Summary of Prompting Implicit Discourse Relation Annotation, by Frances Yung et al.
Prompting Implicit Discourse Relation Annotation
by Frances Yung, Mansoor Ahmad, Merel Scholman, Vera Demberg
First submitted to arxiv on: 7 Feb 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 This study examines the performance of pre-trained large language models like ChatGPT in implicit discourse relation classification tasks. While these models excel in various reasoning tasks without supervision, their performance in this specific task falls short of state-of-the-art supervised approaches. To improve ChatGPT’s recognition of discourse relations, several prompting techniques were tested, including breaking down the classification task into smaller subtasks. Despite these efforts, experiment results showed that inference accuracy remained largely unchanged. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study looks at how well large language models like ChatGPT can understand relationships between ideas in text. These models are really good at doing certain tasks without being taught, but they’re not great at this specific task. To help them do better, researchers tried different ways of asking the model questions. Unfortunately, even with these new techniques, the model still didn’t do very well. |
Keywords
» Artificial intelligence » Classification » Discourse » Inference » Prompting » Supervised
