Update README.md

README.md

@@ -13,7 +13,7 @@ Check out the configuration reference at https://huggingface.co/docs/hub/spaces-
 
 # CV to CSV Extraction App
 
-A Gradio application that extracts publications, talks, and other scholarly accomplishments from faculty CVs (PDFs) using Google's Gemini API.
+A Gradio application that extracts publications, talks, and other scholarly accomplishments from faculty CVs (PDFs) using Google's Gemini API with Pydantic-AI for robust structured data extraction.
 
 ## Features
 
@@ -22,6 +22,7 @@ A Gradio application that extracts publications, talks, and other scholarly acco
 - Display results in a tabular format
 - Download results as CSV
 - Password protection using Hugging Face secrets
+- Robust JSON parsing with Pydantic-AI
 
 ## Installation
 
@@ -39,6 +40,7 @@ A Gradio application that extracts publications, talks, and other scholarly acco
 3. Create a `.env` file in the root directory with your Google API key:
    ```
    GOOGLE_API_KEY=your_google_api_key_here
+   APP_PASSWORD=your_app_password_here
    ```
 
 ## Usage
@@ -74,12 +76,25 @@ A Gradio application that extracts publications, talks, and other scholarly acco
 
 2. **PDF Processing**: The app extracts text from uploaded PDF files using PyPDF2
 
-3. **LLM Processing**: The extracted text is sent to Google's Gemini API to identify faculty names and extract scholarly accomplishments
+3. **LLM Processing with Pydantic-AI**: 
+   - The extracted text is processed using Pydantic-AI with Google's Gemini model
+   - Pydantic models define the structure of the expected data
+   - This approach ensures more robust parsing and validation of the extracted data
+   - If Pydantic-AI processing fails, the app falls back to the standard Gemini API approach
 
 4. **Categorization**: Accomplishments are categorized into different types based on a decision tree approach
 
 5. **Results Display**: The extracted accomplishments are displayed in a tabular format and can be downloaded as CSV
 
+## Pydantic-AI Integration
+
+The app uses Pydantic-AI to improve the reliability of structured data extraction:
+
+- **Defined Data Models**: Clear schema definitions for faculty data and accomplishments
+- **Type Validation**: Ensures fields like years and confidence scores are properly typed
+- **Default Values**: Handles missing fields gracefully with sensible defaults
+- **Fallback Mechanism**: If Pydantic-AI extraction fails, the app falls back to standard extraction
+
 ## Customization
 
 ### Changing the Password
@@ -101,4 +116,4 @@ To modify the categories of scholarly accomplishments, edit the `MAIN_CATEGORIES
 
 ## License
 
-This project is licensed under the MIT License.
+This project is licensed under the MIT License.