🎓 ArXiv Research Assistant (MCP Server)
An automated Research Agent server built with the Model Context Protocol (MCP). This tool helps Data Scientists and Researchers search ArXiv, analyze papers, and discover insights using Machine Learning—all directly from their IDE.
🚀 Features
- automated Search: Query ArXiv for papers on any topic.
- Intelligent Extraction: Uses NLP heuristics to extract the core "Problem", "Method", and "Result" from abstracts.
- Data Science Pipeline:
- Saves findings to a structured dataset (
references.csv). - Visualize Trends: Generate charts of dominating research methods (
analyze_references.py). - Topic Modeling: Uses NMF (Non-Negative Matrix Factorization) to automatically discover hidden research themes (
topic_modeling.py).
- Saves findings to a structured dataset (
- Knowledge Graph: Generates an interactive HTML network graph of author collaborations (
generate_network.py).
Installation
Prerequisites
- Python 3.10+ (Tested on Python 3.14)
Setup
-
Clone the repository:
git clone <your-repo-url> cd mcp-project -
Create a Virtual Environment:
python3 -m venv .venv source .venv/bin/activate -
Install Dependencies:
pip install -r requirements.txt
⚙️ Configuration
Add the server to your IDE's MCP settings (e.g., mcp-servers.json in VS Code or Claude Desktop):
{
"mcpServers": {
"research-assistant": {
"command": "/absolute/path/to/your/mcp-project/.venv/bin/python",
"args": [
"/absolute/path/to/your/mcp-project/research_server.py"
],
"env": {
"PYTHONPATH": "/absolute/path/to/your/mcp-project"
}
}
}
}
Note: Replace /absolute/path/to/your/mcp-project/ with the actual full path on your machine.
💡 Usage
1. The MCP Agent
Once configured, you can ask your AI Assistant commands like:
"Find 5 papers on 'Reinforcement Learning from Human Feedback', extract their key findings, and save them to my bibliography."
2. Analysis Tools
Run these scripts to generate insights from your collected references.csv:
-
Visualize Keyword Trends:
python analyze_references.pyGenerates
method_keywords.png. -
View Author Network:
python generate_network.pyGenerates
author_network.html(interactive). -
Discover Hidden Topics:
python topic_modeling.pyGenerates
references_with_topics.csvwith ML-assigned topic clusters.
📂 Project Structure
research_server.py: The core MCP server logic.analyze_references.py: Visualization script for keyword frequencies.generate_network.py: NetworkX/Pyvis script for knowledge graphs.topic_modeling.py: Scikit-learn script for NMF topic modeling.references.csv: The dataset built by the agent.requirements.txt: Python dependencies.
Built with mcp, arxiv, pandas, scikit-learn, and networkx.