Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.


Dataset Card for Steam 游戏评论时序数据集

Dataset Details

Uses

Direct Use

  • 热度(评论量对数)多步预测
  • 口碑(正向推荐比例)多步预测
  • 衰退风险二分类(未来热度显著下降的预警)
  • 特征消融与模型对比(LSTM vs Transformer,h_only / h_sent / full_no_topic)

Out-of-Scope Use

  • 不含主题特征,不能直接用于主题相关分析。
  • 情绪特征基于英文 VADER,对非英文评论信号有限;不适合跨语言情感对比。

Dataset Structure

  • 粒度:按游戏(app_id)和自然周聚合
  • 特征:热度 H_t、口碑 P_t、情绪(sent_mean/sent_std/neg_ratio)、元数据(价格/免费/EA/语言数/类型 One-Hot 等)
  • 划分:按时间顺序 70% 训练 / 15% 验证 / 15% 测试(在脚本中按窗口切分)

Dataset Creation

Curation Rationale

  • 以用户评论反馈为核心信号,构建可用于时间序列预测和运营预警的多源数据。

Source Data

Data Collection and Processing

  • 原始元数据:games.json(游戏信息)
  • 原始评论:Game Reviews/ 下的多 CSV,文件名 {app_id}_xxxx.csv
  • 预处理步骤:编码容错、字段补全、日期解析、推荐标准化、周聚合;按评论数/跨度/有效周数筛样本;合并情绪与元数据,生成最终周级表。

Annotations

  • 无人工标注;情绪分数由 VADER 自动生成。

Personal and Sensitive Information

  • 评论文本可能包含用户自述信息,使用时应遵守 Steam 条款并避免用于隐私挖掘;本数据集未额外脱敏。

Bias, Risks, and Limitations

  • 语言偏置:情绪特征依赖英文情感模型,非英文信号弱。
  • 样本偏置:仅保留评论数/跨度足够的游戏,长尾游戏未纳入。
  • 主题特征缺失:本版本未包含主题分布,无法提供内容结构信号。

Recommendations

  • 使用情绪特征时注意其英文偏置;口碑/热度结果更可靠于英文评论占比高的样本。
  • 需要内容信号可在后续版本加入主题模型(BERTopic 等)并重训。

Glossary

  • H_t:log(1 + review_count_t),热度指标
  • P_t:positive_ratio_t,口碑指标
  • h_only / h_sent / full_no_topic:特征消融组

More Information

Downloads last month
14