Ontologies are widely used in materials science to describe experiments, processes, material properties, and experimental and computational workflows. Numerous online platforms are available for accessing and sharing ontologies in Materials Science and Engineering (MSE). Additionally, several surveys of these ontologies have been conducted. However, these studies often lack comprehensive analysis and quality control metrics. This paper provides an overview of ontologies used in Materials Science and Engineering to assist domain experts in selecting the most suitable ontology for a given purpose. Sixty selected ontologies are analyzed and compared based on the requirements outlined in this paper. Statistical data on ontology reuse and key metrics are also presented. The evaluation results provide valuable insights into the strengths and weaknesses of the investigated MSE ontologies. This enables domain experts to select suitable ontologies and to incorporate relevant terms from existing resources.
ontology selection process. CEUR
ceur-ws.org
Consequently, MSE domain experts struggle to identify and select suitable ontologies for their applications. A comprehensive review encompassing all existing ontologies, detailing their reuse in the MSE domain, and providing a quality assessment is urgently needed to guide expert decision-making.
This paper aims to address the current lack of comprehensive knowledge about MSE ontologies by providing a thorough review and analysis. Our objectives include identifying available ontologies, understanding their specific purposes, developing criteria for ontology selection, and determining the necessary information for informed decision-making by domain experts. All evaluation results are available online4.
The paper is structured as follows: Section 2 describes the methodology used to collect and evaluate the MSE ontologies. Section 3 presents quantitative results of the review process and provides a thorough discussion of these findings in the context of our study objectives. Finally, Section 4 summarizes the primary outcomes of this study and outlines potential future research directions.
This section outlines the comprehensive methodology employed to evaluate ontologies in the Materials Science and Engineering (MSE) domain. The methodology comprises three Key components: expert insights and surveys, quality requirements and corresponding criteria, and ontology evaluation metrics. Each component significantly contributes to a thorough assessment of ontologies, empowering MSE experts to select the most suitable ontology for their specific needs. The following subsections provide detailed explanations of each methodological aspect:
This subsection details the process of gathering expert insights and conducting a survey to identify
ontology requirements within the MSE domain. Expert insights provide a foundational understanding of
practical needs and challenges faced by domain professionals. To ensure the relevance and efectiveness
of the selected ontologies, an internal survey was conducted as part of the Platform Material Digital
(PMD) project5. The PMD project comprises 13 industry-led pilot projects with the shared requirement of
(re)using ontologies in the MSE domain. A key finding from our qualitative analysis is that the responses
of the 13 projects focused on diferent aspects of the ontologies. This highlights the varied priorities and
perspectives of the participating experts. The publicly available interview results6 highlight the specific
requirements identified by MSE experts. We asked the responsible PIs of the thirteen PMD projects to
gather essential information about MSE-related ontologies within their domain. The survey targeted
MSE experts, including material engineers, scientists, process and application engineers, and simulation
experts, aiming to standardize taxonomy and metadata for materials and their properties. Key survey
questions addressed the desired ontology domains, intended use and requirements of the ontology,
target users, and the specific competency questions (CQ) the ontology should answer. Table 1 presents
quality requirements, their justifications, and corresponding criteria, as determined by survey results.
The criteria mapped to these requirements include Completeness (ensuring suficient information
for specified tasks), Coverage (measuring the breadth of domain information), Availability (assessing
the accessibility of the ontology and its documentation), and Adaptability (evaluating the ontology’s
capacity to accommodate changes without compromising verified definitions). Relevancy measures the
ontology’s alignment to specified tasks, while Accuracy assesses the precision and correctness of its
representations. Compliance ensures adherence to defined rules and standards, and Internal Consistency
guarantees logical coherence. Credibility reflects the ontology’s acceptance and trustworthiness, and
Complexity measures its structural intricacy. Finally, Comprehensibility measures users’ understanding,
and Modularity assesses the ontology’s composability from discrete, manageable units [
This paper specifically focuses on Availability (presence of ontology files and documentation),
Adaptability (number of CQs answered correctly after changes), Complexity (structural properties like depth
and breadth), Internal Consistency (presence of logical/formal contradictions), Compliance (total
number of breached rules), Credibility (number of other ontologies that link to it or positive user feedback),
Comprehensibility (degree of annotations and naming conventions), and Modularity (number of
ontology partitions and root nodes). While the metrics for these criteria are defined, future work will
address Coverage, Completeness, and Accuracy for a more holistic evaluation [
This subsection outlines the detailed metrics employed to evaluate the ontologies within the MSE
domain. Focusing on base, schema, and graph metrics, we measure ontology structure, complexity, and
usability. These metrics, categorized accordingly, were used in conjunction with the ROBOT tool7 [
Table 2 presents the key metrics used in our evaluation including descriptions and significance. These metrics align with W3C Semantic Web standards, specifically OWL and OWL DL. Base metrics include Axioms, Class Count, Object Properties Count, Datatype Properties Count, Annotation Assertions Count, and DL Expressivity. These metrics collectively indicate the overall size, complexity, and breadth of the ontology, as well as the richness of relationships, data values, and applied logical constructs. For schema metrics, Attribute Richness, Inheritance Richness, Relationship Richness, Axiom Class Ratio, and Equivalence Ratio are included. These metrics reflect the detailed knowledge representation, categorization, interconnectedness, logical definition detail, and redundancy among named classes. Complex class definitions are excluded from these metrics. Graph metrics evaluated include Absolute Root Cardinality (NoR), Absolute Leaf Cardinality (NoL), Number of External Classes (NoC), Depth, Breadth, and Tangledness are evaluated. These metrics assess the foundational structure, granularity, interdependence with external ontologies, hierarchical complexity, width, and interconnectivity of the ontology.
Furthermore, the OOPS! (Ontology Pitfall Scanner!) tool [
In our comprehensive analysis of semantic artifacts relevant to the Materials Science and Engineering (MSE) domain, a total of 94 semantic artifacts were identified, comprising 2 vocabularies and 92 ontologies. These ontologies can be categorized as follows: 11 general scientific ontologies, 7 without publicly available files, 4 top-level ontologies, 8 mid-level ontologies, 60 domain-level ontologies, and 2 application-level ontologies. Notably, 7 ontologies could not be evaluated due to issues with their imports. These ontologies include the Virtual Materials Marketplace (VIMMP) [83] Ontology, Chemical Entities of Biological Interest (ChEBI) [84], Chemical Information Ontology (CHEMINF) [85], Metal Alloy (MetalAlloy) [86], tribAIn Ontology [58], Semantic Materials Manufacturing Design (SEMMD) [39], and Chemical Methods Ontology (CHMO) [31]. This highlights the necessity for improved import handling and integration in ontology development processes. Consequently, 60 ontologies, as summarized in Tables 3, 4, and 5, were evaluated using our introduced methodology9. This list comprehensively details each ontology’s name, abbreviated short name, domain, projects utilizing it, purpose, publication 7http://robot.obolibrary.org/ 8https://ontometrics.informatik.uni-rostock.de/ontologymetrics/ 9Link to the list of ontologies
Quality Requirements Justification Mapped Criteria pdRroEemQhae1ni:nssiTvohefetmaoxanotnteorolimoalgysysccomiveuenrscitne.gprnoevciedsesaarycsoumb-- taThnhadetcttohevreemorna”tlcloonlomegcpyersseshhaoeryunlssdiuvbbee”-dcosoummgagpienlesstt.es eCroamgep, lAevteanileasbsi,litCyov
resenting both experimental and simulation types of data indicates a complete and Adaptability, Availdata. adaptable ontology. ability
sent relations between diferent concepts. these relationships matter to the do- Availability main. wuRnEitiQhfy4e:xmiTsethtinaedgoasnttataondldoegasrycdrsmip.utisotnsstainndcaormdipzleiaanncde aSntadnidnatredrnizaaltcioonnsiismtepnlcieys. compliance aCCbooimnlistpyislitaennccey,,IntAevrnaaill
sive representations for experimental settings, Comprehensive representations imply Completeness, Covoutcomes, high-throughput analysis data, and completeness and coverage. erage, Availability literature. wdRaEotQrat.h6:y aTnhdeveorniftioalbolgeyqumaulistyt mfaacnilaitgaetmeetnrutsotf- cTrreudsitbwiloirttyhainnedssacacnudravceyri.fiability imply ACvraedilaibbiilliittyy, Accuracy, REQ7 (more specific): The ontology should Specificity for machine learning makes Relevancy, Complexenable querying specifically designed for ma- it relevant, complex, and comprehen- ity, Comprehensibilchine learning model development. sive for that purpose. ity, Availability rRmeEsoQedne8tls(pm.reodriectsepdecviafilcu)e:sTfhroemonmtoalcohgiynemleuasrtnrienpg- apRnleedpterpenrseeesdnsitcaatnteiddoncporvooefpreamrgtaeiec.shiinmepllieeasrcnoinmg- eCroamgep, lAevteanileasbsi,litCyoveaRnpEopQuli9gc:ahtfiooTrnhd.eivoernsteoplorgojyecstshboeuyldonbdeitsmporidmualaryr iMn oddivuelrasreitpyraonjedcttshiemapbliylitaydatoptbaebiulisteyd. lAadriatpy,taAbvialiitlayb,iMlityoduof competency questions (CQs), licensing, last update date, homepage, ontology category, and a link to the ontology file. Additionally, it includes references to academic papers, citation counts, practical use cases, distinguishing features that contribute to its common use, and any special problems or challenges.
Table 6 highlights the distribution of ontologies across various MSE sub-domains, with Materials Representation and Materials Characterization having the most extensive coverage due to the complexity of these fields. Process Modeling, Nanomaterials, Computational Materials Science, and Materials Data also show significant ontology usage. However, domains such as Batteries, Chemistry, Energy Systems, Tribology, Biomaterials, and Sensors comprise fewer ontologies, indicating a need for further development to enhance their modeling capabilities and support advanced research applications.
Among the ontologies identified, the top-level ontologies include: Basic Formal Ontology (BFO) [
Our analysis highlights the diversity and complexity of MSE ontologies, reflecting their varied applications and the necessity for rigorous evaluation methodologies. BFO is prominently reused in 16 ontologies, EMMO in 12, BWMD-MID in 3, NPO in 2, and SSN, PMD Core, and GPO in one each, indicating its broad applicability and acceptance in the community. This frequent reuse suggests that fTionteadl innutmhebeornotoflcolgays.ses de- isSmehnpotwlyasbwtrhiodeaedbrerarenadgdoetmhoafoincfoccnoocnvecepertapsgt.se caonvdertehde.aHbiilgithyertocoreupnrteseTrottiaelsninumthbeeroonftoolbojgeyc.t prop- IaMnndodircebaeottebtsejertchrteepprvroeaspreieenrtttyiaetsoiofsnuregolgfaectsiootmnrsipchlheipexrsinibnteetterwarcceoteinonnnecscl.atsiosness. pTorotaplerntiuemsibnetrhoefondtaotlaotgyyp.e aRHteitgfrlhiebceutrstcetosh.uenrtasnignediocfadteatdaevtaaliuleeds raespsorecsiaetnetdatwioitnhocflacslasesss. iTnottahlenounmtobleorgoyf. annotations sSMthaoonrwdeasabntihlnietoyt.aamtioonusntenohfadnecsecrdiopctiuvmeemnetatatidoantaanpdrouvniddeedr-. sTihveitydeosfcrtihpetioonntloolgoigcye.xpres- IsHnoidngiihcneagrt.eesxptrheessicvoitmy palleloxwitsyfoorf nluoagniccaeld caonndsctroumcptslexusreead-. tArvibeurategse penrucmlasbse.r of at- itHmioigpnhr,oeevrsisnveagnluutiesaaslbifniolidrtiyccoainntevaedpyepitnliacgialectdioomnksnp.olewxliendfgoermreaptrioesneanntadcAlvaesrsaegsepernculmasbse.r of sub- taHinoednlpsastnriudnchutuinerdreaedrr.csHhtaiicngadhliesnrtgrvuahcloutuwersiknsnugog.wgelestdbgeetitsercacateteggoorirzizead
per class. ratios suggest more detailed and well-defined concepts. aPxroiopmorsttioonallocflaesqseusiv.alence ttReherefmloesnctatsoreloredgdeyfu’isnnidendatneagcsryaetqaiuonindvaaslbeyinnlitto.ynH.yimghye,rivnadliuceastiinmgphroovwe
tologies. A higher count suggests better interoperability, reusability, and alignment with external standards, though it may also introduce complexity.
BFO and EMMO are considered high-quality foundational ontologies representing various aspects of MSE.
The study found that only nine of these ontologies explicitly published their competency questions:
PRovenance Information in MAterials science, Crystal Structure Ontology, Computational Material
Sample Ontology, Metadata4Ing Ontology, Open Energy Ontology, Materials Design Ontology,
Dislocation Ontology, SMART-Protocols, and PMD Core Ontology. Publishing CQs is crucial for clarifying
the ontology’s intent, facilitating adaptability assessment, and ensuring human-readable answers [
FP7, SemsorGrid4Env, Exalted,
CSIRO 7 7 7 7 3 7 7 7 7 7 7 7 3 7 7 3 7 7 7 7 Patterns (ODPs). The list was updated in June, and all the ontologies were found to the best of our knowledge. 7means that the information could not be found either from the ontology repository or from the publication reference of the ontology. While formulating CQs is an integral part of ontology development, they are often not published along with ontologies, hindering evaluation eforts. Although modern design methodologies advocate for user stories, personas, and contextual statements beyond CQs [87], none of the examined ontologies adopted these approaches. This emphasizes the need for a more comprehensive approach to capturing and addressing user needs and requirements.
The evaluation of MSE ontologies reveals varying levels of complexity and detail across diferent sub-domains, as shown in Tables 9 and 10. In
Materials Characterization, ontologies like EMMO Crystallography and CIF-core demonstrate moderate axiom counts and high annotation axiom counts, indicating a balance between detailed representation and descriptive metadata. CHAMEO, with substantial object properties and annotation axioms, emphasizes detailed relationships and descriptive details. CQs 3 3 3 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 3
Lic. 7 CC0 1.0 CC BY
4.0 CC BY
4.0 CC BY
3.0 In Process Modeling, GPO, EXPO, and PMDCO have high axiom and class counts, reflecting their capability to model complex processes comprehensively. The high DL expressivity of GPO ( ℛ ℐ ( ) ) indicates its advanced logical constructs and reasoning capabilities.
In Computational Materials Science, CMSO and MDO show diverse modeling approaches with substantial axiom counts, indicating detailed and comprehensive domain representation. Materials Representation ontologies like MatOnto and MSEO, with high axiom and class counts, suggest rich and detailed modeling capabilities across various sub-domains. For Nanomaterials, NPO stands out with extensive axioms and classes, indicative of detailed nanoparticle interaction modeling. eNanoMapper and NanoMine, with lower axiom counts and simpler DL expressivity, focus on specific nanomaterial aspects. In Mechanical Testing, MOL Brinell has a high axiom count but simpler DL expressivity ( ℒ ( ) ), suggesting a thorough yet straightforward representation. Additive Manufacturing ontologies, such as LPBFO, exhibit detailed and complex representations, whereas AMONTOLOGY ofers a broader but
ogy, Crystallographic Defect Core Ontology, Line Defect Ontology,
tology of Scientific Experiments, Metadata4Ing Ontology, The Open
simpler structure. In the Batteries domain, EMMO BattINFO and EMMO BVC present similar metrics, indicating detailed modeling capabilities.
Schema metrics provide insights into the richness and interconnectedness of these ontologies, as seen in Tables 11 and 12. High ACR and RR values denote detailed and interconnected models, which are essential for capturing the complexity of material characterization. However, these also bring increased computational challenges. High IR values suggest well-structured hierarchies, enhancing knowledge organization but potentially complicating ontology maintenance. The diversity in metrics indicates that while some ontologies are well-suited for detailed and complex modeling, others may ofer more streamlined and eficient structures, highlighting the need for balance depending on specific application requirements. For instance, in the Materials Characterization domain, ontologies like CSO and PLDO show high Attribute Richness (AR) and Axiom Class Ratio (ACR), indicating detailed knowledge representation and logical definitions, making them suitable for applications requiring extensive detail. EMMO Crystallography and CSO excel in Inheritance Richness (IR), suggesting a wellstructured hierarchical categorization, are beneficial for understanding domain taxonomy. CHAMEO and DISO are notable for their high Relationship Richness (RR), implying a well-connected structure that enhances relationship understanding. The choice of ontology depends on the application needs, with CSO, PLDO, and CDCO recommended for detailed logical reasoning, EMMO Crystallography and CSO for hierarchical structuring, and CHAMEO and DISO for comprehensive relationship mapping.
Graph metrics, detailed in Tables 13 and 14, provide a deeper understanding of the structural properties of the ontologies. Higher root cardinality (NoR) indicates a diversified foundational structure, enhancing the ontology’s ability to cover a broad range of concepts. Conversely, a higher number of leaf classes (NoL) suggests greater granularity and specificity, crucial for capturing detailed knowledge within the domain. The number of External Classes (NoC) is calculated by comparing the ontology’s namespace with the namespaces of the referenced classes. Classes that have a diferent namespace than those present in the ontology are considered external. For ontologies that consist of multiple modules, the modules are considered part of the core ontology. Therefore, classes from these modules are not treated as external. To be classified as external, a class must have a namespace that difers from any of the namespaces used within the core ontology and its modules. This ensures that only truly external references are counted, reflecting the ontology’s degree of interdependence with other distinct ontologies. The number of external classes (NoC) highlights the degree of interoperability and alignment with external standards, with a higher count indicating better integration but potentially adding complexity. Together, these metrics reveal the balance between foundational diversity, detail specificity, and interconnectivity, impacting the ontology’s usability and efectiveness in representing complex domains. For instance, in the Materials Characterization domain, ontologies like EMMO Crystallography and MSLE exhibit high root cardinality, which may suggest a broad foundational structure. However, ontologies like CSO and CDCO have fewer root classes, potentially indicating a narrower scope. It is important to note that this metric is just a hint; a thorough examination of the ontologies’ content is necessary to draw valid conclusions. An ontology with fewer root classes might still have a broader scope at deeper levels of the hierarchy, and the reduced number of root classes could result from a higher-level top classification. Ontologies such as EMMO Microstructure and OIE Characterisation Methods show high leaf cardinality, reflecting a high level of detail and specificity. In terms of external class references, CHAMEO and EMMO Microstructure demonstrate better interoperability with higher counts, whereas PODO and PLDO have fewer external references, implying less reliance on external ontologies. Depth and breadth metrics indicate complex hierarchical structures in ontologies like EMMO Crystallography, while others, such as CSO and CDCO, display simpler, more focused structures. Overall, these metrics highlight the diversity in foundational breadth, detail granularity, and external interconnectivity across the Materials Characterization ontologies, impacting their comprehensiveness.
For detailed evaluations in other domains, please refer to Appendix C and D. The appendix includes the same evaluations applied to ontologies in diferent domains such as Biomaterials, Sensors, and Energy, providing a comprehensive understanding of their complexities and structures.
In conclusion, this study provides a comprehensive evaluation of 94 ontologies within the field of Materials Science and Engineering (MSE), ofering a detailed analysis based on quality-control metrics. The findings highlight both the strengths and weaknesses of the evaluated ontologies, emphasizing their structural complexities, domain-specific relevance, and reuse of existing ontological frameworks. However, the study also identifies critical areas for improvement, such as the limited adoption of competency questions and ontology design patterns, and the need for better documentation and user support to address common pitfalls and enhance overall quality and usability.
Future work will focus on several key areas to further advance ontology development in MSE. Eforts will be made to systematically identify and extract Ontology Design Patterns (ODPs) to enhance quality and reusability. Additionally, further research will aim to comprehensively evaluate the completeness, domain coverage, and accuracy of these ontologies by extracting relevant domain terms and concepts within the subdomain of materials science. Moreover, the assessment of FAIRness (Findability, Accessibility, Interoperability, and Reusability) of the ontologies will be incorporated into future studies.
The authors thank the German Federal Ministry of Education and Research (BMBF) for financial support of the project Innovation-Platform MaterialDigital through project funding FKZ no: 13XP5094F (FIZ). This publication was written by the NFDI consortium NFDI-MatWerk in the context of the work of the association German National Research Data Infrastructure (NFDI) e.V.. NFDI is financed by the Federal Republic of Germany and the 16 federal states and funded by the Federal Ministry of Education and Research (BMBF) – funding code M532701 / the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 460247524.
The authors would also like to acknowledge Sven Hertling for his valuable assistance in discussing the reviews. [64] Dislocation simulation and model ontology, 2024. URL: https://github.com/OCDO/DSIM. [65] A. Z. Ihsan, D. Dessì, M. Alam, H. Sack, S. Sandfeld, Steps towards a dislocation ontology for crystalline materials, 2021. arXiv:2106.15136. [66] Atomistic simulation methods ontology, 2024. URL: https://github.com/OCDO/asmo. [67] Point defects ontology, 2024. URL: https://github.com/OCDO/podo. [68] Crystallographic defect core ontology, 2024. URL: https://github.com/OCDO/cdco. [69] Line defect ontology, 2024. URL: https://github.com/OCDO/ldo. [70] Planar defects ontology, 2024. URL: https://github.com/OCDO/pldo. [71] M. Jalali, M. Mail, R. Aversa, C. Kübel, Msle: An ontology for materials science laboratory equipment – large-scale devices for materials characterization, Materials Today Communications 35 (2023) 105532. URL: https://www.sciencedirect.com/science/article/pii/S2352492823002222. doi:https://doi.org/10.1016/j.mtcomm.2023.105532. [72] Open innovation environment domain ontologies, characterisation methods, 2022. URL: https: //github.com/emmo-repo/OIE-Ontologies/. [73] S. Arndt, B. Farnbacher, M. Fuhrmans, S. Hachinger, J. Hickmann, N. Hoppe, M. T. Horsch, D. Iglezakis, A. Karmacharya, G. Lanza, S. Leimer, J. Munke, D. Terzijska, J. Theissen-Lipp, C. Wiljes, J. Windeck, Metadata4Ing: An ontology for describing the generation of research data within a scientific activity., 2022. URL: https://doi.org/10.5281/zenodo.5957104. doi:10.5281/zenodo. 5957104. [74] Materialsmine ontology, 2021. URL: https://github.com/tetherless-world/materialsmine. [75] Periodic table ontology, 2004. URL: http://www.daml.org/2003/01/periodictable/. [76] T. Käfer, S. Lauber, A. Harth, Using workflows to build compositions of read-write linked data apis on the web of things., in: ISWC (P&D/Industry/BlueSky), 2018. [77] A. Bandrowski, R. Brinkman, M. Brochhausen, M. H. Brush, B. Bug, M. C. Chibucos, K. Clancy, M. Courtot, D. Derom, M. Dumontier, et al., The ontology for biomedical investigations, PloS one 11 (2016) e0154556. [78] M. Ashburner, C. A. Ball, J. A. Blake, D. Botstein, H. Butler, J. M. Cherry, A. P. Davis, K. Dolinski, S. S. Dwight, J. T. Eppig, et al., Gene ontology: tool for the unification of biology, Nature genetics 25 (2000) 25–29. [79] G. V. Gkoutos, P. N. Schofield, R. Hoehndorf, The anatomy of phenotype ontologies: principles, properties and applications, Briefings in Bioinformatics 19 (2018) 1008–1021. [80] Material properties ontology, 2020. URL: https://bimerr.iot.linkeddata.es/def/material-properties/. [81] Matolab tensile test ontology, 2021. URL: https://matportal.org/ontologies/MOL_TENSILE. [82] Matolab brinell test ontology, 2022. URL: https://matportal.org/ontologies/MOL_BRINELL. [83] M. T. H. et al., Ontologies for the virtual materials marketplace, 2020. arXiv:1912.01519. [84] K. Degtyarenko, P. De Matos, M. Ennis, J. Hastings, M. Zbinden, A. McNaught, R. Alcántara, M. Darsow, M. Guedj, M. Ashburner, Chebi: a database and ontology for chemical entities of biological interest, Nucleic acids research 36 (2007) D344–D350. [85] Chemical information ontology, 2024. URL: https://github.com/egonw/semanticchemistry. [86] Metal alloy, 2021. URL: https://github.com/arsarkar/ontorepo/tree/master/metal-alloy. [87] V. Presutti, E. Blomqvist, E. Daga, A. Gangemi, Pattern-based ontology design, Ontology Engineering in a Networked World (2012) 35–64.
This supplementary section provides a comprehensive evaluation of various aspects of MSE ontologies. Section A focuses on error detection using the OOPS! tool. Section B presents base metrics, examining structural aspects. Section C evaluates schema metrics. Finally, Section D details graph metrics.
The OOPS! (Ontology Pitfall Scanner!) tool was employed for error-checking evaluations. Identifying and categorizing these pitfalls is essential for the evaluation process. Critical pitfalls impact ontology consistency, reasoning, and applicability, leading to significant issues in how the ontology is used and interpreted. Important pitfalls, while not afecting the core functionality, can still degrade the quality and reliability of the ontology. Minor pitfalls are less problematic but addressing them can enhance the ontology’s organization and user-friendliness. By comparing the presence of these pitfalls across diferent ontologies, MSE domain experts can make informed decisions about which ontology best meets their needs, ensuring that the chosen ontology is robust, reliable, and suitable for their specific applications. Table 7 describes only the pitfalls that exist in the MSE ontologies, along with their descriptions and impact levels.
Table 8 details the results of ontology evaluation using the OOPS! tool. Critical pitfalls afecting ontology consistency and reasoning include P19 (Multiple domains or ranges for properties), P40 (Namespace hijacking), P31 (Incorrect use of owl:equivalentClass), P05 (Incorrect use of owl:inverseOf), P29 (Incorrect use of owl:TransitiveProperty), and P27 (Incorrect use of owl:equivalentProperty). Important pitfalls, although less severe, still afect ontology quality and should be addressed. Minor pitfalls, while not critical, can be improved for better organization and user-friendliness.
The frequent occurrence of critical pitfalls such as P19 and P40 across multiple ontologies suggests a need for more stringent guidelines and validation tools during ontology development. The presence of these pitfalls can significantly impact the usability and accuracy of the ontologies in real-world applications. For instance, P19’s misinterpretation issues can lead to incorrect inferences, while P40 can hinder efective data retrieval and integration.
The Tables 9 and 10 provide an evaluation of the base metrics for MSE ontologies. It includes several columns such as Domain, Ontology Name, and the total number of Axioms. Additionally, it details the Class Count, Object Property (OP) Count, Data Property (DP) Count, and Annotation Axiom (Ann. Axm.) Count. The table also highlights the Description Logic Expressivity10 (DL Expr.) and includes information on the OWL2 profile 11.
Materials Characterization ontologies such as EMMO Crystallography (NoR: 33, NoL: 4) and MSLE (NoR: 63, NoL: 6) show broad foundational structures and detailed coverage, indicated by deep (Max Depth: 4) and broad (Max Breadth: 1760) hierarchies. CHAMEO and OIE Characterisation Methods have high leaf cardinality (NoL: 34-35) and demonstrate good interoperability with moderate tangledness (0.05-0.12). Process Modeling ontologies like GPO (NoR: 115, NoL: 8) and EXPO (NoL: 202) reflect broad foundational coverage and significant depth (Max Depth: 8-12) with moderate tangledness (0.10-0.11). PMDCO shows a balanced structure with extensive roots (NoR: 194) and low tangledness (0.02).
In Computational Materials Science, ASMO (NoR: 28, NoL: 3) and CMSO (NoR: 31, NoL: 2) have broad structures with no tangledness. DSIM (NoR: 17, NoL: 4) exhibits moderate depth and higher tangledness 10https://en.wikipedia.org/wiki/Description_logic 11https://www.w3.org/TR/owl2-profiles/
Imp. Pitfall l a c i t i r C t n a t r o p m I r o n i M
Multiple domains or ranges defined for properties: Leads to misinterpretation as a conjunction in OWL.
Namespace hijacking: Terms from another namespace are used without proper definition, preventing information retrieval.
Incorrect use of owl:equivalentClass: Defining non-equivalent classes as equivalent.
Incorrect use of owl:inverseOf: Defining non-inverse relationships as inverse.
Incorrect use of owl:TransitiveProperty: Defining non-transitive relationships as transitive.
Incorrect use of owl:equivalentProperty: Defining non-equivalent properties as equivalent.
Lack of domain or range definitions: Properties without defined domain or range may cause misunderstandings.
Defining classes as instances: Instances incorrectly defined as classes, causing confusion in class hierarchy.
Misuse of transitive property: Improper use of transitive property afects ontology reasoning.
Misuse of symmetric property: Misuse can lead to incorrect inference.
Using diferent naming conventions: Inconsistent naming conventions reduce ontology clarity.
Use of deprecated classes or properties: Use of deprecated elements afects ontology maintenance. (0.24). Materials Representation ontologies like MatOnto (NoR: 848, NoL: 13) and MSEO (NoR: 100, NoL: 5) suggest deep, wide structures with no tangledness. Nanomaterials ontologies such as NPO (NoR: 65, NoL: 5) and NanoMine (NoR: 1, NoL: 5) show extensive depth (Max Depth: 14) and breadth (Max Breadth: 284-1593) with moderate to low tangledness (0.03-0.31). Mechanical Testing ontologies like EMMO Mechanical Testing (NoR: 13, NoL: 7) and MOL Brinell (NoR: 17, NoL: 3) have moderate depth (Max Depth: 3-7) and tangledness (0.21). Additive Manufacturing ontologies like AMONTOLOGY (NoR: 5, NoL: 3) and LPBFO (NoR: 2, NoL: 4) show moderate depth and breadth with low tangledness. Batteries ontologies like BattINFO (NoR: 10, NoL: 3) and EMMO BVC (NoR: 11, NoL: 4) suggest broad foundational coverage with high horizontal coverage (Max Breadth: 1203-1781) and low tangledness (0.01-0.05).
Materials Characterization ontologies like EMMO Crystallography (NoR: 33, NoL: 4) and MSLE (NoR: 63, NoL: 6) show extensive foundational structures and detailed coverage, with deep (Max Depth: 4) and broad (Max Breadth: 1760) hierarchies. CHAMEO and OIE Characterisation Methods have high leaf cardinality (NoL: 34-35) and good interoperability with moderate tangledness (0.05-0.12). Process Modeling ontologies like GPO (NoR: 115, NoL: 8) and EXPO (NoL: 202) reflect broad foundational coverage and significant depth (Max Depth: 8-12) with moderate tangledness (0.10-0.11). PMDCO shows balanced structure with extensive roots (NoR: 194) and low tangledness (0.02).
In Computational Materials Science, ASMO (NoR: 28, NoL: 3) and CMSO (NoR: 31, NoL: 2) indicate broad structures with no tangledness. DSIM (NoR: 17, NoL: 4) has moderate depth and higher tangledness (0.24). Materials Representation ontologies like MatOnto (NoR: 848, NoL: 13) and MSEO (NoR: 100, NoL: 5) suggest deep, wide structures with no tangledness. Nanomaterials ontologies like NPO (NoR: 65, NoL: 5) and NanoMine (NoR: 1, NoL: 5) show extensive depth (Max Depth: 14) and breadth (Max Breadth: 284-1593) with moderate to low tangledness (0.03-0.31). Mechanical Testing ontologies like EMMO Mechanical Testing (NoR: 13, NoL: 7) and MOL Brinell (NoR: 17, NoL: 3) have moderate depth (Max Depth: 3-7) and tangledness (0.21). Additive Manufacturing ontologies like AMONTOLOGY (NoR: 5, NoL: 3) and LPBFO (NoR: 2, NoL: 4) show moderate depth and breadth with low tangledness. Batteries ontologies like BattINFO (NoR: 10, NoL: 3) and EMMO BVC (NoR: 11, NoL: 4) suggest broad foundational coverage with high horizontal coverage (Max Breadth: 1203-1781) and low tangledness (0.01-0.05). Additive Manufacturing Batteries Biomaterials Sensor Energy
NoC