References:
Use the following command to load this dataset in TFDS:
ds = tfds.load('huggingface:conv_questions')
- Description:
ConvQuestions is the first realistic benchmark for conversational question answering over knowledge graphs.
It contains 11,200 conversations which can be evaluated over Wikidata. The questions feature a variety of complex
question phenomena like comparisons, aggregations, compositionality, and temporal reasoning.
- License: CC BY 4.0
- Version: 1.0.0
- Splits:
| Split | Examples |
|---|---|
'test' |
2240 |
'train' |
6720 |
'validation' |
2240 |
- Features:
{
"domain": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"seed_entity": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"seed_entity_text": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"questions": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"answers": {
"feature": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"answer_texts": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
}
}