The rapidly expanding market for composite materials, driven by applications in automotive, aerospace, and renewable energy, requires a comprehensive framework to describe fiber-reinforced materials. Although versatile, these materials lack standardized data representation, hindering cross-scale integration from material characterization to manufacturing. The OntOMat project aims to bridge this gap by developing a novel ontological framework grounded in FAIR data principles. Through ontology engineering, it enables the collection, linkage, and analysis of data across various processes, focusing on fiber-reinforced polymers. This approach allows for the optimization of material microstructures for enhanced stability and recyclability.
The market for composite materials is growing at double digit percentages annually [
The main goal of the OntOMat project is to develop a novel ontological framework that allows the implementation of FAIR data principles using ontology engineering. This should allow us to collect, link, and analyze data arising from material data sheets, but also from characterization, production, and simulation processes which are collected for a variety of material systems. In terms of material systems, the focus is on fiber-reinforced polymers. The linking of data and parameter arising from these
CEUR Workshop
ISSN1613-0073 processes should then allow us to compare and optimize a material’s microstructure for operational stability and recyclability. The scope can be directly derived from the project goal and includes the creation of an ontology for (a) fiber-reinforced materials and their characteristics, (b) the manufacturing and characterization processes of these materials, and (c) the simulation processes to determine certain characteristics or behavior. In this paper, we first present our ontology development methodology (Section 3), then give the details of the OntOMat ontology (Section 4) and discuss limitations and future work (Section 5).
For an extensive survey of ontologies in materials science and engineering, we refer to [3] and give more details on the relevant top- and midlevel ontologies in Section 3. From the domain-level ontologies presented in [3], the EMMO General Process Ontology (GPO), the Semantic Materials, Manufacturing and Design1 (SEMMD) ontology, the Additive Manufacturing Ontology (AMONTOLOGY), Ontology for Simulation, Modelling, and Optimization (OSMO), Dislocation Simulation and Model Ontology (DSIM), Atomistic Simulation Methods Ontology (ASMO), Metadata4Ing Ontology (M4I), and Material properties ontology (MAT) covers the same domains as our work. Since the OntOMat project is part of Platform MaterialDigital2 that provides a top-level ontology with the BFO[4] and a mid-level ontology with the PMD Core Ontology3 (PMDCo), we focus only on ontologies that align with the BFO or PMDCo, since alignments with other top-level ontologies are beyond the scope of this work. The SEMMD ontology is concerned with the same focus and is already BFO aligned, but any documentation regarding its use is missing. Ontologies that describe either composite materials or VDI/VDE 3682 Formal Process Language (VDI 3682 language) are harder to find. Polymer membrane research and related laboratory experiments are described in the PolyMat ontology [5], but the ontology does not seem expressive enough to capture, e.g., complex simulations. An initial ontology for the VDI 3682 language was introduced in the work of [6], which we considered in our ontology design. However, the work of [6] missed certain design elements, such as process decompositions and alignment with any top-level ontology.
We apply the “Simple Knowledge Engineering Methodology” of [7] to develop our ontology. It ofers good guidelines for an agile modeling process. For brevity, we recapture the main steps introduced by the authors of [7] and give more details in the following subsections: 1. Domain and scope of the ontology that includes competency questions for narrowing the scope; 2. Reuse of existing ontologies that includes an alignment with mid- or top-level ontologies; 3. Collect important terms and define classes and their hierarchies in the ontology, which can be addressed by top-down, bottom-up or combined approach; 4. Define properties of classes including “slots” such as cardinality and domain/range; 5. Define instances, which includes generic instances that should be available for every user, but also includes illustrating examples.
Note that above steps are a non-linear process of incremental interactions (in particular steps 3. to 5.). Hence, the terms for a specific topic are collected by domain experts, mainly material scientists but also simulation engineers, stored in an intermediate source, which then builds the base to define classes and properties. The authors then present the results including class and property hierarchies to the aforementioned domain experts for evaluation.
The first step of [ 7] relates to the collection of more than 50 competency questions with the experts of the OntOMat partners. They are categorized by “material testing/characterization”, “manufacturing process and equipment”, “physical product traceability”, “material consumption”, “product life cycle and design”, and “material testing/characterization”. In the following, we present a list of exemplary competency questions (CQs) and expected exemplary answers:
Competency question Expected answer ebWrethyui?cshedchfoarraactceerriztaaitnionmmateetrhiaoldpsrcoapn- cqCmhueaaartrtneaarticiinaftyelm.,rieza.atget.ir,oiatnhletperfibcoheprneirvqtoiueleussmc.aeAncbCoenTqteusncatannotiffiecdaanbcyboemspupesocesidificteto
erties and values as well as defined geometry to be modeled aWshpaetcpificarsaimmueltaetrisoanr?e required to run ccaohlonanndgiictwiso,intahsrenthneeeedepdhteoydsb.iecIsndteaofindbeddeitaaionnnad,lytthhzeeedgl,oeeao.dgms.,eastntrryducnbtoeueurdnasldmtaoreybe discretized in order to numericial solve the diferential equation.
How can data from the characteriza- Test results can be reduced to efective parameters, e.g., tion be implemented into the simula- the modulus of elasticity, which is then used in a suitable tion model? numerical model.
Durability means change of material properties values over What contributes to product durabil- time. Durability therefore depends on the operational conity? ditions applied as well as the impact on specific material properties.
There is no clear relation between material class and simuWhich simulations can be used for a lation technique, but there is a relation between simulation certain material system? domain and physics to be studied for a certain material class. onWrehesdiametdulfcaohtriaoarnas?cpteecriizficamtioatneriadlamtaodaerle ifcenThrehtdayer.amvTcahatletueerepirszia.aarltaimmonoetdteeerclvhraenlqiuqueuisreecssatnchebarettaqrieuntarpnieatvrifaeymdmfertoaemtresrstipaolebcpeirficodpe-sWpehcaitficsmubatsetarinaclecslaasrse? contained in a aAmsefibtnhetre.-rmeaintrfioxrcaenddecpaorxbyonr efibseinrsh(a7s78e2p-o4x2y-5r)eassinth(6e1r7e8in8-f9o7r-c4e)eWrthieast faorer athsepreeclieficvmanattemriaatlecrliaaslsp?rop- tnThueesmtmerroeisdcuuallltumsscooadfneelb.laestriecdituyc,ewdhtiocheifesctthiveenpuasreadmientearssu,iet.agb.,le cSmohanonwtuenftahtceotuinmrimpnagactpteroroifacrleespcsyreocslp?eedrtmieastearniadl tscCeiogornnimasitlficpaaacnnortetn.ldytetdnoetv,vieairvtgeeinnfoirmfmtahtaeetreimraialaslt,setcrhioaenltpacroinomipnpegrotrsyeitcivyoacnllueidessmkceaap-nt Based on the competency questions, we can derive important concepts, including material properties and compositions, as well as a model for simulation, manufacturing, and characterization processes, including process parameters and results, which play a crucial role for answering the CQs. We also provide the full list of CQs in our project’s Gitlab repository.
The second step of [7] is the alignment with a mid-/top-level ontology. Ontology reuse is along a pyramid, where the top-level has the highest abstraction level and are usually called upper ontologies. Examples are Basic Formal Ontology (BFO) [4] and Industrial Data Ontology (IDO) [8] with a focus on industrial data. Below are mid-level ontologies that describe a specific technical domain such as Quantities, Units, Dimensions and Data Types Ontologies (QUDT)4 or PROV Ontology (PROV-O) [9] which are often orthogonal to upper ontologies. Domain ontologies, such as Elementary Multiperspective Material Ontology (EMMO) [10]5, describe a specific scientific or industrial domain, where the OntOMat ontology would be one particular domain.
As the OntOMat project is situated at the intersection of industrial data/processes and material science, the IDO ontology is a natural starting point for modeling, since simulation and manufacturing processes are an important scope of the vocabulary. Another option would have been the PMDCo 1.0 ontology based on PROV-O, which provides some useful top-level concepts, such as Activity, but it lacks the coverage and standards compliance to known top-level ontologies such as BFO. With the publication of the PMDCo 3.0 ontology that is based on the BFO the coverage and standards compliance is given, and in the discussion we will outline the path towards a tighter integration with the PDMCore 3.0.
For the OntOMat ontology in most cases, the IDO classes and properties build the highest level of class hierarchies, whereby the classes lis:PhysicalArtefact, lis:PhysicalObject, lis:Quality, lis:Activity, and lis:InformationObject are our anchor points. An important characteristic of IDO modeling is the distinction of any object (besides information) between lis:Specified, i.e., any abstract representation of an object such as a definition or simulation model, and lis:Actual, i.e., objects that exist or existed at one point.
The third step of [7] is the collection of important terms that lead to classes and related properties. Initially, we manually collected terms based on standards and guidelines with the support of domain experts. For example, experts suggested a material classification according to the IEC 62474 standard [11] and a formalized process description (FPD) according to the VDI/VDE 3682 guidelines [12] as important sources. For material qualities, the domain experts suggested a list of 150 relevant qualities based on literature research, which we extended with units and symbolic representations. Material classes and qualities were collected as term lists and taxonomies in newly created spreadsheets, while processes, states, and operators of the FPD were directly extracted from the text-based guidelines. With the fourth step, we bring all the pieces together, where we created an initial version of the ontology formulated in OWL 2 DL6. We decided on a modular approach that creates a separation between terms related to material products, to material and properties classes, and to processes that include simulations. Each module was validated by our domain experts for content-related correctness, which led to several rounds of improvements. The ontology described in the next section represents the result after the completion of the fourth step.
The OntOMat ontology (OMO) is conceptually split into three tiers, where the top level is currently IDO, the middle level is a handcrafted structure with the support of domain expert, and the lowest level is automatically generated based on two industry standards concerning material classes and process descriptions. An important part are material classes and material qualities, i.e., material properties, which are then used to define material products based on these material classes. Another important
part describes the characterization, manufacturing, and simulation processes from an “a posterior” perspective, thus the process path (e.g., based on observations stored in logs) of the respective process is captured, and (currently) not the process plan modeled “a priori”. Figure 1 shows the three tiers, where the color of the box represents thematic modularization, the white arrows represent the subclass relations, and the black arrows define the domain / range of object properties. The ontology is published on the MaterialDigital Github repository under https://github.com/materialdigital/ontomat-ontology.
With OMO, we introduce a duality between “real” physical products represented by PhysicalProduct and abstract material classes simply called Material. A physical product is made of one or more materials represented by material classes and is connected by the hasMaterial object property to a physical product.7 Measurable material properties are connected by hasMaterialQuality with the domain lis:PhysicalObject (thus any physical object can have material properties) and the range of IntensivePhysicalProperty and subclasses that define types of material properties. Specific properties can be “real” properties of a physical product, which are measured as values of the designated specimen, or “abstract” properties of the material obtained from a product datasheet / publications. This duality allows us to manage the known material properties for material classes and their values only once and reuse them for every specimen. For example, the density of the material (class) gold is approximately 19.3 g/cm3, but the density of a specific gold bullion is measured as 19.2 g/cm 3 due to some contamination.
The hierarchy of material classes starts with the Material class with a fixed set of subclasses derived from the IEC 62474:2018 standard taxonomy. The left nodes in the IEC taxonomy are automatically generated based on the standard spreadsheet shown in Figure 2a.8 However, the IEC 62474:2018 standard misses some concepts to describe fiber-reinforced materials: 7hasMaterial is a subproperty of IDO’s lis:hasMaterialPart, which has the domain lis:Compound. We extended the domain to include Material, since the ontology should also cover composites. 8https://std.iec.ch/iec62474/iec62474.nsf/Index?open&q=161936 (a) Material classes (b) Material properties
• Classes representing fiber-reinforced composite materials are missing, only filled composites are mentioned; • Only one class for gases and liquids is defined, whereby a clear separation between mixtures and pure gases (similarly for liquids) is missing; • Thermosetting polymers, such as melamine formaldehyde (MF) and vinyl ester (VE) are not considered; • The CAS registry number is missing, which is used to identify the most prominent chemical substance in a material.
With the support of our domain experts, we extended and refined the IEC 62474:2018 taxonomy, thus introducing the paired classes for gas (pure / mixture), liquids (pure / mixture) and solid materials (homogeneous / heterogeneous), as well as dropping all unspecified classes such as “other unfilled thermoplastics”, which are covered by their superclasses. For example, we introduced the new classes SolidHomogeneousMaterial and SolidHeterogeneousMaterial, where from the latter we derived the subclass CompositeMaterial. This subclass is central to OMO and acts as the domain for the following object properties that define possible composite structures: • hasMatrixWith denotes the matrix material, e.g., epoxy resin; • reinforcedWith denotes the material that is used for reinforcement, e.g., fibers; • filledWith denotes the material that is used as a a filler, e.g., air.
Note that for any of the above properties, we also added the transitive closure using a transitive property, since a material structure can have several levels of nesting. For example, we added the property hasMatrixWithTransitive for the hasMatrixWith. Similarly to the material class hierarchy, our experts conducted an extensive literature analysis to collect suitable property classes that were recorded in a reference spreadsheet. The spreadsheet defines the QUDT unit, a commonly used symbol, including aligned index usages, e.g., 1, with respect to direction and symmetry class, e.g., isotropic. An excerpt of the reference spreadsheet is given in Figure 2b and again the respective subclasses such as ElasticModulusE1 for “elastic modulus E1” are generated and aligned to the mid-tier, e.g., ElasticProperty, automatically.
An objective of the OntOMat project is to provide an automatic classification of composite materials according to the structure of these materials. As shown in Figure 3, the classification of composite materials resembles the overlap of two hierarchies starting with CompositeMaterial, namely a matrixbased hierarchy (starting with MatrixBasedCompositeMaterial) and a reinforcement-based hierarchy starting with ReinforcedCompositeMaterial. Both hierarchies are joined again on the leaves, where the outcomes of the classification are added. For example, the leave class FiberReinforcedEpoxyResin is sub-class of PolymerMatrixCompositeMaterial and also of FiberCompositeMaterial. Having a wide range of possible materials that can act as a matrix (examples of suitable polymer classes are shown in Figure 2a) and act as reinforcements or fillings (such as Carbon), we introduced a classification system based on equivalence axioms. Our intuition is the following; on left-hand side of the equivalence is the class for the material system, for instance FiberReinforcedEpoxyResin, and on the right-hand side is an intersection of CompositeMaterial, a existential re- Figure 3: Partial rendering of the composite material classtriction on hasMatrixWithTransitive, sification. in this case, EpoxyResin, and another existential restriction on reinforcedWithTransitive on Material. The generic Material class is used, since any material class could act as a reinforcement. The resulting equivalence axiom: FiberReinforcedEpoxyResin ≡ (CompositeMaterial ⊓ ∃hasMatrixWith.EpoxyResin ⊓ ∃reinforcedWith.Material), with ( hasMatrixWith) and ( reinforcedWith) is written in RDF 1.1 Turtle9 notation as: 1 ontomat : F i b e r R e i n f o r c e d E p o x y R e s i n owl : e q u i v a l e n t C l a s s 2 [ owl : i n t e r s e c t i o n O f ( 3 ontomat : C o m p o s i t e M a t e r i a l 4 [ r d f : type owl : R e s t r i c t i o n ; 5 owl : onProperty ontomat : h a s M a t r i x W i t h T r a n s i t i v e ; 6 owl : someValuesFrom ontomat : EpoxyResin 7 8 9 10 11 ] 12 ) ; 13 r d f : type owl : C l a s s 14 ] ; 15 dcterms : i d e n t i f i e r ”O−402” ; 16 r d f s : l a b e l ” F i b e r − r e i n f o r c e d EpoxyResin ( EP ) ” .
] [ r d f : type owl : R e s t r i c t i o n ; owl : onProperty ontomat : r e i n f o r c e d W i t h T r a n s i t i v e ; owl : someValuesFrom ontomat : M a t e r i a l
Listing 1: Example of an equivalence axiom with the OMO prefix ontomat:.
Taking the axiom from the listing above, reinforcedWithTransitive is replaced with filledWithTransitive to define filled composite material classes. The Material class could be replaced with Carbon to define more specialized systems such as CarbonFiberReinforcedEpoxyResin. Note that these axioms are automatically generated based on the spreadsheets shown in Figure 2a, where the “ClassName” column is parsed to extract the type of reinforcement and the matrix class.
Despite several standards that are used to model processes, e.g., business process modeling and notation (BPMN),10 no language captures all the aspects of technical processes that occur in engineering / industry applications, including their digital representation. The authors of the VDI/VDE 3682 Formal Process Language (VDI 3682 language) outline the following aspects that should be captured by a process language [12] used in engineering / industry applications:
(a) An single process step and its decomposition (b) The eight graphical symbols
• Modeling the life cycle of a technical systems, including its digital representation (usually called the digital twin); • Applicable to all kinds of processes including technical / non-technical, continuous / batch processes in all fields of technical applications; • Simple, neutral regarding the industry, and easily understandable; • Containing all information necessary for engineering and normal operation based on modern IT tooling and methods;
The resulting VDI 3682 language is a visual language based on a predefined set of eight symbols (shown in Figure 4b) that include three states, a system limit (dashed frame), a process operator (green rectangle) and a technical resource (gray rectangle). The three kinds of states are product (red circle), energy (blue, rhombus), information (blue hexagon), which act as input and output of a process operator and are required to evaluate the operator. As shown in Figure 4a left, a process flow is always an alteration between states and process operators with reference to the technical resources used in that operator. In addition, the modeler has the choice to define parallel (the default) or alternatively running (similar to an XOR condition) process flows. A powerful feature of the VDI 3682 language is the ability to decompose a process operator into a new subprocess (shown in Figure 4a right), thus introducing modularity similar to functions in programming languages. The VDI 3682 guideline also introduces an UML-based data model and XML-based rendering of the visual language. We believe that the UML-based data model is a helpful sketch but is ambiguous regarding its serialization. In the second part of the guideline, the authors introduced an XML-based representation, which is suficient for simple processes. However, this representation is underspecified and does not capture the full complexity of VDI 3682 process flows, for example, cyclic flows or subprocesses. Similarly to the authors of [ 6], we suggest that an ontology-based representation for the VDI 3682 symbols and workflows, as well as an RDF-based representation of (executed) workflows, are the most promising methods to capture it. Furthermore, material states can be defined more fine-grained by material classes, and similarly information states can be modeled by document and process parameter classes.
The ontology represented by [6] does not reuse any top-level ontology. Thus, we anchor our representation of the VDI 3682 language in IDO, where processes and operators are subclasses of lis:Activity, information states are directly mapped to lis:InformationObject, and production assets are subclasses of lis:PhysicalArtefact. As shown in Figure 5, the class VDIProcess is directly derived and includes subclasses such as ManufacturingProcess and SimulationProcess. Similarly the class VDIProcessOperator is directly derived with the subclasses MoldingOperator, and SimulationRun. A production asset is captured by VDIProductionAsset with subclasses such as Workstation and SimulationTool. A larger challenge is to capture the process flows and alteration of the input/output states. A complete process workflow is called VDIProcess and is a wrapper that includes the system limit, which has one or more hasOperator properties pointing to its initial VDIProcessOperators. The properties next (and the inverse previous) define the next (and previous) evaluation order of a specific operator. The three states are linked to a VDIProcessOperator by the following object properties, whereas at least one state is required to be linked: • product state: by hasInputProduct and hasOutputProduct to PhysicalProduct thus connecting products and their related materials to processes; • information state: by hasInputInformation and hasOutputInformation linking to VDIInformationObject, which has a property source to underlying documents and can combine several process parameter or result items by the property contains; • energy state: by hasInputStream and hasOutputStream to VDIEnergyStream class; • technical resource: by producedBy with the domain ProductionAsset.
Our representation does not fully align with the VDI 3682 language, as some definitions seem redundant. We have not introduced any concept for the system border, since the VDIProcess represents the border. For example, any state connected directly to VDIProcess, e.g., by hasInputProduct, is the interface to the exterior of a process. Note that the VDI 3682 language does not have a next and previous relation to represent an explicit process flow. Hence, we introduced the next and previous object properties, since they should simplify the query of explicit process flows using property paths. The next property can be deduced by the property chain: hasIP− ∘ hasOP ⊑ next, where hasIP, hasOP and next represents hasInputProduct(y,z), hasOutputProduct(x,z), and next(x,y).
To illustrate the use of the OMO, we give a simple example of a knowledge base (KB) that describes a test specimen used in the OntOMat project. A test sample is produced by resin transfer molding (RTM), where a liquid epoxy resin is injected under pressure into a closed mold containing dry glass ifber, which is then cured to form a composite part, i.e., the resulting product.
First, we describe the material and its composition, and add the material quality density and its value measured in a material characterization process: 1 : epoxy −M_LY556_GFKUD0_01 r d f : type ontomat : C o m p o s i t e M a t e r i a l ; 2 r d f s : l a b e l ” Siemens Epoxy GFK UD” ; 3 ontomat : manufacturedBy ” Siemens Technology ” ; 4 ontomat : hasMatrixWith : a r a l d i t e _ L Y _ 5 5 6 ;
Second, we outline the RTM process and a single process operator according to the VDI 3682 standard.
Note that the RTM process has nine operators, but we only show the first one for brevity: 1 : epoxy −M_LY556_GFKUD0− M a t e r i a l F o r m i n g −Main r d f : t y p e ontomat : M a n u f a c t u r i n g P r o c e s s ; 2 r d f s : l a b e l ” Epoxy GFK UD P r o c e s s ” ; 3 ontomat : h a s O p e r a t o r : epoxy −M_LY556_GFKUD0− M a t e r i a l F o r m i n g −PO01 , : epoxy −M_LY556_GFKUD0−
M a t e r i a l F o r m i n g −PO05 ; ontomat : h a s I n p u t P r o d u c t : a r a l d i t e _ L Y _ 5 5 6 − product −1 , : a c r y s t a l _ L _ 4 0 4 0 _ B B − product −1 ; ontomat : h a s O u t p u t P r o d u c t : epoxy −M_LY556_GFKUD0_01− f i n a l − p a r t . 4 5 6 7 : epoxy −M_LY556_GFKUD0− M a t e r i a l F o r m i n g −PO01 r d f : t y p e ontomat : V D I P r o c e s s O p e r a t o r ; 8 r d f s : l a b e l ” Epoxy GFK UD p r o c e s s f o r c o m p r e s s i n g ” ; 9 ontomat : h a s I n p u t P r o d u c t : a r a l d i t e _ L Y _ 5 5 6 − product −1 ; 10 ontomat : h a s O u t p u t P r o d u c t : a r a l d i t e _ L Y _ 5 5 6 − product −2 ; 11 ontomat : producedBy : p r e s s u r e _ v e s s e l − a s s e t −01 ; 12 ontomat : n e x t : epoxy −M_LY556_GFKUD0− M a t e r i a l F o r m i n g −PO05 .
Third, we modeled a generic simulation process according to the VDI 3682 standard. This process model can capture several types of simulations, such as structural mechanics simulations, using ifnite element analysis software. As illustrated in Figure 6, a generic simulation process can be described as a preprocessing, a simulation (that is, running the simulation), and a postprocessing operator. A material instance (e.g., a specific ifber-reinforced epoxy resin), a geometry (e.g., the shape based on a given standard) and a set of input parameters are the input states, while the simulation results (e.g., additional material qualities), information (e.g., logs), and a simulated specimen (of a material) are the output states of the simulation process. Note that we deliberately modeled a material instance as a (specified) physical object instead of an information object. Furthermore, this process model does not sufice to describe multi-scale simulation processes, which could be seen as a parent process with the above simulation process as one of several compositions. Figure 6: A generic model of a simulation process.
From this simple example, one can see that it is not feasible to create and maintain instances for more complex processes manually. Thus, the next step in the OntOMat project is to provide tools for data onboarding of VDI 3682 process models created with the FPB.JS tool11 of Fay et al. [13]. Additional sources such as input parameter, source, and log files that describe the boundary conditions or process parameters of a specific simulation run can be used to enrich the imported models.
In this paper, we outline the ontology development process in the OntOMat project and introduce the main elements of the resulting OntOMat ontology. Describing base material classes and material properties is as expected, but modeling and import of the IEC 62474:2018 standard, as well as the related equality axiom-based classification of composite materials, based on their base materials, is a novel approach to material classification. At the core of the OntOMat ontology is the representation of the VDI 3682 (Formal Process Language) standard, which is a visual language to describe any kind of real and digital processes. Due to the complexity of the VDI 3682 standard, it is not trivial to find an appropriate semantic representation of it. However, this should allow us to describe complex multiscale simulations, integrate the results of these simulations, and align them with underlying material systems.
We recognize that the current ontology has limitations and still needs to be extensively evaluated with real characterization, simulation, and manufacturing processes. One limitation is that we do not fully support parallel process flows, which require additional concepts such as synchronization points. The axioms for automatically deducing the object properties next / previous and the automatic classification that the processing states are in the limit of the system are yet missing. Furthermore, technical resources are coarse-grained modeled, and a taxonomy based on the ISA-9512 standard would be valuable.
The future work focuses on several directions: (a) finalizing the ontology addressing the above limitations and a complementary alignment to the PMD Core Ontology; (b) providing an extensive tool set for injecting data for specific material systems based on input spreadsheets; (c) supporting the integration of existing VDI 3682 visual process modeling tools; (d) in addition to an existing SPARQL interface, also provide a GenAI-based copilot to interact with the knowledge base combining the ontology with instances for each material system.
During the preparation of this work, the authors used Writefull for Overleaf13 to check grammar, spell check, and paraphrase to improve the readability of the text. The authors reviewed and edited every suggested content improvement and assume full responsibility for the content of the publication.
This work was supported by the OntOMat project (no. 20) supported by the MaterialDigital initiative of the Bundesministerium für Bildung und Forschung. 11https://github.com/HamiedNabizada/FPB.JS 12https://www.isa.org/standards-and-publications/isa-standards/isa-95-standard 13https://www.writefull.com/writefull-for-overleaf
O. Dumon, S. Edmunds, C. Evelo, R. Finkers, B. Mons, The FAIR guiding principles for scientific data management and stewardship, Scientific Data 3 (2016). doi: 10.1038/sdata.2016.18. [3] E. Norouzi, J. Waitelonis, H. Sack, The landscape of ontologies in materials science and engineering: a survey and evaluation, in: Proceedings of the First International Workshop on Semantic Materials Science (SeMatS 2024): Harnessing the Power of Semantic Web Technologies in Materials Science, volume 3760 of CEUR Workshop Proceedings, CEUR-WS.org, 2024, pp. 78–100. URL: https://ceur-ws.org/Vol-3760/paper3.pdf. [4] J. N. Otte, J. Beverley, A. Ruttenberg, S. Borgo, A. Galton, O. Kutz, Bfo: Basic formal ontology1,
Appl. Ontol. 17 (2022) 17–43. doi:10.3233/AO- 220262. [5] M. Dembska, M. Held, S. Schindler, Polymat - bringing semantics to polymer membrane research, in: A. Valdestilhas, H. Li, P. Lambrix, H. Sack (Eds.), 1st International Workshop on Semantic Materials Science: Harnessing the Power of Semantic Web Technologies in Materials Science, SeMatS 2024, CEUR Workshop Proceedings, CEUR-WS.org, 2024. URL: https://elib.dlr.de/206770/. [6] T. Jeleniewski, H. Nabizada, J. Reif, A. Köcher, A. Fay, A semantic model to express process parameters and their interdependencies in manufacturing, in: 2023 IEEE 32nd International Symposium on Industrial Electronics (ISIE), 2023, pp. 1–6. doi:10.1109/ISIE51358.2023.10228021. [7] N. Noy, D. McGuinness, et al., Ontology development 101: A guide to creating your first ontology, 2001. URL: http://www.ksl.stanford.edu/people/dlm/papers/ ontology-tutorial-noy-mcguinness-abstract.html. [8] I. O. for Standardization, Part 3: Industrial data ontology, Automation systems and integration — Ontology based interoperability (ISO/DIS 23726-3) (2025). URL: https://www.iso.org/standard/ 87560.html. [9] K. Belhajjame, J. Cheney, D. Corsar, D. Garijo, S. Soiland-Reyes, S. Zednik, J. Zhao, PROV-O: The
PROV Ontology, Technical Report, 2012. URL: http://www.w3.org/TR/prov-o/. [10] P. D. Nostro, J. Friis, E. Ghedini, G. Goldbeck, O. Holtz, O. Roscioni, F. Zaccarini, D. Toti, Elementary multiperspective material ontology: Leveraging perspectives via a showcase of emmo-based domain and application ontologies, in: Proceedings of the 16th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management - Volume 2: KEOD, INSTICC, SciTePress, 2024, pp. 135–142. doi:10.5220/0012910200003838. [11] I. E. Commission, Iec 62474:2018 - material declaration for products of and for the electrotechnical industry, IEC Norms (2018). URL: https://tc111.iec.ch/tc-activity/material-declaration/. [12] V. D. INGENIEURE, VDI/VDE 3682:2: Formalised process descriptions - information model Blatt 1, VDI/VDE-RICHTLINIEN (2015). URL: https://www.vdi.de/en/home/vdi-standards/details/ vdivde-3682-blatt-2-formalisierte-prozessbeschreibungen-informationsmodell. [13] A. Fay, X. L. Hoang, C. Diedrich, M. Dubovy, C. Eck, C. Hildebrandt, A. Scholz, T. Schröder, R. Wiegand, Abschlussbericht – SemAnz40. Vorhandene Standards als semantische Basis für die Anwendung von Industrie 4.0., 2018.