App Reviews Labeled with Kano Model Factors for Classification

A replication package contains manually labeled app reviews for Kano model factors. The data includes two source datasets named Statik and Brunotte, as well as five downsampled versions used for classifier evaluation. The package was created by Michelle Binder, Annika Vogt, Adrian Bajraktari, and Andreas Vogelsang to accompany their research paper.

Use Cases

• Training classifiers to automatically categorize app reviews based on Kano model factors.
• Evaluating the performance of different machine learning models on imbalanced text data.
• Studying the relationship between user feedback language and product feature satisfaction.

Strengths

• Includes manually applied Kano model labels, providing a ground truth for training.
• Provides five downsampled datasets specifically for evaluating classifier performance on imbalanced data.

Limitations

• Row count and dataset scale are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Replication package for an academic paper by Michelle Binder et al.

Collection Method

Manually labeled app reviews from two source datasets.