12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

A Knowledge Graph-Based RAG-GLLM Approach for LiteratureReview and Knowledge Discovery

Yancong Xie, Yuanyuan Song, Richard Watson · Journal of the Association for Information Systems · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10
Relevance
0/4
Quality (LMQS)
C
Evidence
0
Citations
0.00
FWCI

Methodology & findings

Study design

Research-in-progress using conceptual design, development planning, and early prototyping of a knowledge graph-based RAG-GLLM system for literature review automation

Main result

The proposed knowledge graph-based RAG-GLLM approach "combines structured knowledge representation with generative reasoning" and "provides persistent context, reduce[s] hallucination, and enable[s] analysis across five levels of literature review, from concepts to themes." The research presents early prototyping results demonstrating the feasibility of this methodological foundation for scholarly literature reviews.

Research paradigm

Design science / computational methodology

Author conclusions

The authors conclude that "our research-in-progress outlines the conceptual design, development plan, and early prototyping results, contributing a methodological foundation for scalable, reliable, and theory-informed literature reviews in Information Systems research."

Limitations

  • The paper explicitly acknowledges that current RAG-GLLM applications "remain focused on fact retrieval rather than higher-level conceptual integration" and notes that GLLMs suffer from "hallucination, limited context windows, and weak evidential grounding" when applied to scholarly contexts
  • The work is presented as research-in-progress with only early prototyping results, not yet fully validated.

Open questions raised

  • The paper identifies that current RAG-GLLM applications remain focused on fact retrieval rather than higher-level conceptual integration, and that traditional and computational approaches are being pushed to their limits by the rapid growth of academic publications.
  • The paper identifies the gap that current RAG applications are "focused on fact retrieval rather than higher-level conceptual integration" and addresses the challenge of conducting literature reviews given "the rapid growth of academic publications" that pushes "traditional and computational approaches to their limits."
  • The paper identifies gaps in applying GLLMs to scholarly contexts: current RAG applications focus on fact retrieval rather than higher-level conceptual integration; there is a need for approaches that address hallucination, limited context windows, and weak evidential grounding in literature review processes.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 70%

Explore related topics

Systematic Review AutomationLiterature DiscoveryConceptual

Related papers

← Browse all papers