Table 4.
Datasets for Goal Oriented CAs | ||||
---|---|---|---|---|
Schema Guided | dialogue simulator+ | multi-domain, | 20 k | intent prediction, |
Dialogue [232] | paid | task-oriented | conversations | lang. generation, |
crowd-workers | human-agent convev. | dialogue tracking | ||
MultiWOZ | turkers working | human-human | 10 k dialogues | Task-oriented |
[233] | conversations | dialogue modelling | ||
Taskmaster-1 | crowd workers | spoken & written | 5507 spoken & | dialogue systems |
[234] | users and | technical | 7708 written | research, dev. |
center operators | dialogs | dialogs | and design | |
MultiDoGo | crowd workers | human to human, | 1 K dialogues | virtual assistants |
[235] | paired with | services dialogues | across 6 domains, | development |
trained annotators | ||||
Datasts for Supporting CAs | ||||
COVID-19 dialogue | online healthcare | conversations between | 603 Eng. + | medical dialogue |
dataset [176] | platform | doctors and | 1088 Chinese | system |
patients | consultations | systems | ||
MedDialog | medical dialogue | doctors–patients | 1.1 M Chinese + | medical dialogue |
[236] | platform | conversations | 0.3 M English | systems |
dialogues | ||||
SEMAINE | human–human | emotionally coloured | 25 recordings, | eliciting non-verbal |
[239] | conversation | conversations video | 0 min | signals in |
experiment | recordings | long | human-computer | |
interactions | ||||
EmpatheticDialogues | 810 crowd workers | conversations | 25 k conversations | recognizing |
[238] | select an emotion | grounded in | human’s feelings | |
and talk about it | emotional situations | |||
Offensive response | input–response | input–response | 110 K | improve CA |
dataset [241] | records from SimSimi | pairs and | chat pairs | abilities |
offensivity annotated | their annotation | |||
by crowd workers | ||||
BURCHAK dataset | dialogues of | chat outputs of | 177 dialogues | learning |
[242] | pairs of participants, | dialogues | 2454 turns | visually grounded |
discussing visual | word meanings | |||
attributes of 9 objects | in a foreign language | |||
The CIMA collection | conversations between | tutoring interactions | 2970 tutor | tutoring conversation |
[246] | crowd workers playing | and accompanying | responses | based on |
as students and tutors. | responses | to 350 exercises. | a provided strategy. |