This project provides the official code for the "Chain of Ideas" (CoI) Agent, designed to revolutionize research by facilitating novel idea development using Large Language Model (LLM) agents. It targets researchers and academics seeking to enhance their ideation process.

How It Works

The CoI Agent leverages LLMs to generate and refine research ideas through a structured "chain of thoughts" process. It integrates with external tools like SciPDF Parser for PDF analysis and GROBID for academic paper parsing, enabling it to process and synthesize information from existing research to propose novel concepts.

Quick Start & Requirements

• Installation: Clone the repository, install Python dependencies (pip install -r requirements.txt), install SciPDF Parser (pip install git+https://github.com/titipata/scipdf_parser), and download a spaCy model (python -m spacy download en_core_web_sm).
• Prerequisites: Java 11 (JDK 11.0.2 recommended) for GROBID, and API keys for LLM services (Azure OpenAI or OpenAI).
• Setup: Requires configuring API keys in config.yaml. GROBID needs to be running separately.
• Running: Execute python main.py --topic {your research topic}.
• Links: arXiv Paper, Online Demo

Highlighted Details

• Official implementation of the Chain of Ideas (CoI) Agent for LLM-driven research ideation.
• Integrates PDF parsing (SciPDF Parser) and academic document processing (GROBID).
• Supports both Azure OpenAI and OpenAI APIs for LLM interactions.
• Includes evaluation data and results within the dataset folder.

Maintenance & Community

The project is presented as the "first version" (as of Oct 2024). No specific community channels or roadmap details are provided in the README.

Licensing & Compatibility

The repository does not explicitly state a license. The code is provided for research purposes.

Limitations & Caveats

The setup requires manual installation and configuration of external services like GROBID and Java, which can be complex. API keys for LLMs are mandatory for operation.