README.md

metadata

license: apache-2.0
task_categories:
  - text-generation
language:
  - zh
tags:
  - novel
  - structured
  - multimodal
size_categories:
  - 1K<n<10K

中文互动小说结构化语料

This dataset contains uncleaned (!) 3534 structured Chinese interactive novels (中文互动小说), accounting for around 0.25B (gpt-3.5) tokens in total.

All contents are parsed from certain online sources.

Usage

This dataset can be potentially used for LLM training. But be aware that you'd better clean the data yourself to remove undesired low-quality contents.

Each novel is a dict structured as follows:

class Novel:
    book_title: str
    book_author: str
    book_tag: list[str]
    book_intro: str
    collect: int    # 收藏
    popularity: int # 热度
    book_chapter: list[Chapter]  # Main content

class Chapter:
    chapter_title: str
    content: list[Content]

class Content:
    content_tag: str    # can be dialog, img, or narration
    role: str
    content: str

• Example snippet

Corresponding visual screenshot:

{
    "book_title": "庞博文：你喜欢我吗",
    "book_author": "李李李李李李李李",
    "book_tag": [
        "现代",
        "庞博文",
        "刘佳梁",
        "抖音网红",
        "马建博"
    ],
    "book_intro": "简介：意外爆火的新晋主播安然与抖音网红庞博文的相识相恋",
    "collect": 632,
    "popularity": 417220,
    "book_chapter": [
        ...
        {
            "chapter_title": "喝多了",
            "content": [
                ...
                {
                    "content_tag": "dialog",
                    "role": "庞博文",
                    "content": "拜拜"
                },
                {
                    "content_tag": "dialog",
                    "role": "安然",
                    "content": "拜拜"
                },
                {
                    "content_tag": "narration",
                    "role": "",
                    "content": "安然跟粉丝唠了一会儿磕后，屏幕上又出现了另一个人的面孔"
                },
                {
                    "content_tag": "dialog",
                    "role": "安然",
                    "content": "你好呀"
                },
                {
                    "content_tag": "dialog",
                    "role": "安然",
                    "content": "王子"
                },
                {
                    "content_tag": "dialog",
                    "role": "刘佳梁",
                    "content": "晚上好啊"
                },
                {
                    "content_tag": "dialog",
                    "role": "刘佳梁",
                    "content": "怎么着来一局吗"
                },
                {
                    "content_tag": "dialog",
                    "role": "安然",
                    "content": "再叫两个人吧"
                },
                ...
            ]
        },
        ...
    ]
}

Limitations

• As depicted, this is an uncleaned dataset. There might be null values (parser failure) throughout novels. There are content noises as well (e.g., author saying thanks or asking for likes and subscriptions)
• Not all pieces of work are of excellent quality. However, you can always refer to the popularity and collect meta info as a quality indicator.