Research Assistant Agent

Features

Installation

# Clone the repository
git clone https://github.com/davidburton/ResearchAssistantAgent.git
cd ResearchAssistantAgent
# Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install in development mode
pip install -e .

Quick Start

Command Line Interface

# Basic search
research-assistant search "transformer neural networks"
# Search only ArXiv
research-assistant search "quantum computing" --source arxiv --limit 20
# Search by author
research-assistant advanced-search --author "Yoshua Bengio" --limit 10
# Search by category (ArXiv)
research-assistant advanced-search --category cs.AI --limit 15
# Store results in vector database (requires OpenAI API key for embeddings)
research-assistant search "large language models" --store

Python API

import asyncio
from research_assistant import ArxivCollector, SemanticScholarCollector
async def search_papers():
 # Search ArXiv
 async with ArxivCollector() as arxiv:
 papers = await arxiv.search("cat:cs.LG transformer", max_results=5)
 for paper in papers:
 print(f"{paper.title} - {paper.arxiv_id}")
 
 # Search Semantic Scholar 
 async with SemanticScholarCollector() as s2:
 papers = await s2.search("deep learning", limit=5)
 for paper in papers:
 print(f"{paper.title} - Citations: {paper.citation_count}")
asyncio.run(search_papers())

Architecture

src/research_assistant/
├── collectors/ # API clients for paper sources
│ ├── arxiv_collector.py
│ └── semantic_scholar_collector.py
├── analyzers/ # LLM-based paper analysis
│ └── paper_analyzer.py
├── vector_store/ # FAISS similarity search
│ └── faiss_store.py
└── utils/ # Rate limiting and helpers
 └── rate_limiter.py

Configuration

export OPENAI_API_KEY="your-api-key" # For paper analysis and embeddings
export SEMANTIC_SCHOLAR_API_KEY="your-key" # Optional, for higher rate limits

Development

# Install development dependencies
pip install -r requirements-dev.txt
# Run tests
pytest
# Format code
black src/ tests/
# Type checking
mypy src/

API Rate Limits

Advanced Usage

Using the Rate Limiter

from research_assistant import RateLimiter, AdaptiveRateLimiter
# Fixed rate limiting
limiter = RateLimiter(max_calls=10, time_window=60) # 10 calls per minute
# Adaptive rate limiting (adjusts based on server responses)
adaptive = AdaptiveRateLimiter(
 initial_rate=10.0,
 min_rate=1.0,
 max_rate=50.0,
 backoff_factor=0.5
)
# Use with async context manager
async with limiter:
 # Your API call here
 pass

Paper Analysis with LLMs

from research_assistant import PaperAnalyzer, AnalysisType
analyzer = PaperAnalyzer(api_key="your-openai-key")
# Analyze a paper
analysis = await analyzer.analyze_paper(
 paper_text="Paper abstract or full text...",
 paper_id="arxiv.2301.00001",
 paper_title="Attention Is All You Need",
 analysis_type=AnalysisType.METHODOLOGY
)
print(analysis.methodology)
print(analysis.key_contributions)

Vector Store Operations

from research_assistant import FAISSVectorStore, Document
# Initialize vector store
store = FAISSVectorStore(dimension=1536, index_type="flat")
# Add documents
doc = Document(
 id="paper_001",
 text="Paper content...",
 metadata={"title": "Paper Title", "authors": ["Author 1"]},
 embedding=[0.1, 0.2, ...] # 1536-dimensional vector
)
store.add_documents([doc])
# Search similar documents
results = store.search(query_embedding, k=10)
# Save and load
store.save("./my_index")
loaded_store = FAISSVectorStore.load("./my_index")

Testing

# Run all tests
pytest
# Run with coverage
pytest --cov=research_assistant tests/
# Run specific test file
pytest tests/unit/utils/test_rate_limiter.py

Project Status

Contributing

License

Acknowledgments