real-persona-chat.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.


"""
RealPersonaChat: A Realistic Persona Chat Corpus with Interlocutors' Own Personalities

This script is based on
https://github.com/huggingface/datasets/blob/d69d1c654c4645a0474731794a20d4c012d2d214/templates/new_dataset_script.py
"""


import json
from pathlib import Path

import datasets


_CITATION = """\
@inproceedings{yamashita-etal-2023-realpersonachat,
    title = "{R}eal{P}ersona{C}hat: A Realistic Persona Chat Corpus with Interlocutors{'} Own Personalities",
    author = "Yamashita, Sanae  and
      Inoue, Koji  and
      Guo, Ao  and
      Mochizuki, Shota  and
      Kawahara, Tatsuya  and
      Higashinaka, Ryuichiro",
    booktitle = "Proceedings of the 37th Pacific Asia Conference on Language, Information and Computation",
    year = "2023",
    pages = "852--861"
}

@inproceedings{yamashita-etal-2024-realpersonachat-ja,
    title = "{R}eal{P}ersona{C}hat: 話者本人のペルソナと性格特性を含んだ雑談対話コーパス",
    author = "山下 紗苗 and 井上 昂治 and 郭 傲 and 望月 翔太 and 河原 達也 and 東中 竜一郎",
    booktitle = "言語処理学会第30回年次大会発表論文集",
    year = "2024",
    pages = "2738--2743"
}
"""

_DESCRIPTION = """\
RealPersonaChat: A Realistic Persona Chat Corpus with Interlocutors' Own Personalities
"""

_HOMEPAGE = "https://github.com/nu-dialogue/real-persona-chat/"

_LICENSE = "CC BY-ND 4.0"

_VERSION = "1.0.0"

_URL = f"https://github.com/nu-dialogue/real-persona-chat/archive/refs/tags/v{_VERSION}.zip"


class RealPersonaChat(datasets.GeneratorBasedBuilder):
    """RealPersonaChat consists of dialogues and interlocutor information."""

    VERSION = datasets.Version(_VERSION)

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(
            name="dialogue",
            version=VERSION,
            description="This part contains dialogues"
            ),
        datasets.BuilderConfig(
            name="interlocutor",
            version=VERSION,
            description="This part contains interlocutor information"
            )
    ]

    DEFAULT_CONFIG_NAME = "dialogue"

    def _info(self):
        if self.config.name == "dialogue":
            features = datasets.Features(
                    {
                        "dialogue_id": datasets.Value("int32"),
                        "interlocutors": datasets.Sequence(datasets.Value("string"), length=2),
                        "utterances": datasets.Sequence(
                            {
                                "utterance_id": datasets.Value("int32"),
                                "interlocutor_id": datasets.Value("string"),
                                "text": datasets.Value("string"),
                                "timestamp": datasets.Value("timestamp[us]")
                                }
                            ),
                        "evaluations": datasets.Sequence(
                            {
                                "interlocutor_id": datasets.Value("string"),
                                "informativeness": datasets.Value("int32"),
                                "comprehension": datasets.Value("int32"),
                                "familiarity": datasets.Value("int32"),
                                "interest": datasets.Value("int32"),
                                "proactiveness": datasets.Value("int32"),
                                "satisfaction": datasets.Value("int32")
                                }
                            )
                        }
                    )

        elif self.config.name == "interlocutor":
            features = datasets.Features(
                    {
                        "interlocutor_id": datasets.Value("string"),
                        "persona": datasets.Sequence(datasets.Value("string"), length=10),
                        "personality": {
                            "BigFive_Openness": datasets.Value("float32"),
                            "BigFive_Conscientiousness": datasets.Value("float32"),
                            "BigFive_Extraversion": datasets.Value("float32"),
                            "BigFive_Agreeableness": datasets.Value("float32"),
                            "BigFive_Neuroticism": datasets.Value("float32"),
                            "KiSS18_BasicSkill": datasets.Value("float32"),
                            "KiSS18_AdvancedSkill": datasets.Value("float32"),
                            "KiSS18_EmotionalManagementSkill": datasets.Value("float32"),
                            "KiSS18_OffenceManagementSkill": datasets.Value("float32"),
                            "KiSS18_StressManagementSkill": datasets.Value("float32"),
                            "KiSS18_PlanningSkill": datasets.Value("float32"),
                            "IOS": datasets.Value("int32"),
                            "ATQ_Fear": datasets.Value("float32"),
                            "ATQ_Frustration": datasets.Value("float32"),
                            "ATQ_Sadness": datasets.Value("float32"),
                            "ATQ_Discomfort": datasets.Value("float32"),
                            "ATQ_ActivationControl": datasets.Value("float32"),
                            "ATQ_AttentionalControl": datasets.Value("float32"),
                            "ATQ_InhibitoryControl": datasets.Value("float32"),
                            "ATQ_Sociability": datasets.Value("float32"),
                            "ATQ_HighIntensityPleasure": datasets.Value("float32"),
                            "ATQ_PositiveAffect": datasets.Value("float32"),
                            "ATQ_NeutralPerceptualSensitivity": datasets.Value("float32"),
                            "ATQ_AffectivePerceptualSensitivity": datasets.Value("float32"),
                            "ATQ_AssociativeSensitivity": datasets.Value("float32"),
                            "SMS_Extraversion": datasets.Value("float32"),
                            "SMS_OtherDirectedness": datasets.Value("float32"),
                            "SMS_Acting": datasets.Value("float32"),
                            },
                        "demographic_information": {
                            "gender": datasets.ClassLabel(names=["Male", "Female", "Other"]),
                            "age": datasets.ClassLabel(names=["-19", "20-29", "30-39", "40-49", "50-59", "60-69"]),
                            "education": datasets.ClassLabel(names=["High school graduate", "Two-year college", "Four-year college", "Postgraduate", "Other"]),
                            "employment_status": datasets.ClassLabel(names=["Employed", "Homemaker", "Student", "Retired", "Unable to work", "None"]),
                            "region_of_residence": datasets.ClassLabel(names=["Hokkaido", "Aomori", "Iwate", "Miyagi", "Akita", "Yamagata", "Fukushima", "Ibaraki", "Tochigi", "Gunma", "Saitama", "Chiba", "Tokyo", "Kanagawa", "Niigata", "Toyama", "Ishikawa", "Fukui", "Yamanashi", "Nagano", "Gifu", "Shizuoka", "Aichi", "Mie", "Shiga", "Kyoto", "Osaka", "Hyogo", "Nara", "Wakayama", "Tottori", "Shimane", "Okayama", "Hiroshima", "Yamaguchi", "Tokushima", "Kagawa", "Ehime", "Kochi", "Fukuoka", "Saga", "Nagasaki", "Kumamoto", "Oita", "Miyazaki", "Kagoshima", "Okinawa"]),
                            },
                        "text_chat_experience": {
                            "age_of_first_chat": datasets.ClassLabel(names=["-9", "10-19", "20-29", "30-39", "40-49", "50-59"]),
                            "frequency": datasets.ClassLabel(names=["Every day", "Once every few days", "Once a week", "Less frequent than these"]),
                            "chatting_partners": datasets.Sequence(datasets.ClassLabel(names=["Family", "Friend", "Colleague", "Other"])),
                            "typical_chat_content": datasets.Value("string"),
                            }
                        }
                    )

        else:
            raise ValueError(f"Config name `{self.config.name}` is invalid.")

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        data_dir = dl_manager.download_and_extract(_URL)

        if self.config.name == "dialogue":
            filepath_list = Path(data_dir, f"real-persona-chat-{_VERSION}", "real_persona_chat", "dialogues").glob("*.json")
            filepath_list = list(sorted(filepath_list, key=lambda x: int(x.stem)))

        elif self.config.name == "interlocutor":
            filepath_list = Path(data_dir, f"real-persona-chat-{_VERSION}", "real_persona_chat").glob("interlocutors.json")

        else:
            raise ValueError(f"Config name `{self.config.name}` is invalid.")

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepath_list": filepath_list
                }
            )
        ]

    def _generate_examples(self, filepath_list):
        if self.config.name == "dialogue":
            for filepath in filepath_list:
                key = filepath.stem

                with open(filepath, encoding="utf-8") as f:
                    example = json.load(f)

                for utterance_id in range(len(example["utterances"])):
                    timestamp = example["utterances"][utterance_id]["timestamp"]

                    if timestamp == "NaT":
                        example["utterances"][utterance_id]["timestamp"] = "0001-01-01T00:00:00.000000"

                yield key, example

        elif self.config.name == "interlocutor":
            for filepath in filepath_list:
                with open(filepath, encoding="utf-8") as f:
                    interlocutors = json.load(f)

                for key, example in interlocutors.items():
                    yield key, example

        else:
            raise ValueError(f"Config name `{self.config.name}` is invalid.")