Flexible and Extensible Competency Management with Knowledge Graphs

Abstract

Especially in the diverse and fast-paced eld of Arti cial Intelligence it is imperative to have a clear picture of relevant competencies and how they are distributed within or over organisations. For this purpose, we have developed a generic competency ontology that can be used to describe competencies of people and organisational structures in the Arti cial Intelligence domain. The ontology is embedded in an application to create, manage, and utilize a Competency Knowledge Graph. In our presentation we show concrete application scenarios, advantages, and challenges.

-

{ Discover persons or organisations having relevant competencies; suggest related competencies to identify knowledge gaps; nd completed projects to foster reuse (based on competencies de ned using the ontology).

The Competency Ontology The competency ontology provides the three generic classes Competency, CompetencyHolder, and CompetencyTarget. These are supposed to be extended by sub-classes in order to describe a concrete scenario. A Competency is a particular skill or piece of know-how that one might possess or apply. To be able to describe competencies in the AI domain, we incorporate the data from AI-KG [ 1 ], which is a knowledge graph automatically extracted from the Microsoft Academic Graph.3 The graph describes 850K research entities (i.e. speci c Tasks, Methods, Metrics, and Materials in the AI domain) and their relations. We include those as sub-classes of Comptency in our ontology. CompetencyHolder s are entities that possess or apply competencies. In our case, those are persons and organisations. To model them, we re-use de nitions from existing ontologies like FOAF4 and ORG5 as sub-classes. CompetencyTarget s serve to describe the context in which concrete competencies are applied in. In a business setting, these are concepts like projects, processes, or applications. Here, we again rely on well-established ontologies like FOAF.

Application Design We use metaphactory [ 3 ] for the design of the competency management application as it provides several advantages; primarily its ability to guide the input of competency descriptions based on the underlying ontology through intuitive components like ontology-driven semantic forms and search. Furthermore, metaphactory's low-code approach makes it possible to rapidly develop custom dashboards. In general, a knowledge graph-based approach to de ne competencies reduces e ort with respect to data integration, scalability and exibility. Apart from the re-use of established ontologies to cover the modeling requirements, external data sources can also be tapped using federated queries. For instance, we display information about publications of AI competencies from an external SPARQL endpoint of the Microsoft Academic Graph (as it would be too large to integrate). The approach is also easily extendable to domains outside of AI due to the open nature of the RDF data model and metaphactory's low-code dashboarding capabilities. 3 https://www.microsoft.com/en-us/research/project/microsoft-academic-graph/ 4 http://www.foaf-project.org 5 https://www.w3.org/TR/vocab-org/

References

1. Dess , D. , et al.: AI-KG: an automatically generated knowledge graph of arti cial intelligence . In: International Semantic Web Conference. pp. 127 { 143 . Springer ( 2020 )

2. Galkin , M. , Auer , S. , Vidal , M.E. , Scerri , S. : Enterprise knowledge graphs: A semantic approach for knowledge management in the next generation of enterprise information systems . In: ICEIS (2) . pp. 88 { 98 ( 2017 )

3. Haase , P. , Herzig , D.M. , Kozlov , A. , Nikolov , A. , Trame , J.: metaphactory: A platform for knowledge graph management . Semantic Web 10 ( 6 ), 1109 { 1125 ( 2019 )