An intelligent enterprise-grade document analysis dashboard engineered to parse, validate, and isolate critical differences across multi-page PDF documents. Built using Streamlit and Google Gemini LLM.
- Deterministic Text Validation: Checks document density page-by-page, identifying blank contents or scanned images requiring OCR prior to analysis.
- Strict Anti-Hallucination Framework: Enforces strict boundary prompts, ensuring variations are only isolated from provided document text.
- Granular Page-Tracing Citations: Automatically targets, appends, and maps source data tags (e.g.,
[Page X]) for total data auditability. - Side-by-Side Parameter Matrix: Organizes granular differences (deadlines, salaries, roles) into a comparative display table.
- Production Report Exporter: Formats and compiles the entire markdown analysis stream into a structured Microsoft Word (.docx) file with one click.
ai-document-comparison-assistant/
├── app.py # Main Streamlit Dashboard UI & state memory coordinator
├── document_reader.py # Density checker, validator, and page-by-page parser
├── comparison.py # Gemini LLM connection framework & response builder
├── report_generator.py # Word document processing (.docx) export layer
├── requirements.txt # Production package dependency mapping
└── .gitignore # Security filter separating local files (.env, .venv)
git clone https://github.com
cd ai-document-comparison-assistantCreate a .env file in the root directory:
GEMINI_API_KEY=your_actual_alphanumeric_api_key_herepython -m venv .venv
source .venv/bin/activate # On Windows use: .venv\Scripts\activate
pip install -r requirements.txtpython -m streamlit run app.pyThe application will automatically initialize a web application session at http://localhost:8501.