Mobile App Reviews with Appraisal-Theory Sentiment Labels

A manually annotated dataset for sentiment analysis of mobile app reviews, created by Ruping Zhang and last updated in June 2026. The dataset contains reviews labeled as High or Low based on performance parameters aligned with appraisal theory. It is a small dataset, 5.5 KB in size, and is available in XLS format.

Use Cases

• Training sentiment classifiers based on appraisal theory labels (appreciation and judgment).
• Benchmarking lightweight transformer models like MobileBERT for real-time text analysis.
• Developing explainable AI (XAI) methods for sentiment classification using SHAP.
• Applying data augmentation techniques using synonym replacement for NLP robustness.

Strengths

• Manual annotation accuracy was ensured using the SHAP method of Explainable AI.
• The associated model achieved a reported mean fold accuracy of 88.63% and ROC AUC of 90.91%.
• Labels are based on a detailed framework of seven performance parameters (e.g., insightfulness, precision).

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 5.5 KB, indicating a limited scope of reviews.

Provenance

Source

Ruping Zhang via figshare.

Collection Method

Reviews were manually labeled based on specified performance parameters.

Freshness

Last updated 2026-06-01 17:52:54; freshness should be verified.