Niharika Ramanath, Ranjitkumar Gudder, David Hoffmann and Arndt Lüder (Otto von Guericke University Magdeburg)
Presented at
2026 IEEE 31st International Conference on Emerging Technologies and Factory Automation (ETFA), Västerås, Sweden, 8 to 11 September 2026
Type
Conference paper, IEEE ETFA 2026, 8 pages
Full text
Read on IEEE Xplore (doi:10.1109/ETFA68377.2026.11724844)
Summary
Aligning domain-specific engineering data with AutomationML concepts is still largely manual and error-prone, and the mapping knowledge is lost from one project to the next. The paper presents Mapping Memory, a semantic integration tool developed following the Design Science Research paradigm. It combines Sentence-BERT based attribute matching, rule-based instance classification and a persistent PostgreSQL knowledge base, so that mapping knowledge accumulates across projects, with a human-in-the-loop layer for refinement.
On a synthetic dataset of 408 instances across 24 SystemUnitClasses derived from real engineering vocabularies, the tool reached about 90% matching accuracy with a 60% auto-acceptance rate and reduced manual mapping effort by 25%.
Summary written by AutomationML e.V.; the authors‘ abstract is available on IEEE Xplore. Listed as part of AutomationML at Academic Conferences 2026.
How to Cite
Niharika Ramanath, Ranjitkumar Gudder, David Hoffmann, Arndt Lüder: A Hybrid-AI Semantic Tool for AutomationML with Database Persistence and a Human-in-the-Loop Refinement Layer. In: 2026 IEEE 31st International Conference on Emerging Technologies and Factory Automation (ETFA), Västerås, Sweden, 2026, pp. 1-8. doi: 10.1109/ETFA68377.2026.11724844
BibTeX
@inproceedings{Ramanath2026Hybrid,
author = {Niharika Ramanath and Ranjitkumar Gudder and David Hoffmann and Arndt Lüder},
title = {A Hybrid-AI Semantic Tool for AutomationML with Database Persistence and a Human-in-the-Loop Refinement Layer},
booktitle = {2026 IEEE 31st International Conference on Emerging Technologies and Factory Automation (ETFA)},
address = {Västerås, Sweden},
publisher = {IEEE},
year = {2026},
pages = {1--8},
doi = {10.1109/ETFA68377.2026.11724844}
}