PERELMAN: Pipeline for scientific literature meta-analysis. Technical report

2025-12-25

AI Summary

Scientific literature exhibits high heterogeneity, and manual meta-analyses are inefficient and error-prone. Method: This paper proposes an intelligent agent pipeline tailored for systematic reviews. It introduces a novel, expert-knowledge-guided, multi-stage collaborative agent architecture that integrates structured human–agent dialogue, cross-modal information extraction (from text, tables, and figures), and standardized semantic mapping—enabling one-time domain knowledge injection and end-to-end knowledge-driven processing. Contribution/Results: We present the first reusable, end-to-end evidence structuring framework that automatically transforms unstructured scientific papers into unified, machine-readable, standardized evidence tables. Evaluated on a meta-analysis task for NMC811 lithium-ion battery cathode materials, our pipeline reduces analysis time from months to minutes while ensuring high reproducibility. This significantly advances the automation, scalability, and reliability of large-scale literature synthesis.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Search and Optimization: Metareasoning and Metaheuristics
• Multiagent Systems: Agent/AI Theories and Architectures

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

We present PERELMAN (PipEline foR sciEntific Literature Meta-ANalysis), an agentic framework designed to extract specific information from a large corpus of scientific articles to support large-scale literature reviews and meta-analyses. Our central goal is to reliably transform heterogeneous article content into a unified, machine-readable representation. PERELMAN first elicits domain knowledge-including target variables, inclusion criteria, units, and normalization rules-through a structured dialogue with a subject-matter expert. This domain knowledge is then reused across multiple stages of the pipeline and guides coordinated agents in extracting evidence from narrative text, tables, and figures, enabling consistent aggregation across studies. In order to assess reproducibility and validate our implementation, we evaluate the system on the task of reproducing the meta-analysis of layered Li-ion cathode properties (NMC811 material). We describe our solution, which has the potential to reduce the time required to prepare meta-analyses from months to minutes.

Problem

• Extract specific information from large scientific literature corpus
• Transform heterogeneous article content into unified machine-readable representation
• Reduce meta-analysis preparation time from months to minutes

Innovation

• Agentic framework extracts information from scientific articles
• Structured dialogue elicits domain knowledge from experts
• Coordinated agents aggregate evidence across studies consistently