EmotionShift: Chinese Text Data for Emotion Shift Annotation

A consolidated repository for data and experiments on Chinese emotion shift annotation. The dataset includes experiments using BERT, RoBERTa, and LLM frameworks like LLaMA-Factory. It was authored by Xuemei TANG and last updated on 2026-05-30.

Use Cases

• Training emotion shift detection models based on Chinese text annotations.
• Benchmarking BERT and RoBERTa models on the emotion shift task described.
• Fine-tuning LLMs via the LLaMA-Factory framework for emotion analysis.

Strengths

• Explicitly mentions experiments with BERT, RoBERTa, and LLaMA-Factory, providing a clear technical scope.
• Has a specific Creative Commons license (CC-BY-4.0) for clear reuse terms.

Limitations

• The dataset size is listed as 0.0 B, indicating a potentially limited or placeholder scope.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

figshare

Freshness

Last updated 2026-05-30 12:18:35; freshness should be verified.

Geography

Chinese language context.