Commit
·
473d560
1
Parent(s):
d9afa77
Create README.md (#2)
Browse files- Create README.md (2030eb705bb40ba2653fdcfea105b05406ddea80)
Co-authored-by: F <Hugging-zheng@users.noreply.huggingface.co>
README.md
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- zh
|
| 4 |
+
license: other
|
| 5 |
+
task_categories:
|
| 6 |
+
- text-classification
|
| 7 |
+
- text-generation
|
| 8 |
+
- conversational
|
| 9 |
+
- feature-extraction
|
| 10 |
+
tags:
|
| 11 |
+
- social-media
|
| 12 |
+
- xiaohongshu
|
| 13 |
+
- narrative-analysis
|
| 14 |
+
- research-only
|
| 15 |
+
size_categories:
|
| 16 |
+
- n<1K
|
| 17 |
+
configs:
|
| 18 |
+
- config_name: default
|
| 19 |
+
data_files:
|
| 20 |
+
- split: train
|
| 21 |
+
path: "debug_dataset.jsonl"
|
| 22 |
+
---
|
| 23 |
+
|
| 24 |
+
# Dataset Card for XHS Research Dataset
|
| 25 |
+
|
| 26 |
+
> **⚠️ Access Notice / 访问须知**
|
| 27 |
+
>
|
| 28 |
+
> **Status:** Closed / Restricted Access (目前闭源)
|
| 29 |
+
>
|
| 30 |
+
> This dataset is currently **private** and intended strictly for academic research purposes. It is not available for public download to strictly adhere to data privacy regulations and platform terms of use.
|
| 31 |
+
>
|
| 32 |
+
> 本数据集目前处于**闭源状态**,仅供特定学术研究使用。出于对隐私合规及平台使用条款的严格遵守,暂不向公众开放下载。
|
| 33 |
+
|
| 34 |
+
## Dataset Description (数据集简介)
|
| 35 |
+
|
| 36 |
+
本数据集包含一组从小红书(Xiaohongshu)采集的**公开笔记(Public Notes)**及其关联评论。数据采集目标旨在支持**强化学习(Reinforcement Learning, RL)训练**,特别是用于构建基于真实社交互动的奖励模型(Reward Model)与用户模拟环境。
|
| 37 |
+
|
| 38 |
+
该数据集在采集过程特别实施了严格的隐私过滤机制。
|
| 39 |
+
|
| 40 |
+
- **维护者 (Curated by):** AgentAlpha
|
| 41 |
+
- **语言 (Language):** 中文 (zh)
|
| 42 |
+
- **许可证 (License):** 仅供科研使用 (Research Use Only) - 非商业用途
|
| 43 |
+
|
| 44 |
+
### Supported Tasks (支持的任务)
|
| 45 |
+
- **奖励模型构建 (Reward Modeling):** 利用评论互动的点赞数、回复层级构建基于人类偏好(Human Preference)的奖励函数。
|
| 46 |
+
- **用户行为模拟 (User Behavior Simulation):** 训练 Agent 模拟真实用户在社交媒体环境下的浏览路径与互动反馈,构建 RL 训练环境。
|
| 47 |
+
- **对话策略优化 (Conversational Policy Optimization):** 基于真实的笔记-评论对(Post-Comment Pairs),使用 RL 算法优化对话 Agent 的回复策略与社交适应性。
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
## Dataset Structure (数据结构)
|
| 51 |
+
|
| 52 |
+
虽然数据本身不可见,但为了方便合作者理解数据形态,以下是数据模式(Schema)的详细说明。
|
| 53 |
+
|
| 54 |
+
### Data Instances (数据样例结构)
|
| 55 |
+
|
| 56 |
+
数据集中的每一条记录代表一篇笔记,包含其元数据(Metadata)和嵌套的评论列表。
|
| 57 |
+
|
| 58 |
+
```json
|
| 59 |
+
{
|
| 60 |
+
"title": "笔记标题示例",
|
| 61 |
+
"content": "笔记的完整正文内容...",
|
| 62 |
+
"tags": ["标签1", "标签2"],
|
| 63 |
+
"meta": {
|
| 64 |
+
"publish_date": "YYYY-MM-DD",
|
| 65 |
+
"original_date_location": "编辑于 2024-01-01 上海",
|
| 66 |
+
"author": "[已脱敏/已隐藏]"
|
| 67 |
+
},
|
| 68 |
+
"comments": [
|
| 69 |
+
{
|
| 70 |
+
"user_nick": "匿名用户_A1B2",
|
| 71 |
+
"content": "这是一条评论内容...",
|
| 72 |
+
"comment_like_count": 10,
|
| 73 |
+
"parent_id": null,
|
| 74 |
+
"date": "YYYY-MM-DD"
|
| 75 |
+
}
|
| 76 |
+
]
|
| 77 |
+
}
|
| 78 |
+
````
|
| 79 |
+
|
| 80 |
+
### Data Fields (字段说明)
|
| 81 |
+
|
| 82 |
+
| Field | Type | Description |
|
| 83 |
+
| :--- | :--- | :--- |
|
| 84 |
+
| `title` | `String` | 笔记的标题 / Title of the post. |
|
| 85 |
+
| `content` | `String` | 笔记的正文内容 / Main body text. |
|
| 86 |
+
| `tags` | `List[String]` | 笔记关联的话题标签 / Hashtags. |
|
| 87 |
+
| `meta` | `Dict` | 元数据集合 (Metadata). |
|
| 88 |
+
| 🔹 `publish_date` | `String` | 发布日期 (Standardized). |
|
| 89 |
+
| 🔹 `author` | `String` | **[Anonymized]** 作者昵称已脱敏或隐藏. |
|
| 90 |
+
| `comments` | `List[Dict]` | 评论区数据 (含二级评论). |
|
| 91 |
+
| 🔹 `user_nick` | `String` | **[Anonymized]** 评论者昵称 (e.g., "匿名用户_25854F"). |
|
| 92 |
+
| 🔹 `content` | `String` | 评论内容. |
|
| 93 |
+
| 🔹 `parent_id` | `String` | 父评论ID (用于构建回复树). |
|
| 94 |
+
|
| 95 |
+
---
|
| 96 |
+
|
| 97 |
+
### Considerations for Using the Data (使用考量)
|
| 98 |
+
|
| 99 |
+
#### Social Impact (社会影响)
|
| 100 |
+
本数据集捕捉了生活方式分享平台上的真实用户互动,反映了当前中文社交媒体环境下的消费行为、流行趋势以及社区叙事风格。
|
| 101 |
+
|
| 102 |
+
#### Discussion of Biases (偏见讨论)
|
| 103 |
+
- **来源偏差 (Source Bias):** 数据是基于特定的搜索关键词和推荐流采集的,可能无法代表平台用户的整体全貌。
|
| 104 |
+
- **选择性偏差 (Selection Bias):** 采集的关键词是由研究者根据小红书往年的搜索热词所决定的,与全量爬虫相比,可能会引入特定的行为选择偏差。
|
| 105 |
+
|
| 106 |
+
### Privacy & Ethical Standards (隐私与伦理)
|
| 107 |
+
|
| 108 |
+
本数据集严格遵守以下科研伦理规范:
|
| 109 |
+
|
| 110 |
+
1. **匿名化处理 (Anonymization):** 评论中的用户昵称均已替换为匿名 ID(如 `匿名用户_XXXX`)或已隐藏。
|
| 111 |
+
2. **仅限公开数据 (Public Data Only):** 仅包含公开可见的笔记,不涉及私信或受限内容。
|
| 112 |
+
3. **禁止商用 (No Commercial Use):** 本数据严禁用于商业挖掘或营销目的。
|