PaperAgent: Academic Multi-Agent Collaboration Framework

Your AI-Powered Research Assistant for the Complete Academic Lifecycle

Features • Quick Start • Documentation • Demo • Contributing

📖 Overview

PaperAgent is a production-grade, academic-level multi-agent collaboration framework designed for graduate students, researchers, and academic teams. It covers the entire research workflow:

Research Topic Selection → Literature Review → Experiment Design → Paper Writing → Formatting → Submission → Results Management

Built on the KtR (Knowledge-to-Role) framework, PaperAgent systematically decomposes research tasks to avoid common multi-agent pitfalls like coordination overhead, performance redundancy, and debugging difficulties.

🎯 Core Goals

✅ Local Privacy Protection: All data stays on your machine
✅ Compliant Data Collection: Respects API terms of service (arXiv, Google Scholar)
✅ Academic Rigor: Multi-layer quality control for publication standards
✅ Efficient Collaboration: KtR-driven agent coordination

✨ Features

🔍 Literature Research (Literature Agent)

Smart Topic Recommendation: AI-powered research topic suggestions based on field analysis
Multi-source Paper Collection: Integrated arXiv & Google Scholar search
Deep Paper Analysis: Automatic summarization, clustering, and citation tracking
Research Gap Identification: Discover underexplored areas in your field
Citation Management: BibTeX generation and formatting

🔬 Experiment Management (Experiment Agent)

Experiment Design: AI-guided experimental methodology
Data Analysis: Statistical testing and visualization recommendations
Results Visualization: Automatic figure and table generation
Experiment-Paper Sync: Seamless integration with paper writing

✍️ Academic Writing (Writing Agent)

Structured Drafting: Generate complete paper outlines
Section Writing: AI-assisted writing for all paper sections
Academic Polishing: Grammar, style, and clarity improvements
Multi-format Citations: Support for IEEE, APA, MLA, GB/T 7714, etc.
Bilingual Support: English and Chinese academic writing

📝 Formatting & Submission (Formatting Agent)

Journal Templates: 50+ pre-built templates (IEEE, Elsevier, Springer, etc.)
LaTeX Integration: Full LaTeX support with Overleaf sync
Format Conversion: Word ↔ LaTeX ↔ PDF
Submission Checklist: Automated compliance checking

🎯 Orchestration (Boss Agent)

Task Decomposition: Intelligent workflow planning
Progress Monitoring: Real-time project tracking
Quality Control: Multi-layer validation and review
Error Recovery: Automatic retry and fallback mechanisms

🛠️ Additional Features

Local LLM Support: Ollama integration (Llama 3, Qwen2, etc.)
Cloud LLM Support: OpenAI GPT-4, Anthropic Claude
Docker Deployment: One-command setup
Web Interface: Intuitive Streamlit UI
REST API: Full-featured FastAPI backend

🎨 NEW - Enhanced Capabilities

📊 Advanced Statistical Analysis

Deep Statistical Methods: scipy, statsmodels, scikit-learn integration
Comprehensive Tests: t-tests, ANOVA, regression, chi-square, Mann-Whitney U
Machine Learning: Random Forest, cross-validation, feature importance
Effect Size Calculation: Cohen's d, eta-squared, Cramér's V

🎓 Advanced LaTeX Formatting

Complex Layouts: Multi-column layouts, custom environments
Algorithm Formatting: Professional algorithm and pseudocode environments
Theorem Environments: Theorems, lemmas, proofs, definitions with automatic numbering
Advanced Tables: Booktabs, multi-row/column, long tables spanning pages
Mathematical Formatting: Matrices, aligned equations, cases, integrals, derivatives

🖼️ Multimodal Analysis

Chart Understanding: Automatic chart type detection (bar, line, pie, scatter, histogram)
Data Extraction: Extract data points and trends from chart images
Formula Recognition: OCR-based mathematical formula recognition with LaTeX conversion
Table Analysis: Table structure detection and data extraction from images
Document Structure: Section extraction, citation parsing, reference analysis
Code Analysis: Complexity metrics, quality assessment, structure analysis

📈 Publication-Quality Visualizations

Static Plots: Scatter, bar, line, heatmap, box, violin, histogram plots
Interactive Visualizations: Plotly-based interactive charts and 3D plots
Correlation Matrices: Beautiful correlation heatmaps with statistical significance
Academic Style: Publication-ready figures with 300 DPI, color-blind friendly palettes

📄 Deep PDF Analysis

Full Text Extraction: Complete text extraction with NLP analysis
Table Extraction: Automatic table detection and conversion to DataFrames
Image Analysis: Extract and analyze all images and figures
Structure Analysis: Document hierarchy, sections, and cross-references
Metadata Extraction: Author, title, keywords, abstract parsing

🔗 Integrated Workflow

One-Stop Analysis: Unified interface for all multimodal content
Auto-Detection: Automatically detect content type (PDF, image, code, text)
Seamless Integration: All tools work together seamlessly

See ENHANCED_FEATURES.md for detailed usage guide and examples.

🚀 Quick Start

Prerequisites

Python 3.10 or higher
Docker and Docker Compose (optional but recommended)
PostgreSQL 16+ (or use Docker)
Redis 7+ (or use Docker)

Option 1: Docker Deployment (Recommended)

# Clone the repository
git clone https://github.com/yourusername/paperagent.git
cd paperagent

# Copy environment file
cp .env.example .env

# Edit .env with your API keys (optional, can use local LLM)
nano .env

# Start all services
docker-compose up -d

# Access the web interface
# http://localhost:8501

# Access the API documentation
# http://localhost:8000/docs

Option 2: Local Installation

# Clone repository
git clone https://github.com/yourusername/paperagent.git
cd paperagent

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Set up environment
cp .env.example .env
# Edit .env file with your configuration

# Initialize database
python -c "from paperagent.database import init_db; init_db()"

# Start API server
uvicorn paperagent.api.main:app --host 0.0.0.0 --port 8000

# In another terminal, start web interface
streamlit run paperagent/web/app.py

Option 3: Local LLM Setup (Privacy-Focused)

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model
ollama pull llama3

# Update .env
DEFAULT_LLM_PROVIDER=ollama
OLLAMA_MODEL=llama3

# Start PaperAgent
docker-compose up -d

📚 Usage Examples

Example 1: Complete Research Workflow

from paperagent.agents import BossAgent

# Initialize boss agent
boss = BossAgent()

# Create project
project = boss.execute({
    'action': 'create_project',
    'name': 'Deep Learning for Medical Imaging',
    'research_field': 'Computer Science',
    'description': 'Investigating CNN architectures for medical image diagnosis',
    'keywords': ['deep learning', 'medical imaging', 'CNN']
})

# Execute full workflow
result = boss.execute({
    'action': 'execute_workflow',
    'project_id': project['project_id']
})

# Check progress
progress = boss.execute({
    'action': 'monitor_progress',
    'project_id': project['project_id']
})

print(f"Progress: {progress['progress_percentage']}%")

Example 2: Literature Search Only

from paperagent.agents import LiteratureAgent

lit_agent = LiteratureAgent()

# Search papers
result = lit_agent.execute({
    'action': 'search_literature',
    'query': 'transformer models for NLP',
    'max_results': 50,
    'sources': ['arxiv', 'google_scholar'],
    'project_id': 1
})

print(f"Found {result['total_papers']} papers")

# Identify research gaps
gaps = lit_agent.execute({
    'action': 'identify_gaps',
    'project_id': 1,
    'research_field': 'Natural Language Processing'
})

for gap in gaps['research_gaps']:
    print(f"Gap: {gap['gap']}")
    print(f"Importance: {gap['importance']}")

Example 3: Paper Writing

from paperagent.agents import WritingAgent

writer = WritingAgent()

# Create paper structure
structure = writer.execute({
    'action': 'create_structure',
    'title': 'Advances in Transformer Architectures',
    'objective': 'Survey recent transformer improvements',
    'project_id': 1,
    'findings': ['Efficient attention mechanisms', 'Sparse transformers']
})

# Write introduction
intro = writer.execute({
    'action': 'write_section',
    'section': 'introduction',
    'context': 'Survey paper on transformers',
    'paper_id': structure['paper_id'],
    'key_points': ['Background', 'Motivation', 'Contributions']
})

# Generate abstract
abstract = writer.execute({
    'action': 'write_abstract',
    'paper_id': structure['paper_id'],
    'word_limit': 250
})

🏗️ Architecture

PaperAgent/
├── paperagent/
│   ├── agents/              # AI Agents
│   │   ├── boss_agent.py         # Central orchestrator
│   │   ├── literature_agent.py   # Literature research
│   │   ├── experiment_agent.py   # Experiment management
│   │   ├── writing_agent.py      # Paper writing
│   │   └── base_agent.py         # Base agent class
│   ├── api/                 # FastAPI backend
│   │   ├── main.py
│   │   ├── schemas.py
│   │   └── routers/
│   ├── web/                 # Streamlit frontend
│   │   └── app.py
│   ├── core/                # Core utilities
│   │   ├── config.py
│   │   ├── llm_manager.py
│   │   └── prompts.py
│   ├── database/            # Database models
│   │   ├── models.py
│   │   └── database.py
│   └── tools/               # Tool modules
│       ├── literature_collector.py
│       ├── latex_processor.py
│       └── document_processor.py
├── data/                    # Data storage
│   ├── papers/
│   ├── experiments/
│   └── literature/
├── docker-compose.yml
├── Dockerfile
├── requirements.txt
└── README.md

🔧 Configuration

Environment Variables

Key configuration options in .env:

# Database
DATABASE_URL=postgresql://paperagent:password@localhost:5432/paperagent

# LLM Provider (openai, anthropic, ollama)
DEFAULT_LLM_PROVIDER=ollama

# OpenAI (optional)
OPENAI_API_KEY=your_key_here
OPENAI_MODEL=gpt-4o

# Anthropic (optional)
ANTHROPIC_API_KEY=your_key_here
ANTHROPIC_MODEL=claude-3-5-sonnet-20241022

# Ollama (local)
OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_MODEL=llama3

# Literature Collection
USE_PROXY=false  # Enable for Google Scholar
ARXIV_MAX_RESULTS=50

🆚 Comparison

vs. Traditional Tools

Feature	Zotero+Word+Grammarly	PaperAgent
Full Workflow	❌ Fragmented	✅ Integrated
AI Assistance	❌ Limited	✅ End-to-end
Local Privacy	⚠️ Cloud-based	✅ Local-first
Multi-agent	❌ No	✅ KtR-driven

vs. General Multi-Agent Frameworks

Feature	AutoGen/CrewAI	PaperAgent
Domain-specific	❌ Generic	✅ Academic-focused
Agent Design	⚠️ Manual	✅ KtR-optimized
Academic Tools	❌ Limited	✅ Comprehensive
Quality Control	❌ Basic	✅ Multi-layer

🛣️ Roadmap

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Setup

# Clone repo
git clone https://github.com/yourusername/paperagent.git
cd paperagent

# Install dev dependencies
pip install -r requirements.txt
pip install -r requirements-dev.txt

# Run tests
pytest tests/

# Code formatting
black paperagent/
flake8 paperagent/

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

LangChain - LLM orchestration
LangGraph - Multi-agent workflows
arXiv API - Open access to research papers
Scholarly - Google Scholar API
FastAPI - Modern web framework
Streamlit - Rapid UI development

📞 Contact

Issues: GitHub Issues
Discussions: GitHub Discussions
Email: 2905153124@qq.com

⭐ Star History

📜 Citation

If you use PaperAgent in your research, please cite:

@software{paperagent2024,
  title = {PaperAgent: Academic Multi-Agent Collaboration Framework},
  author = {bingdongni},
  year = {2026},
  url = {https://github.com/bingdongni/paperagent}
}

Built with ❤️ for the research community

⬆ Back to Top

Name		Name	Last commit message	Last commit date
examples		examples
paperagent		paperagent
tests		tests
.env.example		.env.example
.gitignore		.gitignore
CONTRIBUTING.md		CONTRIBUTING.md
Dockerfile		Dockerfile
ENHANCED_FEATURES.md		ENHANCED_FEATURES.md
ENV_CONFIG_GUIDE.md		ENV_CONFIG_GUIDE.md
LICENSE		LICENSE
PROJECT_GUIDE_CN.md		PROJECT_GUIDE_CN.md
README.md		README.md
TESTING.md		TESTING.md
docker-compose.yml		docker-compose.yml
pytest.ini		pytest.ini
requirements.txt		requirements.txt
run.bat		run.bat
run.sh		run.sh
setup.py		setup.py
start.bat		start.bat
start.sh		start.sh

Latest commit

History

Folders and files

Repository files navigation