Agentic AI System for Automated Systematic Literature Reviews
Context
This project operates within the domain of agentic AI, information retrieval, and research methodology automation. Systematic literature reviews (SLRs) are fundamental to evidence-based research, yet their execution remains largely manual and time-intensive. The project builds on an existing paper collection system that queries academic databases (arXiv, Semantic Scholar, DBLP) using structured configurations, and aims to transform it into a multi-agent pipeline where researchers provide a natural language topic description and receive a draft survey report with human oversight at critical decision points.
Motivation
A typical systematic literature review takes 6 to 18 months. The process involves crafting Boolean search queries across multiple databases, deduplicating results, screening papers, reading full texts, and writing a synthesis. Current tool support is fragmented: individual tools exist for searching, screening, or reference management, but no system ties the full pipeline together. Recent progress in agentic AI, where LLM-powered agents use tools, browse the web, and coordinate tasks, makes end-to-end automation feasible. No such open-source system exists today.
Goal
The student will design and prototype a multi-agent system that automates the systematic literature review pipeline. The system will include a planning agent that converts natural language input into structured search configurations, a collection agent that queries academic databases and uses browser automation for sources without APIs, a screening agent using active learning to prioritize relevant papers, an analysis agent that parses and extracts information from full-text PDFs, a synthesis agent that clusters papers semantically to identify themes and gaps, and a report generation agent that produces a draft survey with a PRISMA flow diagram. Human-in-the-loop checkpoints ensure research rigor at key stages. All components should use open-source tools and locally-hosted LLMs.
Requirements
The student should have solid programming skills in Python. Familiarity with web development (Next.js/TypeScript or React) is needed for the monitoring dashboard. Experience with or willingness to learn LLM orchestration frameworks (LangGraph, CrewAI) is important. The student should be comfortable working with REST APIs and have basic understanding of information retrieval concepts. Interest in research methodology and systematic reviews is beneficial.
Pointers
- LangGraph and CrewAI documentation cover multi-agent orchestration patterns.
- BrowserUse provides LLM-driven browser automation for sources without APIs.
- Ollama enables local LLM hosting with models such as DeepSeek-V3 and Qwen3.
- LlamaIndex supports retrieval-augmented generation for document analysis.
- Semantic Scholar, arXiv, and OpenAlex API documentation describe available academic data endpoints.
- PRISMA 2020 guidelines define the reporting standard for systematic reviews.