💻 computer science

Exploring Ethical Concerns of Mobile Applications from App Reviews: A Literature Survey

This paper presents a comprehensive survey of 37 studies analyzing mobile app reviews to identify user-reported ethical concerns regarding privacy, security, and accessibility, while highlighting current research gaps and proposing a future agenda for automated detection and improved development practices.

Original authors: Aakash Sorathiya, Gouri Ginde

Published 2026-04-21

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Aakash Sorathiya, Gouri Ginde

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you've just downloaded a new app to help you track your mood, order food, or manage your finances. You use it for a week, and something feels "off." Maybe it's asking for your location when it doesn't need it, maybe the text is too small to read, or maybe the subscription fees are hidden until the very end.

You go to the App Store, leave a one-star review, and vent your frustration. You think, "Surely the developers will read this and fix it."

But here's the problem: Developers are often overwhelmed. They have thousands of reviews, and most are just saying "Great app!" or "Crashed on launch." Finding the specific complaints about ethics (like privacy, safety, or fairness) is like trying to find a specific needle in a haystack made of other needles.

This paper is a massive detective report written by researchers at the University of Calgary. They didn't just look at one app; they looked at 37 different scientific studies that tried to solve this exact problem. They wanted to answer: How can we automatically find and understand the ethical complaints hidden inside millions of app reviews?

Here is the breakdown of their findings, explained with some everyday analogies.

1. The "Seven Deadly Sins" of Apps

The researchers identified seven main categories of ethical problems that users complain about. Think of these as the "Seven Deadly Sins" of the app world:

Privacy: The app is a nosy neighbor who won't stop peeking through your curtains. (e.g., "Why does this flashlight app need my contact list?")
Security: The app has a broken lock on the front door. (e.g., "My password was stolen," or "The app has a virus.")
Accessibility: The app is a building with no ramps or elevators. (e.g., "I can't use this because the text is too small," or "It doesn't work with my screen reader.")
Transparency: The app is a magician hiding its tricks. (e.g., "I thought this was free, but then they charged me $50.")
Fairness: The app is a referee who favors one team. (e.g., "The algorithm shows different prices to different people," or "The chat bot is sexist.")
Accountability: The app is a ghost that disappears when you have a problem. (e.g., "I emailed support three times and got no reply.")
Safety: The app is a playground with a broken slide. (e.g., "This app encourages dangerous behavior," or "I was harassed by other users.")

2. The Detective Work (How they did it)

The researchers acted like archaeologists. They dug through 553 potential studies and carefully selected the 37 best ones.

The Dig Site: They looked at reviews from Google Play and the Apple App Store.
The Tools: Some studies used robots (computer programs) to scan millions of reviews for keywords like "hack" or "privacy." Others used human detectives who read reviews one by one to understand the nuance.
The Language Barrier: Most of the time, the robots only spoke English. This is a big problem because it ignores the complaints of billions of people who speak Spanish, Mandarin, or Hindi. It's like trying to understand a global party by only listening to the people speaking English in the corner.

3. What They Found (The "Aha!" Moments)

The researchers discovered a few surprising things:

The "Privacy Obsession": Most studies focus heavily on Privacy and Security. It's like everyone is worried about the house being burglarized, but they are ignoring the fact that the roof is leaking (Accessibility) or the landlord is being unfair (Fairness).
The "Silent Majority": Many ethical issues are reported by users, but because they are buried in the noise of "Great app!" reviews, developers often miss them.
The Root Causes: Why do these bad apps exist?
- The "Checklist" Trap: Developers often treat ethics like a checklist ("Did we check the privacy box? Yes.") rather than a mindset. They follow the rules on paper but fail in real life.
- The "Speed vs. Quality" Race: App stores are like a fast-food drive-thru. Developers are in a rush to get their app out the door before the competition, so they skip the "ethical quality control" steps.
- The "Black Box" of Stores: The App Stores themselves are a bit mysterious. They don't always help users report ethical issues effectively, and their rating systems can hide bad behavior.

4. The Future Plan (How to fix it)

The paper suggests a four-step roadmap to make apps more ethical:

Better Detective Tools: We need smarter AI that can understand context. Instead of just looking for the word "privacy," the AI should understand why a user is angry about data collection. It needs to speak many languages, not just English.
The "Ethical Dashboard": Imagine a dashboard for app developers (like a car dashboard) that lights up red when an app is collecting too much data or has accessibility bugs. This tool would alert them while they are building the app, not after users complain.
Listening to Everyone: We need to include voices from all over the world, not just the US or Europe. We also need to talk to the developers and the policymakers, not just the users.
Ethics as a Habit: Instead of treating ethics as a final step before launching, it needs to be baked into the recipe from the start. Just like you wouldn't build a house without a foundation, you shouldn't build an app without ethical safeguards.

The Big Picture

Think of the mobile app ecosystem as a giant city.

Users are the citizens.
Developers are the builders.
App Reviews are the town hall meetings where citizens shout out their problems.

Right now, the town hall is so noisy that the builders can't hear the specific complaints about safety or fairness. This paper is a guide on how to build a soundproof microphone system (better tools and methods) so that the builders can finally hear the citizens, fix the broken bridges, and make the city a safer, fairer place for everyone.

In short: Apps are everywhere, and they need to be trustworthy. This paper shows us how to listen to the people who use them to make sure they are safe, fair, and respectful.

1. Problem Statement

Mobile applications (apps) are increasingly embedded in critical domains (healthcare, finance, governance), raising significant ethical concerns regarding privacy, security, accessibility, transparency, fairness, accountability, and safety. While these are often categorized as non-functional requirements (NFRs), developers frequently prioritize functional features to meet tight market deadlines, leading to the neglect of these ethical attributes.

Although app reviews in stores (e.g., Google Play, Apple App Store) serve as a rich source of user feedback regarding bugs and feature requests, ethical concerns reported by users are often buried among general feedback and remain unidentified by automated analysis tools. While extensive research exists on mining app reviews for functional requirements, there is a lack of a comprehensive synthesis specifically focused on extracting and analyzing ethical concerns from these reviews. This gap hinders developers from proactively addressing ethical risks during the requirements engineering phase.

2. Methodology

The authors conducted a Systematic Literature Review (SLR) following the guidelines of Kitchenham [26].

Search Strategy:
- Databases: Scopus, Web of Science, ACM Digital Library, and IEEE Xplore.
- Search Query: A comprehensive string combining terms for "app reviews" (e.g., user feedback, comments) AND specific ethical concerns (privacy, security, accessibility, transparency, fairness, accountability, safety) AND "mobile applications."
- Timeframe: Studies published between 2012 and early 2025.
Selection Criteria:
- Inclusion: Papers must analyze app reviews specifically to assess ethical concerns and provide empirical findings on the seven defined ethical dimensions.
- Exclusion: Papers published before 2012, non-English, non-peer-reviewed, conference abstracts without detail, or studies analyzing ethical concerns without using app reviews as a data source.
Process:
- Initial retrieval: 553 papers.
- Screening: Three rounds of filtering (Title/Abstract, Introduction/Conclusion, Full Text) plus a snowball search (backward and forward citation tracking).
- Final Corpus: 37 primary studies were selected for detailed analysis.
Data Extraction & Synthesis:
- A custom extraction form captured data on dataset characteristics, methodologies, resources used, and specific user-reported issues.
- Data was synthesized qualitatively to answer four Research Questions (RQs), focusing on identifying patterns, gaps, and root causes.

3. Key Contributions

The paper makes three primary contributions to the field of Software Engineering (SE) and Human-Computer Interaction (HCI):

Comprehensive Synthesis of Ethical App Review Analysis: It is the first survey to integrate app review analysis with the specific domain of ethical concerns (privacy, security, accessibility, transparency, fairness, accountability, safety). It maps how researchers currently identify, classify, and validate these concerns.
Identification of Research Gaps and Imbalances: The survey reveals a heavy research bias toward privacy and security (appearing in >50% of studies), while fairness, accountability, and safety are significantly under-researched. It highlights that current methods often rely on technical measurability, overlooking complex socio-technical ethical issues.
Proposed Research Agenda & Integration Framework:
- A 4-point research agenda focusing on automated mining, AI/NLP advancements, developer tooling, and community engagement.
- A Taxonomy for Integrating Ethical Concerns across the entire Software Engineering (SE) lifecycle (Requirements, Design, Implementation, Testing, Deployment, Maintenance, and Project Management), moving ethics from a post-hoc check to a continuous process.

4. Key Results and Findings

A. Dataset Characteristics (RQ1)

Scale: Studies varied widely, analyzing from 500 to 626 million reviews and 1 to 1.3 million apps.
Sources: The Google Play Store was the dominant source, followed by the Apple App Store.
Domains: Mental health apps were the most frequent focus, followed by finance, education, and sharing economy apps.
Language Bias: 89% of studies analyzed only English reviews, creating a significant gap in understanding non-English user perspectives.
Collection Methods: Ranged from custom web scraping and public APIs to manual extraction and reuse of existing datasets.

B. Methodologies and Resources (RQ2)

Analysis Stages: Studies typically followed a 7-stage pipeline: Data Collection $\rightarrow$ Preprocessing $\rightarrow$ Feature Extraction $\rightarrow$ Classification/Labeling $\rightarrow$ Validation $\rightarrow$ Interpretation.
Techniques: A mix of Quantitative (Machine Learning, NLP, Sentiment Analysis, Topic Modeling) and Qualitative (Thematic Analysis, Manual Coding) methods.
Validation: Many studies validated findings against external frameworks like WCAG (accessibility), GDPR/CCPA (privacy), and App Privacy Policies. Some used static/dynamic analysis tools (e.g., MobSF, AndroBugs) to verify user claims.

C. User-Reported Ethical Issues (RQ3)

The survey categorized specific user complaints into the seven ethical dimensions:

Privacy: Unauthorized data access, excessive permissions (e.g., location), and data sharing with third parties.
Security: Malware, phishing, lack of encryption, and account hacking.
Accessibility: Poor UI design, lack of screen reader support, and incompatibility with assistive technologies.
Transparency: Hidden costs, unclear subscription models, and misleading app descriptions.
Fairness: Algorithmic bias, discriminatory moderation, and unequal service quality across platforms.
Accountability: Unresponsive customer support and lack of redressal for harms.
Safety: Harmful content (e.g., self-harm communities), predatory users, and physical safety risks (e.g., unsafe rides).

Root Causes Identified:

Policy-Practice Misalignment: Apps claim compliance (e.g., GDPR) but fail in implementation.
Design Shortcomings: Ethical NFRs are treated as afterthoughts rather than core requirements.
Platform Constraints: Opaque app store algorithms and lack of structured reporting channels.
Trust Gaps: Users perceive hidden fees and biased systems.

D. Future Challenges (RQ4)

The authors identified 78 recommendations from the primary studies, categorized into 8 themes (e.g., dataset diversity, automated extraction, content analysis, user studies). These were consolidated into a Research Agenda:

Mining: Improve dataset diversity (non-English, iOS, niche apps) and move beyond keyword-based search to NLI (Natural Language Inference).
AI/NLP: Leverage Large Language Models (LLMs) and Knowledge Graphs for deeper semantic understanding of ethical nuances.
Tooling: Develop IDE plugins and dashboards to alert developers of ethical risks in real-time.
Engagement: Foster multi-stakeholder collaboration (users, developers, policymakers) via crowdsourcing and user studies.

5. Significance and Implications

For Developers & Architects: The paper provides a roadmap to integrate ethical NFRs early in the Requirements Engineering (RE) phase. By using the proposed taxonomy, teams can systematically translate user complaints into actionable requirements (e.g., "Add opt-in consent" or "Fix screen reader compatibility").
For Researchers: It highlights critical gaps, particularly the need to study fairness, accountability, and safety with the same rigor as privacy. It calls for moving beyond English-only datasets and purely quantitative methods.
For Policymakers: The findings offer empirical evidence of where regulations (like GDPR or accessibility laws) are failing in practice, suggesting areas for stricter enforcement or clearer guidelines.
Broader Impact: The survey bridges the gap between technical implementation and human values, advocating for a shift from "functional-only" development to "ethically-by-design" mobile ecosystems.

In conclusion, this survey establishes that while app reviews are a vital resource for identifying ethical failures, current research is fragmented and biased. The proposed research agenda and lifecycle taxonomy offer a structured path forward for creating more trustworthy, inclusive, and ethically robust mobile applications.