SAP researchers release Plant Memory system for maintenance work-order analysis
Plant Memory is a system designed to process and analyze maintenance work orders from SAP Plant Maintenance (PM) exports. The system uses technical language processing (TLP) to extract structured data from unstructured, often abbreviated or misspelled text in work orders. It builds an ISO 14224-aligned maintenance ontology where each extracted statement is a verifiable fact with a confidence…
Key points
- Plant Memory extracts structured data from SAP PM work orders using rule-based TLP and optional LLM review
- System builds an ISO 14224 ontology with verifiable facts, confidence scores, and technician approval workflows
- On **400 synthetic work orders**, achieved **96.2% component accuracy** and **82.1% diagnostic question correctness**
The system combines rule-based extraction with an optional LLM as a secondary reviewer. Technicians verify facts via a review queue, address knowledge gaps in repeated failures, and approve or reject new terms. Results are accessible via a web app for humans and an Model Context Protocol (MCP) server for AI agents, ensuring consistent, citable answers. On a synthetic dataset of 400 work orders from a parboiled-rice plant, rule-based extraction achieved 96.2% component accuracy, 88.9% failure-mode accuracy, and 100% root-cause precision at 62% coverage. For 28 diagnostic questions, the ontology’s ranked root causes matched the correct answer at 82.1%—double the 53.6% accuracy of raw SAP root-cause data.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- OpenAI offers U.S. university labs free ChatGPT Business seats with Rosalind access · 2 src
- Anthropic reports small AI models may self-replicate on IoT devices within months · 29 src
- Researchers propose LLM-referenced C-test for evaluating supra-intelligence · 8 src
- Researchers study permutation diffusion LLMs for recommendation reranking · 1 src
- Researchers compare Non-Generative verifier to Astra LLM for RAG claim checks · 1 src
Comments
via GitHub Discussions