We are proud to have participated in the grand finale of the Dark Patterns Buster Hackathon hosted by the Department of Consumer Affairs, Government of India, at IIT BHU, Varanasi. The hackathon provided a unique platform to contribute to the fight against deceptive practices in online spaces. Our project, Guard’s Eye, is a state-of-the-art browser extension aimed at promoting transparency and fairness in e-commerce platforms.
The hackathon reinforced the importance of ethical digital practices and the power of teamwork in driving meaningful change in the digital landscape.
Guard's Eye is a browser extension designed to improve online shopping experiences by tackling common issues such as Fake Reviews, User Interface Deception, and Misleading Product Information. The extension leverages machine learning and image recognition technologies to provide users with comprehensive tools to verify the authenticity of product-related information. It also safeguards users from deceptive practices like Fake Urgency, Privacy Intrusion, and Subscription Trickery, while ensuring transparency in Data Governance policies.
-
Misleading Product Information & UI Deception
- Comprehensive product analysis by cross-referencing images, descriptions, and reviews.
- Identification of discrepancies to ensure trustworthy information.
-
Fake Reviews
- Detection and filtering of fake reviews using advanced tools.
- Assigns credibility scores to reviews for better decision-making.
-
Fake Urgency
- Compares current prices with historical data to validate time-sensitive offers.
- Provides context to determine if urgency is real or artificially created.
-
User Interface Deception
- Identifies UI elements that mislead or hide crucial information from users.
-
Data Governance
- Evaluates websites for compliance with data privacy regulations (e.g., GDPR).
- Provides simplified overviews of data handling for better transparency.
-
Privacy Intrusion
- Scrutinizes website data practices and ensures compliance with privacy laws.
- Simplifies privacy policies for user clarity.
-
Subscription Trickery
- Detects misleading subscription offers and complex cancellation procedures.
- Helps users navigate subscription services confidently.
Price History Graph
A dynamic feature allowing users to view historical price data for products on e-commerce platforms. It helps users identify genuine discounts and avoid deceptive pricing tactics.
- Graphical Representation: Displays price trends over time with interactive features.
- Time Range Flexibility: Users can choose time frames (1 month, 3 months, 6 months, or year-to-date) to analyze price movements.
-
Data Scraping
- Utilizes Beautiful Soup and Selenium to scrape product data (images, reviews, text) from e-commerce websites.
-
Machine Learning Model
- Analyzes scraped data for dark patterns, fake reviews, privacy policy compliance, and subscription trickery.
-
Dark Pattern Detection
- Highlights deceptive tactics identified during analysis, empowering users to avoid manipulative practices.
-
Fake Review Identification
- Detects fake reviews and provides a credibility score for each.
-
User-Friendly Interface
- Delivers a clear and concise analysis of dark patterns, making it easy for users to make informed decisions.
- Blip Image Captioning: Generates captions for product images, enhancing visual content analysis.
- Pytesseract OCR: Extracts text from product images for detailed analysis.
- roBERTa Classifier: Detects and filters fake reviews from the dataset.
- Distill roBERTa+ Classifier: Analyzes textual content from e-commerce websites to identify dark patterns.
- Mistral LLM Model: Integrates analysis from image captions, extracted text, and reviews, ensuring data integrity. It also evaluates privacy policies and subscription tricks against legal standards.
Guard's Eye represents a robust solution to tackle various dark patterns in e-commerce. Through teamwork and cutting-edge technology, we aim to empower users to make informed choices and protect themselves from manipulative practices online.