Mitigating Hallucinations in Large Language Models via Retrieval Augmented Generation: A Systematic Review of n8n-Based Implementations

Abstract

This study systematically examines hallucination phenomena in Large Language Models (LLMs), focusing on their characteristics, causal factors, and mitigation strategies through Retrieval-Augmented Generation (RAG) and low-code orchestration platforms such as n8n. Using a Systematic Literature Review (SLR) approach based on PRISMA 2020 guidelines, this study analysed 40 peer-reviewed articles published between 2020 and 2025 from major scientific databases. The findings reveal that hallucinations are multidimensional, consisting of factual, semantic, and contextual hallucinations influenced by static training data, probabilistic token prediction, prompt ambiguity, and insufficient validation mechanisms. The review further demonstrates that RAG significantly improves factual accuracy by integrating external retrieval systems with LLM generation processes. Recent innovations such as Hybrid Retrieval and GraphRAG enhance contextual relevance and knowledge representation. A major finding of this study is the identification of “Conflict of Information” between external retrieved data and internal LLM knowledge in automated RAG pipelines. Furthermore, this study proposes a novel conceptual framework and taxonomy for hallucination mitigation in low-code AI environments, integrating retrieval, validation, conflict resolution, and workflow orchestration mechanisms. These findings contribute to the development of more reliable, transparent, and scalable AI systems.