Turk dialogues - Dialogues invented by Amazon Mechanical Turk workers
---------------------------------------------------------------------
https://aactext.org/turk/

This dataset is a collection of two-sided dialogues invented by Amazon Mechanical Turk workers.
Different workers invented each turn in the dialogues.

Further details can be found in this ASSETS '17 poster:
@inproceedings{vertanen_aacdialogue,
  author       = {Keith Vertanen},
  title        = {Towards Improving Predictive AAC using Crowdsourced Dialogues and Partner Context},
  booktitle    = {ASSETS '17: Proceedings of the ACM SIGACCESS Conference on Computers and Accessibility (poster)},
  year         = {2017},
  pages        = {347--348},
}

We created a filtered version of the dialogues removing potentially offensive content as part of this paper:
@inproceedings{adhikary_speech,
  author       = {Jiban Adhikary and Robbie Watling and Crystal Fletcher and Alex Stanage and Keith Vertanen},
  title        = {Investigating Speech Recognition for Improving Predictive AAC},
  booktitle    = {SLPAT '19: Proceedings of the Workshop on Speech and Language Processing for Assistive Technologies},
  location     = {Minneapolis, MN},
  month        = {June},
  year         = {2019},
  pages        = {37--43},
}

The Turk dialogue dataset is licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License, http://creativecommons.org/licenses/by-nd/3.0/

Details of included files
-------------------------
readme.txt                   - This file

turk-dialogues.txt           - Original dialogue set of 1,420 dialogues. Each line has 8 columns:
                                * set - Specifies if dialogue in the train, dev, or test set in ASSETS '17 poster
                                * id - Unique ID for this dialogue 
                                * turn1 - First worker's invented dialogue turn
                                * turn2 - Second worker's response to previous turn
                                ...
                                * turn6 - Sixth worker's response to previous five turns

turk-dialogues-filtered.txt  - Filtered version of 1,155 dialogues. Each line has 7 columns:
                                * id - Unique ID for this dialogue, corresponds to ID in original dialogue set
                                * turn1 - First worker's invented dialogue turn
                                * turn2 - Second worker's response to previous turn
                                ...
                                * turn6 - Sixth worker's response to previous five turns

Have fun!
Keith Vertanen

Revision history
----------------
Nov 18, 2022    Created a packaged version of the original and filtered Turk dialogues.