E-Com Sentinel

🛡️ A Chrome extension designed to analyze online shopping risks on platforms like Shopee using AI. It provides users with a quantitative Trust Score and highlights common issues found in 1-star reviews, helping them make safer purchasing decisions.

✨ Features

• AI-Powered Analysis: Leverages the Google Gemini API to analyze 1-star review text.
• Risk Classification: Identifies specific risk categories: NonDelivery, Counterfeit, WrongItem, Damaged.
• Bayesian Trust Score: Calculates a quantitative Trust Score using a hybrid Bayesian average formula, combining star ratings and AI-identified risks.
• Risk Level Indication: Provides clear "Low", "Medium", or "High" risk labels based on the Trust Score.
• On-Page Summary: Displays a concise summary and risk level directly on the product page via a non-intrusive banner (with minimize option).
• Detailed Popup Analytics: Offers a detailed view in the extension popup, including:
  • A short, one-sentence summary of 1-star reviews.
  • Sentiment analysis of 1-star reviews.
  • A visual horizontal bar chart showing the top 3 identified issues.
  • Original product star rating and total review count.
• Caching: Caches analysis results locally to avoid repeated API calls for the same product within a time window.
• Privacy Focused: Only analyzes publicly visible review data; no personal user data is collected or transmitted.

🛠️ Technologies Used

• JavaScript (ES6+): Core logic for content scripts, background script, and popup.
• HTML5: Structure for the popup UI.
• CSS3: Styling and layout, including flexbox and animations.
• Chrome Extension APIs: For content injection, background processing, popup, storage, and messaging.
• Google Gemini API: For AI-powered review analysis and classification.
• Git: Version control.

📦 Installation

Prerequisites: A Google Chrome browser.

1. Download: Click the green "Code" button on this repository and select "Download ZIP".
2. Extract: Unzip the downloaded file to a folder on your computer (e.g., e-com-sentinel-main).
3. Open Chrome: Launch Google Chrome.
4. Navigate to Extensions: Type chrome://extensions in the address bar and press Enter.
5. Enable Developer Mode: Toggle the "Developer mode" switch in the top-right corner.
6. Load Unpacked: Click the "Load unpacked" button.
7. Select Folder: Navigate to and select the extracted folder (e.g., e-com-sentinel-main).
8. Install: The extension should now appear in your list of extensions.

🔐 API Key Configuration (Important!)

Warning

This step is crucial for the extension to work.

1. Obtain an API Key: Visit Google AI Studio and create an API key for the Gemini API.
2. Edit background.js: Inside the extracted folder, open the background.js file in a text editor (like Notepad, VS Code, etc.).
3. Find the API Key Line: Look for the line containing const GEMINI_API_KEY = "YOUR_API_KEY_HERE";.
4. Replace the Placeholder: Replace "YOUR_API_KEY_HERE" with your actual Google Gemini API key.

   const GEMINI_API_KEY = "your_actual_api_key_here";

5. Save the File: Save the changes to background.js.
6. Reload Extension: Go back to chrome://extensions, find E-Com Sentinel, and click the "Reload" button (or press Ctrl+R on the extensions page).

The extension is now ready to use.

🎯 Usage

1. Navigate to a product page on Shopee (or any supported platform if the script is adapted).
2. The extension will automatically start analyzing the 1-star reviews (if enabled).
3. A banner will appear at the top of the page showing the risk level, a short summary, and the Trust Score.
4. Click the extension icon in the Chrome toolbar to open the popup for detailed analytics.

🚀 How It Works

1. Data Fetching: The page-context-script.js fetches 1-star review text from the Shopee product page API.
2. Data Processing: The content.js script receives this data and sends it to the background.js script.
3. AI Analysis: The background.js script uses the Google Gemini API to:
   • Classify review text into risk categories (NonDelivery, Counterfeit, etc.).
   • Generate a short summary of the 1-star reviews.
4. Trust Score Calculation: A Bayesian formula is applied, combining the product's overall star rating, the number of reviews, and the severity/quantity of issues found by the AI in 1-star reviews.
5. UI Display: The calculated Trust Score, risk level, summary, and top issues are sent back to content.js (for the page banner) and stored locally for the popup (popup.html, popup.js).
6. Caching: The analysis results are cached using chrome.storage.local for the current product URL.