PDF PROFESSOR 1.0

Overview

PDF Professor 1.0 extracts text from PDFs using PyMuPDF (with a Poppler fallback), then processes the text in chunks through an Ollama model using a user-defined prompt. This makes it ideal for summarization, data extraction, and text analysis.

Key Features:

• Text Extraction & Chunking – Efficiently processes large PDFs into manageable sections.
• Custom Prompt Processing – Guides the Ollama model dynamically with your input.
• Logging & Resume Support – Tracks progress to continue where it left off.
• Timeout & Concurrency Control – Optimizes performance by adjusting processing limits.

---

Installation & Setup

1. Clone the Repository

git clone https://github.com/YourUsername/PDF-Professor.git
cd PDF-Professor

2. Set Up Environment

conda create -n pdfprofessorENV python=3.10
conda activate pdfprofessorENV

3. Install Dependencies

pip install -r requirements.txt

---

Configuration

All settings are stored in config.json:

{
    "pdf_directory": "PDF",
    "output_directory": "Scripts",
    "log_directory": "Logs",
    "chunk_storage_directory": "ProcessedChunks",
    "ollama_command": ["ollama", "run", "wizardlm2:7b"],
    "chunk_size": 2000
}

---

Usage

1. Run the Program

python pdfprofessor.py

2. Enter Your Prompt

Enter your prompt for Ollama: Summarize key points.

3. Processing & Output

• Chunks PDFs and processes them via Ollama.
• Saves processed content in ProcessedChunks and combined scripts in Scripts.

---

Model Selection & Setup

Recommended Models:

• WizardLM 2:7B – Primary model optimized for deep text analysis.
• DeepSeek-R1:7B – Ideal for case law reasoning and cybersecurity reports.
• Mistral or LLaMA 3.1 – Best for speed-focused processing of large PDF batches.

Changing Models:

1. Modify config.json

   • Locate the ollama_command setting and change the model name. Example:

   "ollama_command": ["ollama", "run", "deepseek-r1:7b"]

2. Ensure the Model is Downloaded Locally

   ollama pull <model-name>

   Example:

   ollama pull deepseek-r1:7b

3. Verify Available Models

   ollama list

4. Test Model Performance

   ollama run <model-name>

   Example:

   ollama run mistral:latest

Once the model is set in config.json, restart PDF Professor to apply the changes.

---

File Structure

PDF-Professor/
├── pdfprofessor.py              # Main script
├── config.json                  # Configuration file
├── requirements.txt             # Dependencies
├── PDF/                         # Source PDFs
├── Logs/                        # Logs
├── ProcessedChunks/             # Chunk outputs
└── Scripts/                     # Final scripts

---

Prompt List

OSINT & Investigations

Fraud & Financial Crimes:

• Identify and analyze patterns in fraud schemes, financial crimes, and risk assessment methodologies.
• Detect and analyze trends in whistleblower disclosures related to fraud and misconduct.
• Convert whistleblower testimonies or case disclosures into structured datasets for fraud detection and AI-assisted risk analysis.
• Parse and analyze financial records, including bank statements and transaction histories, to detect anomalies and fraudulent activities.
• Examine corporate and government financial reports to identify trends, discrepancies, and regulatory compliance issues.

Social Media & Misinformation:

• Analyze patterns in social media misinformation campaigns and their impact.
• Extract social media investigation techniques and categorize them by platform, methodology, and risk factors.
• Identify and categorize social media manipulation tactics used in misinformation campaigns.
• Identify references to AI-generated misinformation and structure them into datasets for detection and mitigation training.

Surveillance & Intelligence Gathering:

• Identify and categorize surveillance techniques used in corporate and law enforcement settings, including digital vs. physical methodologies.
• Extract intelligence-gathering methodologies applicable to business investigations and OSINT.
• Identify intelligence-gathering techniques used in OSINT investigations and legal contexts.
• Categorize intelligence cycle references into collection, analysis, dissemination, and operational use.

Case Studies & Incident Analysis:

• Analyze crash reports to identify contributing factors, patterns, and potential liabilities.
• Review and summarize incident reports to highlight key events, responsible parties, and procedural gaps.
• Extract relevant statements from witness testimonies and analyze inconsistencies for case evaluation.
• Parse case studies of past cyberattacks and summarize them as structured incident reports.
• Identify trends and patterns in criminal behavior based on case study data.
• Extract real-world OSINT case studies and format them for AI-assisted investigation workflows.

Legal

Legal Analysis & Case Law:

• Extract and structure legal statutes, case laws, and precedents for AI-driven legal analysis.
• Identify and analyze patterns in legal case outcomes to predict case trajectories.
• Analyze historical patterns in legal cases to determine causation and liability, identifying contributing factors and responsible parties.

Witness & Evidence Analysis:

• Identify recurring themes and inconsistencies in witness statements to assess reliability.
• Examine, analyze, and detail inconsistencies in depositions and court transcripts.
• List witnesses and categorize their statements based on relevance and reliability.
• Organize legal evidence and testimonies for structured case-building.

Regulatory & Compliance:

• Extract all legal definitions, jurisdictional laws, and regulatory frameworks for different types of cybercrime.
• Identify all data privacy laws and format them as structured training sets for compliance analysis.
• Summarize all regulations on electronic surveillance and data retention laws across different jurisdictions.

Cybersecurity

Threat Analysis & Incident Response:

• Analyze patterns in cybersecurity incidents, including intrusion attempts and attack methodologies.
• Analyze threat actors involved in ransomware and breach data attacks.
• Parse the document for mentions of threat actors, hacking groups, or cybercriminal organizations and summarize their activities.
• Summarize all incident response frameworks and standard operating procedures for AI-assisted cybersecurity training.

Attack Vectors & Defenses:

• Generate a structured list of attack vectors, defenses, and mitigation strategies.
• Convert hacking methodologies and defense mechanisms into red team vs. blue team training sets.
• Identify cybersecurity frameworks like NIST, ISO 27001, and CIS controls and format them into structured datasets.

Insider Threats & Counterintelligence:

• Extract and categorize all references to insider threats and mitigation strategies.
• Extract all discussions on counterintelligence tactics and cyber deception strategies.

Phishing & Misinformation:

• Identify structured datasets related to phishing techniques and misinformation campaigns.
• Extract structured datasets on phishing techniques, categorized by attack type and mitigation strategies.

---

Troubleshooting

• Timeout Errors – Increase timeout settings.
• Slow Performance – Reduce chunk size or adjust concurrency.
• Model Choice – Use a smaller Ollama model if needed.

---

License

This project is available under the MIT License. Contributions are welcome!