Chinese Social Media Travel Experience Corpus with Multi-Label Annotations, 2023-2025

135,578 Sina Weibo posts describing urban travel experiences from January 2023 to December 2025. The corpus was constructed by Siyu Long from over 3.4 million raw posts using rule-based cleaning and LLM-assisted semantic filtering, with each record including text, timestamp, location, and nine traffic evaluation labels. It is validated through manual checking and inter-annotator agreement analysis.

Use Cases

• Sentiment analysis and affective computing based on labeled travel experience descriptions.
• Sandbox simulation of heterogeneous traveler populations based on multi-dimensional traffic evaluation labels.
• Text-based world modeling for human-centric mobility research based on geotagged and timestamped social media posts.

Strengths

• 135,578 annotated posts provide a substantial text corpus for analysis.
• Multi-label annotation across nine traffic evaluation dimensions offers detailed categorization.
• Dataset construction involved rigorous validation including manual relevance checking and inter-annotator agreement analysis.
• Covers a three-year time range from 2023 to 2025.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Data is sourced exclusively from Sina Weibo, which may reflect platform-specific user biases.
• The 42.3 MB file size indicates a relatively small text dataset for modern NLP tasks.

Provenance

Source

Sina Weibo social media platform.

Collection Method

Data acquisition, rule-based cleaning, LLM-assisted semantic filtering, and multi-label annotation.

Time Range

January 2023 to December 2025

Freshness

Last updated 2026-05-21 07:39:37

Geography

China (province-level IP locations indicated)