Semantic Interoperability in Federated Digital Twin Systems: A Systematic Review Through the Levels of Conceptual Interoperability Model

Manonmani Sekar; Nasim Nezamoddini

cs.SENov 1, 2026

Semantic Interoperability in Federated Digital Twin Systems: A Systematic Review Through the Levels of Conceptual Interoperability Model

Authors: Manonmani Sekar, Nasim Nezamoddini

Organizations: Department of Industrial and Systems Engineering, Oakland University, Rochester, MI 48309 USA

Abstract

When two companies connect their digital twins, software models of their plants kept up to date with live data, each twin must translate the other's data into its own terms, and that translation rule, called a mapping, can stop being correct when one company changes the meaning of its data without telling the other. This PRISMA 2020 systematic review screened 5,833 records and read 119 full texts to determine whether any published system detects such a broken mapping and repairs it while the twins operate. In all 32 included studies the mapping is fixed before operation, and the two ingredients of run-time repair appear only separately: cross-company matching without any check that the match still holds, and detection and repair of changed meaning within a single company. The review turns this gap into six requirements, each illustrated on one running example, that a system claiming run-time semantic interoperability between independently owned twins should meet.

Figures & tables

Level	Test: the paper shows that…	In the capacity example
L1 Technical	data travel between the twins over a working connection	The assembler receives the supplier’s messages.
L2 Syntactic	the receiver parses the data with a shared format or information model	The assembler reads capacity as a number in a known field of a known model.
L3 Semantic	the receiver interprets terms through a shared or formally mapped vocabulary	The assembler knows that capacity corresponds to its own throughput rate .
L4 Pragmatic	the receiver uses context such as units, time base, or provenance	The assembler knows the value is per eight-hour shift and converts it.
L5 Dynamic	a change of meaning during operation is detected, the mapping revised, and correct exchange restored	The supplier switches to units per hour; the assembler detects the change, replaces the conversion, and its plans become correct again.
L6 Conceptual	the parties reconcile their underlying model assumptions	The two firms agree what counts as available capacity, for instance whether maintenance time is included.

TABLE I: LCIM levels as tests that can be applied to a paper, illustrated on the supplier–assembler example.

Panel of Figure 2 — Fig. 1: PRISMA 2020 flow diagram of study selection [ 11 ] .

Study	Parties exchanging information	Semantic artefact or mechanism	Validation	LCIM
Liu et al. [ 23 ]	Heterogeneous watershed platforms	Ontology; embedding-based matching	Prototype on case data	L3
Akram et al. [ 24 ]	Medical twins at separate institutions	Federated MARL; model updates only	Simulated ICU	L2
Ricci et al. [ 25 ]	Linked twins of many owners	RDF knowledge graph	Illustrative use cases	L3
Xu et al. [ 26 ]	Supply-chain organisations	Agent negotiation protocol	Food supply-chain prototype	L2
Akroyd et al. [ 16 ]	Cross-domain agents, one platform †	Dynamic knowledge graph	Use cases	L3
Elshafei et al. [ 20 ]	Robots of two vendors, one lab †	AAS submodels	Lab experiments	L2

TABLE II: Included studies coded at the highest LCIM level demonstrated.

ID	Requirement	What it demands	In the capacity example
R1	Explicit mapping	The mapping $\phi_{ij}$ is a run-time object that records its target, confidence, provenance, validation status, and the model versions on both sides.	The rule “ capacity $\div 8$ ” is stored with the supplier model version it was written for, so a version change can be seen.
R2	Representable non-translation	A term that cannot be translated safely is marked $\bot$ and is distinguishable from a real value.	While the rule is in doubt, the assembler receives “capacity unknown” rather than a wrong number.
R3	Drift detection	A change in meaning is detected from observed data, without notice from the other party, within the planning horizon.	The assembler notices that reported capacity has jumped relative to delivered quantities.
R4	Repair under partial disclosure	A new mapping is found without either party exposing its whole model, and governance states who may approve it.	The assembler asks the supplier only for the time base of capacity , and the supplier’s answer is logged before the new rule goes live.
R5	Separated credit assignment	A learner can tell a translation error from a policy error, so that ( 2 ) stays well posed under drift.	The scheduler does not retrain its policy to compensate for an eightfold unit error.
R6	Interoperability-specific evaluation	Benchmarks inject unannounced changes of meaning and report how quickly and correctly they are repaired.	The test suite switches capacity to units per hour mid-run and measures time to detect, time to repair, and plans made on wrong data.

TABLE III: Requirements for run-time repair of cross-party mappings, each shown on the capacity example.

Review	Scope	Method	Owners	LCIM	Change
Liu et al. [ 63 ]	DT concepts and applications	Narrative review	No	No	No
Piroumian [ 64 ]	DT interoperability	Position article	Partly	No	No
Hildebrandt et al. [ 65 ]	Data integration for DTs	Syst. review	No	No	No
Listl et al. [ 66 ]	Knowledge graphs in DTs	Syst. review	No	No	No
Huebner and Ihlenfeldt [ 67 ]	Semantic interop. in I4.0	Mapping study	No	No	No
Gebhard and Wang [ 68 ]	Semantic interop. maturity	Maturity model	No	Adjacent	No

TABLE IV: This review compared with earlier ones. “Owners” separates independently owned from single-authority twins; “LCIM” organises findings by interoperability level; “Change” examines what happens when a meaning changes after deployment.

Manonmani Sekar received the B.E. degree in electronics and communication engineering from Madurai Kamaraj University, India, in 1992, and the M.E. degree in embedded systems from Anna University, Chennai, India, in 2014. She is currently working toward the Ph.D. degree in systems engineering with the Department of Industrial and Systems Engineering, Oakland University, Rochester, MI, USA, where she is a research assistant in the Complex Systems Laboratory. Before her doctoral studies she spent more than two decades in engineering practice in manufacturing and telecommunications. Her research interests include semantic interoperability in federated digital twin systems, multi-agent reinforcement learning, and safe coordination of multi-vendor industrial automation.

Table 6

Nasim Nezamoddini is with the Department of Industrial and Systems Engineering, Oakland University, Rochester, MI, USA. Her research interests include multi-agent systems, intelligent manufacturing, and digital twins.

Table 7

Digital Twins

Digital twins are virtual models synchronized with real systems to support monitoring, prediction, simulation, and decision-making.

See all 269 papers

From Brain Models to Executable Digital Twins: Execution Semantics and Neuro-Neuromorphic Systems
Apr 15, 2026Alexandre MuzyDigital Twins
ILLS · International Laboratory on Learning Systems (ILLS) CNRS – McGill – ETS – MILA – Université Paris-Saclay
From Certified Pairs to a Certified Hub: Reasoner-Checked Crosswalks for the Information Exchange Standard
Nov 4, 2026Fabio RovaiData Integration Ontology-Based Reasoning
The Tesseract Academy (Kampakis and Co Ltd), London, United Kingdom
Optimal sequential decision-making for error propagation mitigation in digital twins
Apr 24, 2026Annice Najafi, Shokoufeh MirzaeiMarkov Decision Processes Partially Observable Markov Decision Processes
Department of Industrial and Manufacturing Engineering California State Polytechnic University, Pomona Pomona, CA, 91768, USA

Apr 15, 2026J/K move · Enter open · S save

From Brain Models to Executable Digital Twins: Execution Semantics and Neuro-Neuromorphic Systems

Alexandre Muzy

ILLS · International Laboratory on Learning Systems (ILLS) CNRS – McGill – ETS – MILA – Université Paris-Saclay

Digital Twins

Open paper AskPDF arXiv

Brain digital twins aim to provide faithful, individualized computational representations of brains as dynamical systems, enabling mechanistic understanding and supporting prediction of clinical interventions. Yet current approaches remain fragmented across data pipelines, model classes, temporal scales, and computing platforms, which prevents the preservation of execution semantics across the end-toend workflow. This survey introduces physically constrained executability as a unifying perspective for comparing approaches at the level of execution: whether an execution state is persistent, which events are permitted to update it (simulation, measurement, actuation), and how strongly execution is temporally and causally coupled to neurobiological dynamics. Building on modeling and simulation theory, I propose a taxonomy of execution regimes ranging from isolated offline models to coordinated co-simulation, to continuously executing digital twins sustained by online data assimilation, and ultimately to neuro-neuromorphic physical systems in which biological and computational dynamics are co-executed under shared physical constraints. The executability concept clarifies why accuracy alone is insufficient, and motivates an agenda centered on semantic interoperability, hybrid-time correctness, evaluation protocols, scalable reproducible workflows, and safe closed-loop validation. This survey adopts a systems and runtime-oriented perspective, enabling comparison of heterogeneous approaches based on their execution semantics rather than on model form or application domain alone.

Abstract

Digital Twins

Explore similar work

From Brain Models to Executable Digital Twins: Execution Semantics and Neuro-Neuromorphic Systems

From Certified Pairs to a Certified Hub: Reasoner-Checked Crosswalks for the Information Exchange Standard

Optimal sequential decision-making for error propagation mitigation in digital twins