SOPRAG: Multi-view Graph Experts Retrieval for Industrial Standard Operating Procedures

2026-02-02

AI Summary

This work addresses the challenge of retrieving industrial standard operating procedures (SOPs), which are difficult for conventional semantic retrieval-augmented generation (RAG) methods to handle due to their rigid structure, complex conditions, and strong action dependencies. To this end, we propose SOPRAG, a novel multi-expert graph-based RAG framework tailored for SOPs. SOPRAG models SOP logic through three specialized graph experts—entity, causal, and procedural—and dynamically fuses them via intent-aligned gating guided by large language models (LLMs), enhanced with Procedure Card pruning for efficiency. To mitigate domain data scarcity, we introduce a multi-agent pipeline for automated benchmark construction. Evaluated across four industrial scenarios, SOPRAG significantly outperforms lexical, dense, and graph-based RAG baselines, achieving 100% accuracy on critical task execution and setting new state-of-the-art results in both retrieval precision and response utility.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Knowledge Representation and Reasoning: Logic Programming

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Standard Operating Procedures (SOPs) are essential for ensuring operational safety and consistency in industrial environments. However, retrieving and following these procedures presents unique challenges, such as rigid proprietary structures, condition-dependent relevance, and actionable execution requirement, which standard semantic-driven Retrieval-Augmented Generation (RAG) paradigms fail to address. Inspired by the Mixture-of-Experts (MoE) paradigm, we propose SOPRAG, a novel framework specifically designed to address the above pain points in SOP retrieval. SOPRAG replaces flat chunking with specialized Entity, Causal, and Flow graph experts to resolve industrial structural and logical complexities. To optimize and coordinate these experts, we propose a Procedure Card layer that prunes the search space to eliminate computational noise, and an LLM-Guided gating mechanism that dynamically weights these experts to align retrieval with operator intent. To address the scarcity of domain-specific data, we also introduce an automated, multi-agent workflow for benchmark construction. Extensive experiments across four industrial domains demonstrate that SOPRAG significantly outperforms strong lexical, dense, and graph-based RAG baselines in both retrieval accuracy and response utility, achieving perfect execution scores in real-world critical tasks.

Problem

• Standard Operating Procedures
• Retrieval-Augmented Generation
• Industrial Knowledge Retrieval
• Multi-view Graph
• Expert Retrieval

Innovation

• Multi-view Graph Experts
• Retrieval-Augmented Generation (RAG)
• Mixture-of-Experts (MoE)
• LLM-Guided Gating
• Procedure Card