jmultiwoz.py

# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# TODO: Address all TODOs and remove all explanatory comments
"""JMultiWOZ: Japanese Multi-Domain Wizard-of-Oz dataset for task-oriented dialogue modelling"""

import json
import os

import datasets


# TODO: Add BibTeX citation
# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """\
@inproceedings{ohashi-etal-2024-jmultiwoz,
    title = "JMultiWOZ: A Large-Scale Japanese Multi-Domain Task-Oriented Dialogue Dataset",
    author = "Ohashi, Atsumoto and Hirai, Ryu and Iizuka, Shinya and Higashinaka, Ryuichiro",
    booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation",
    year = "2024",
    url = "",
    pages = "",
}
"""

# TODO: Add description of the dataset here
# You can copy an official description
_DESCRIPTION = """\
JMultiWOZ is a large-scale Japanese multi-domain task-oriented dialogue dataset. The dataset is collected using 
the Wizard-of-Oz (WoZ) methodology, where two human annotators simulate the user and the system. The dataset contains 
4,246 dialogues across 6 domains, including restaurant, hotel, attraction, shopping, taxi, and weather. Available 
annotations include user goal, dialogue state, and utterances.
"""

# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = "https://github.com/nu-dialogue/jmultiwoz"

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = "CC BY-SA 4.0"

# TODO: Add link to the official dataset URLs here
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
_URLS = {
    "original_zip": "https://github.com/nu-dialogue/jmultiwoz/raw/master/dataset/JMultiWOZ_1.0.zip",
}


def _flatten_value(values) -> str:
    if not isinstance(values, list):
        return values
    flat_values = [
        _flatten_value(v) if isinstance(v, list) else v for v in values
    ]
    return "[" + ", ".join(flat_values) + "]"

# TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
class JMultiWOZDataset(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version("1.0.0")

    def _info(self):
        # TODO: This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
        features = datasets.Features({
            "dialogue_id": datasets.Value("int32"),
            "dialogue_name": datasets.Value("string"),
            "system_name": datasets.Value("string"),
            "user_name": datasets.Value("string"),
            "goal": datasets.Sequence({
                "domain": datasets.Value("string"),
                "task": datasets.Value("string"),
                "slot": datasets.Value("string"),
                "value": datasets.Value("string"),
            }),
            "goal_description": datasets.Sequence({
                "domain": datasets.Value("string"),
                "text": datasets.Value("string"),
            }),
            "turns": datasets.Sequence({
                "turn_id": datasets.Value("int32"),
                "speaker": datasets.Value("string"),
                "utterance": datasets.Value("string"),
                "dialogue_state": {
                    "belief_state": datasets.Sequence({
                        "domain": datasets.Value("string"),
                        "slot": datasets.Value("string"),
                        "value": datasets.Value("string"),
                    }),
                    "book_state": datasets.Sequence({
                        "domain": datasets.Value("string"),
                        "slot": datasets.Value("string"),
                        "value": datasets.Value("string"),
                    }),
                    "db_result": {
                        "candidate_entities": datasets.Sequence(datasets.Value("string")),
                        "active_entity": datasets.Sequence({
                            "slot": datasets.Value("string"),
                            "value": datasets.Value("string"),
                        })
                    },
                    "book_result": datasets.Sequence({
                        "domain": datasets.Value("string"),
                        "success": datasets.Value("string"),
                        "ref": datasets.Value("string"),
                    }),
                }
            }),
        })


        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            features=features,  # Here we define them above because they are different between the two configurations
            # If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
            # specify them. They'll be used if as_supervised=True in builder.as_dataset.
            # supervised_keys=("sentence", "label"),
            # Homepage of the dataset for documentation
            homepage=_HOMEPAGE,
            # License for the dataset if available
            license=_LICENSE,
            # Citation for the dataset
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        # TODO: This method is tasked with downloading/extracting the data and defining the splits depending on the configuration
        # If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name

        # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
        # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
        # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
        data_dir = dl_manager.download_and_extract(_URLS["original_zip"])
        split_list_path = os.path.join(data_dir, "JMultiWOZ_1.0/split_list.json")
        dialogues_path = os.path.join(data_dir, "JMultiWOZ_1.0/dialogues.json")
        with open(split_list_path, "r", encoding="utf-8") as f:
            split_list = json.load(f)
        with open(dialogues_path, "r", encoding="utf-8") as f:
            dialogues = json.load(f)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "dialogues": [dialogues[dialogue_name] for dialogue_name in split_list["train"]],
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "dialogues": [dialogues[dialogue_name] for dialogue_name in split_list["dev"]],
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                # These kwargs will be passed to _generate_examples
                gen_kwargs={
                    "dialogues": [dialogues[dialogue_name] for dialogue_name in split_list["test"]],
                },
            ),
        ]

    # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
    def _generate_examples(self, dialogues):
        # TODO: This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
        # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
        
        for id_, dialogue in enumerate(dialogues):
            example = {
                "dialogue_id": dialogue["dialogue_id"],
                "dialogue_name": dialogue["dialogue_name"],
                "system_name": dialogue["system_name"],
                "user_name": dialogue["user_name"],
                "goal": [],
                "goal_description": [],
                "turns": [],
            }

            for domain, tasks in dialogue["goal"].items():
                for task, slot_values in tasks.items():
                    if task == "reqt":
                        slot_values = {slot: None for slot in slot_values}
                    for slot, value in slot_values.items():
                        example["goal"].append({
                            "domain": domain,
                            "task": task,
                            "slot": slot,
                            "value": value,
                        })
            
            for domain, texts in dialogue["goal_description"].items():
                for text in texts:
                    example["goal_description"].append({
                        "domain": domain,
                        "text": text,
                    })
            
            for turn in dialogue["turns"]:
                example_turn = {
                    "turn_id": turn["turn_id"],
                    "speaker": turn["speaker"],
                    "utterance": turn["utterance"],
                    "dialogue_state": {
                        "belief_state": [],
                        "book_state": [],
                        "db_result": {},
                        "book_result": [],
                    },
                }
                if turn["speaker"] == "SYSTEM":
                    for domain, slots in turn["dialogue_state"]["belief_state"].items():
                        for slot, value in slots.items():
                            example_turn["dialogue_state"]["belief_state"].append({
                                "domain": domain,
                                "slot": slot,
                                "value": value,
                            })

                    for domain, slots in turn["dialogue_state"]["book_state"].items():
                        for slot, value in slots.items():
                            example_turn["dialogue_state"]["book_state"].append({
                                "domain": domain,
                                "slot": slot,
                                "value": value,
                            })
                    
                    candidate_entities = turn["dialogue_state"]["db_result"]["candidate_entities"]
                    active_entity = turn["dialogue_state"]["db_result"]["active_entity"]
                    if not active_entity:
                        active_entity = {}
                    example_turn["dialogue_state"]["db_result"] = {
                        "candidate_entities":candidate_entities,
                        "active_entity": [{
                            "slot": slot,
                            "value": _flatten_value(value),
                        } for slot, value in active_entity.items()]
                    }
                    
                    for domain, result in turn["dialogue_state"]["book_result"].items():
                        example_turn["dialogue_state"]["book_result"].append({
                            "domain": domain,
                            "success": result["success"],
                            "ref": result["ref"],
                        })
                
                example["turns"].append(example_turn)

            yield id_, example