View PDF

Abstract:In todays digital landscape, end-user feedback plays a crucial role in the evolution of software applications, particularly in addressing issues that hinder user experience. While much research has focused on high-rated applications, low-rated applications often remain unexplored, despite their potential to reveal valuable insights. This study introduces a novel dataset curated from 64 low-rated applications sourced from the Amazon Software Appstore (ASA), containing 79,821 user reviews. The dataset is designed to capture the most frequent issues identified by users, which are critical for improving software quality. To further enhance the dataset utility, a subset of 6000 reviews was manually annotated to classify them into six district issue categories: user interface (UI) and user experience (UX), functionality and features, compatibility and device specificity, performance and stability, customer support and responsiveness, and security and privacy issues. This annotated dataset is a valuable resource for developing machine learning-based approaches aiming to automate the classification of user feedback into various issue types. Making both the annotated and raw datasets publicly available provides researchers and developers with a crucial tool to understand common issues in low-rated apps and inform software improvements. The comprehensive analysis and availability of this dataset lay the groundwork for data-derived solutions to improve software quality based on user feedback. Additionally, the dataset can provide opportunities for software vendors and researchers to explore various software evolution-related activities, including frequently missing features, sarcasm, and associated emotions, which will help better understand the reasons for comparatively low app ratings.

Submission history

From: Nek Dil Khan [view email] [v1] Tue, 6 Jan 2026 13:32:49 UTC (328 KB)