|
AirDialogue Data v1.2 |
|
|
|
This package contains the train and dev set for the AirDialogue dataset. |
|
|
|
This dataset is meant to be used in combination with the AirDialogue tookit: |
|
https://github.com/google/airdialogue |
|
|
|
News in v1.2: |
|
- Fix alignment problem between kbs and data |
|
- Fix some incorrect name problems |
|
|
|
News in v1.1: |
|
- Standardize data: Combine consecutive user/agent setences into one. |
|
- Fix typos |
|
- Remove emtpy lines in data files |
|
|
|
Reference: |
|
Wei Wei, Quoc V Le, Andrew M Dai and Li-Jia L. AirDialogue: An Environment for Goal-Oriented Dialogue Research. Empirical Methods in Natural Language Processing (EMNLP) 2018 |
|
https://www.aclweb.org/anthology/D18-1419/ |
|
|
|
A reference implementation can be found in the link below: |
|
https://github.com/google/airdialogue_model |
|
|
|
To report an issue related to the dataset, please contact |
|
Wei Wei |
|
wewei@google.com |
|
|