linguistic Perspective in the LLM Era| title | Towards a Reappraisal of the Linguistic Perspective in the LLM Era |
|---|
| start_date | 2024/07/09 |
|---|
| schedule | 09h30 |
|---|
| online | no |
|---|
| location_info | Grand Amphi |
|---|
| summary | Currently, Large Language Models (LLMs) have quickly become the norm and have shown impressive results for language applications in different research domains. Various discourse-related tasks usually requiring in-depth knowledge of the world, however, seem to remain largely understudied. Focusing on tasks such as event coreference resolution, emotion detection and irony detection, I would like to discuss how the rich history of Computational Linguistics can help us in identifying knowledge gaps in current methodologies and in attempting to address these. I will touch upon the issue of dataset selection and labeling and a heavy reliance on benchmark datasets, and discuss opportunities to unify both expert knowledge and modern-day learning frameworks and architectures. |
|---|
| responsibles | NC |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2024/06/27 13:50 UTC |
| |
|