参考文献:
対話
次のコマンドを使用して、このデータセットを TFDS にロードします。
ds = tfds.load('huggingface:schema_guided_dstc8/dialogues')
- 説明：
The Schema-Guided Dialogue dataset (SGD) was developed for the Dialogue State Tracking task of the Eights Dialogue Systems Technology Challenge (dstc8).
The SGD dataset consists of over 18k annotated multi-domain, task-oriented conversations between a human and a virtual assistant.
These conversations involve interactions with services and APIs spanning 17 domains, ranging from banks and events to media, calendar, travel, and weather.
For most of these domains, the SGD dataset contains multiple different APIs, many of which have overlapping functionalities but different interfaces,
which reflects common real-world scenarios.
- ライセンス: CC BY-SA 4.0
- バージョン: 0.0.0
- 分割:
| スプリット | 例 |
|---|---|
'test' | 4201 |
'train' | 16142 |
'validation' | 2482 |
- 特徴：
{
"dialogue_id": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"services": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"turns": {
"feature": {
"speaker": {
"num_classes": 2,
"names": [
"USER",
"SYSTEM"
],
"names_file": null,
"id": null,
"_type": "ClassLabel"
},
"utterance": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"frames": {
"feature": {
"service": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"slots": {
"feature": {
"slot": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"start": {
"dtype": "int32",
"id": null,
"_type": "Value"
},
"exclusive_end": {
"dtype": "int32",
"id": null,
"_type": "Value"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"state": {
"active_intent": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"requested_slots": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"slot_values": {
"feature": {
"slot_name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"slot_value_list": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"actions": {
"feature": {
"act": {
"num_classes": 18,
"names": [
"AFFIRM",
"AFFIRM_INTENT",
"CONFIRM",
"GOODBYE",
"INFORM",
"INFORM_COUNT",
"INFORM_INTENT",
"NEGATE",
"NEGATE_INTENT",
"NOTIFY_FAILURE",
"NOTIFY_SUCCESS",
"OFFER",
"OFFER_INTENT",
"REQUEST",
"REQUEST_ALTS",
"REQ_MORE",
"SELECT",
"THANK_YOU"
],
"names_file": null,
"id": null,
"_type": "ClassLabel"
},
"slot": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"canonical_values": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"values": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"service_results": {
"feature": {
"service_results_list": {
"feature": {
"service_slot_name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"service_canonical_value": {
"dtype": "string",
"id": null,
"_type": "Value"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"service_call": {
"method": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"parameters": {
"feature": {
"parameter_slot_name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"parameter_canonical_value": {
"dtype": "string",
"id": null,
"_type": "Value"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
}
スキーマ
次のコマンドを使用して、このデータセットを TFDS にロードします。
ds = tfds.load('huggingface:schema_guided_dstc8/schema')
- 説明：
The Schema-Guided Dialogue dataset (SGD) was developed for the Dialogue State Tracking task of the Eights Dialogue Systems Technology Challenge (dstc8).
The SGD dataset consists of over 18k annotated multi-domain, task-oriented conversations between a human and a virtual assistant.
These conversations involve interactions with services and APIs spanning 17 domains, ranging from banks and events to media, calendar, travel, and weather.
For most of these domains, the SGD dataset contains multiple different APIs, many of which have overlapping functionalities but different interfaces,
which reflects common real-world scenarios.
- ライセンス: CC BY-SA 4.0
- バージョン: 0.0.0
- 分割:
| スプリット | 例 |
|---|---|
'test' | 21 |
'train' | 26 |
'validation' | 17 |
- 特徴：
{
"service_name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"description": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"slots": {
"feature": {
"name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"description": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"is_categorical": {
"dtype": "bool",
"id": null,
"_type": "Value"
},
"possible_values": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"intents": {
"feature": {
"name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"description": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"is_transactional": {
"dtype": "bool",
"id": null,
"_type": "Value"
},
"required_slots": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"optional_slots": {
"feature": {
"slot_name": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"slot_value": {
"dtype": "string",
"id": null,
"_type": "Value"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
},
"result_slots": {
"feature": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"length": -1,
"id": null,
"_type": "Sequence"
}
},
"length": -1,
"id": null,
"_type": "Sequence"
}
}