ArgSciChat is an argumentative dialogue dataset. It consists of 498 messages collected from 41 dialogues on 20 scientific papers. It can be used to evaluate conversational agents and further encourage research on argumentative scientific agents.
Variants: ArgSciChat
This dataset is used in 2 benchmarks:
| Task | Model | Paper | Date |
|---|---|---|---|
| Fact Selection | TF-IDF | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Fact Selection | S-BERT | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Fact Selection | LED(Q,P) | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Fact Selection | LED(Q,P,H) | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Response Generation | LED(Q,F) | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Response Generation | LED(Q,P,H) | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
| Response Generation | LED(Q,P) | ArgSciChat: A Dataset for Argumentative … | 2022-02-14 |
Recent papers with results on this dataset: