commonDost
Bilingual Human-Computer Dialogue Framework
Data
The data/
folder consists of three metafiles and a folder corresponding for each batch of data collection.
The metafiles contain information about the chat and corresponding survey files. They have the following format
chat_json survey_json
<chat_json_filename> <survey_json_filename>
The data was collected via two sources: mechanical turk (mturk) and local community (community). The labels of the folders are in the format {data-source}_data{#id}_{cs_strategy}_{gender}_{date}
.
Each folder contains the following files:
chat.html
: Visualization of all the chats and surveysclean_transcript.json
: The raw data in a json format. This is only for filtered chats.clean_survey.json
: The surveys for the filtered chats in a json format.
Citations and Contributions
If you use the data or code, please use the following citation:
Parekh, T., Ahn, E., Tsvetkov, Y. and Black, A., 2020. "Understanding Linguistic Accommodation in Code-Switched Human-Machine Dialogues". In Proceedings of the 24th Conference on Computational Natural Language Learning.
@inproceedings{parekh-etal-2020-understanding,
title = "Understanding Linguistic Accommodation in Code-Switched Human-Machine Dialogues",
author = "Parekh, Tanmay and Ahn, Emily and Tsvetkov, Yulia and Black, Alan W",
booktitle = "Proceedings of the 24th Conference on Computational Natural Language Learning",
year = "2020",
publisher = "Association for Computational Linguistics",
url = "https://www.aclweb.org/anthology/2020.conll-1.46",
}
For any queries, please contact/email: tparekh[at]andrew.cmu.edu