Skills knowledge graph curation: the case of digitalization design domain

Skills are now the foundational element of HR practices and HR Tech. Skills can be used to identify what employees can do, should do and what are the gaps, such gaps can be used for recommending learning content and career development paths. Skills are also used in recruiting part of human capital management - for specifying requirements for candidates and matching CVs with vacancies.||According to [Bersin, 2021] one of the hottest markets in HR Tech is the market for what he calls “SkillsTech,” tools that help identify, categorize, assess, manage, and improve skills at work. Examples of tools in this category: SkillsDNA within EDCast, Cornerstone's Clustree, Workday's Skills Cloud, Emsi Skills. Skills within SkillsTech are organized using taxonomies or skills graphs. Every domain usually has various skills taxonomies and reference models, besides they are changing constantly.||Although AI-driven services such as automatic skills extraction and taxonomy generation are provided by major SkillsTech vendors, the quality of final results is not ideal and input from human experts is still required.||So there is a need for the skills data/knowledge curation activities in each domain, which include skills taxonomy development and support, mappings between different skills taxonomies and reference models, creation and support of annotated "golden corpus" etc.||Within the talk, such curation activities for the digitalization design area will be described. This area includes skills of digital architects (enterprise architect, application architect, data architect, solution architect etc.), analysts (business analyst, system analyst), IT product managers, etc. The following components of the skills knowledge graph in the selected area were developed:||- a library of skills taxonomies and reference models, which are relevant for this area (for example TOGAF Architecture Skills Framework, SFIA etc)||- the own reference skills taxonomy (reference model) and mappings to other skills taxonomies and reference models,||- reference job profiles, which link jobs (roles) with skills,||- a fragment of the annotated dataset with job postings ("golden corpus") using the suggested skills taxonomy, such corpus can be used by AI and ML tools for training and evaluation purposes.||The suggested skills knowledge graph can be reused by HR Tech vendors, semantic technology vendors, or by employers, which want to speed up the development of their talent management solutions and make them more cost-effective.||The components of the skills knowledge graph were tested within two pilot projects – 1) for the IT department of a mining company and 2) for IT consultancy. The case study for the first pilot project will be also covered in the presentation.