Mobile App Reviews with Appraisal-Theory Sentiment Labels

2.3 MB of mobile app review text manually labeled for sentiment based on appraisal theory parameters. The dataset was created by Ruping Zhang and last updated on June 1, 2026. It was used to train a model achieving a mean fold accuracy of 88.63% and ROC AUC of 90.91%.

Use Cases

• Training sentiment classifiers based on the described appraisal theory labels (High/Low).
• Benchmarking lightweight transformer models like MobileBERT for real-time text analysis.
• Studying the application of explainable AI (XAI) methods like SHAP for validating manual data annotation.
• Experimenting with data augmentation techniques using synonym replacement via WordNet.

Strengths

• Manual labels are validated using the SHAP method of Explainable AI.
• Model trained on this data reported a mean fold accuracy of 88.63% and ROC AUC of 90.91%.
• Released under a permissive CC-BY-4.0 license.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The 2.3 MB file size suggests a relatively small dataset.

Provenance

Source

Ruping Zhang via figshare.

Collection Method

Manually labeled mobile app reviews based on performance parameters aligned with appraisal theory.

Freshness

Last updated 2026-06-01 17:52:28; freshness should be verified.