Materials are either enabler or bottleneck for the vast majority of technological innovations. The digitization of materials and processes is mandatory to create live production environments which represent physical entities and their aggregations and thus allow to represent, share, and understand materials changes. However, a common standard formalization for materials knowledge in the form of taxonomies, ontologies, or knowledge graphs has not been achieved yet. This paper sketches the e orts in modelling an ontology prototype to describe Materials Science experiments. It describes what is expected from the ontology by introducing a use case where a process chain driven by the ontology enables the curation and understanding of experiments.
The discipline Materials Science and Engineering (MSE) promises solutions to modern societal challenges, including climate change and resource scarcity. However, the complexity of the lifecycles of materials and their diversity poses several challenges in the management of materials' knowledge for a comprehensive sharing and understanding among various MSE disciplines.
Many experiments are conducted to study materials' behavior, which
generates a variety of data, describing manufacturing process settings, material
properties, material structures, and further MSE parameters. The sharing and
interoperability of MSE ndings are mainly achieved through the exchange of not
standard and often not well-documented les [
Existing attempts try to represent top-level knowledge about materials
properties and structures [
Imagine having a process chain. An object undergoes processes that transform the object's material structures and material properties (Figure 1). In this scenario, it is crucial to track how processes are performed and how objects change when describing material transformations. In detail, transformative processes (e.g., manufacturing processes) lead to changes in objects' status (i.e., material properties and material structures) according to their individual process parameters, thus creating new object entities as an output. Figure 2 shows a high-level
7 https://emmc.eu/, accessed on June 10, 2021 8 https://github.com/emmo-repo/EMMO, accessed on June 10, 2021 sketch that represents the main top classes and object properties of the ontology under development. Every process comes with its own parameters that represent all the required variables. Processes that can transform material structures are represented by the class pmd:ManufacturingProcess, processes which perform analysis are represented by the class pmd:AnalysisProcess; however, they can still transform the object (e.g., a Tensile Test process), and therefore, they might also be transformative. Process parameters might have various e ects on the materials and, therefore, the relations between processes to material structures are required. This can also be done at di erent granularity levels (e.g., nano, microscopic, mesoscopic, etc.) depending on the requirements of the application scenario. For example, the problem of locally heterogeneous materials in an object requires separating the object into a certain number of volumetric sub-areas made by the same material (pmd:Material), so called voxels (pmd:Voxel). Therefore, the voxel becomes the object of the ontology and describes the conditions of the material with its individually experienced process parameters. Object properties are de ned to describe how pmd:Process modi es pmd:MaterialStructure a ecting pmd:MaterialProperty, thus allowing machine and humans to understand what is performed in the experiment. 4
MSE ontologies will enable MSE scientists to curate, describe, share, and optimize experiments. An application example is given in Figure 3. There are 2 processes: c0 and h0 instances of pmd:Cutting and pmd:Heating, respectively.
Fig. 2: Visualization of the main elements that constitute the ontology. There are 3 pmd:Object namely o0; o1; o2; o1 is originated from o0 and o2 is originated from o1. o0 has a pmd:Geometry g0 and a pmd:Microstructure m0. In the example, pmd:Cutting class represents cutting processes that are transformative for the geometry, pmd:Heating processes that are transformative for the microstructure. When o0 undergoes c0, it is transformed in o1; since c0 does not transform the microstructure, it is preserved in o1 (edge a). However, o1 will have a di erent geometry i.e., g1. When o1 undergoes h0, it is transformed in o2. In this case, the geometry is not transformed and, therefore, o2 has the same geometry of o1 i.e., g1 (edge b). Thus, the preservation of material structures and material properties can be de ned by means of description logics and Semantic Web Rule Language (SWRL) rules, which enable automatic reasoning on experimental data, e.g., to nd inconsistencies. For example, an object with a di erent microstructure after a cutting process raises an inconsistency. At the same time, this semantics helps to create connections between objects involved in a process chain, thus enabling reasoning on the process-object relations (e.g., an object after a cutting process preserves its microstructure). The reader can nd the prototype ontology as well as the SWRL rules used by this practical application in GitHub9.
The presented prototype provides a modelling philosophy to describe process chains. Furthermore, it shows how Materials Science semantics can be injected into a formal model to represent material tranformations. The prototype addresses the RQs by providing the following insights.
RQ1. The ontology prototype includes the rst e orts to represent core elements (pmd:Process, pmd:Object, pmd:MaterialStructure, pmd:MaterialProperty,
etc.) and their relationships that a MSE ontology must represent in order to describe MSE process chains.
RQ2. The semantics within the model shows how data consistency can be obtained. More speci cally, this is shown by propagating material structures and material properties through SWRL rules. If a user injects erroneous data (e.g., a di erent microstructure for an object after a cutting process), the reasoner is able to spot the introduced error.
RQ3. The prototype ontology provides the semantics to reason on experiments, and links objects to material structures and material properties that are not directly generated by the experiment itself, and thus supports MSE experts in fully representing and understanding material transformations. 6
This paper introduces the vision to model process chains and MSE experiments
through an ontology with the long term goal of studying material
transformations. It provides three RQs that are driving the research and shows an ontology
prototype. Perspectively, the data modelled by speci c use case ontologies will
enable the curation and preservation of data as well as the possibility to
interpret various outcomes. These ontologies are being developed within the
Plattform MaterialDigital. They will enable a substantial step towards the provision
of ndable, accessible, interoperable, and reusable [
Acknowledgements The authors thank the German Federal Ministry of Education and Research (BMBF) for nancial support through the project InnovationPlatform MaterialDigital (FKZ no: 13XP5094A and 13XP5094B and 13XP5094D).