Skills Ontology

A skills ontology is a structured taxonomy of skills, their relationships to each other, their relationships to roles and tasks, and their proficiency levels. It is the data backbone for skills-based hiring, internal mobility, learning recommendations, and workforce planning — the layer that lets HR systems reason about people in terms of capabilities rather than just job titles.

It is the HR analogue of a knowledge graph: nodes (skills, roles, tasks, certifications) and edges (prerequisite-for, similar-to, used-in, taught-by) with explicit semantics that AI systems can traverse and reason over.

Core components

Skill nodes

Each skill is a structured entity with attributes: name, definition, category, and ideally an evidence signal (what does demonstrating this skill look like). Skills can be technical (Python, financial modeling, GDPR compliance), domain (industrial control systems, supply-chain optimization), or transversal (negotiation, project management, technical writing).

Skill relationships

The graph captures relationships between skills: prerequisites (calculus → linear algebra), similarity (TypeScript ≈ JavaScript), substitutability, and progression. This structure is what enables non-obvious internal-mobility suggestions — "this candidate has 80% of the skills for this role and lacks two skills that they could acquire in 6 weeks of training."

Role and task mappings

Skills are mapped to roles (the skill profile of a "senior data engineer") and to tasks (the skill demand of "build a customer churn model"). Job descriptions, learning content, and project assignments are all expressed in terms of the same skill vocabulary.

Proficiency levels

Skills aren't binary. Most ontologies use a 4–6-level scale (e.g. novice, advanced beginner, competent, proficient, expert) with explicit behavioral descriptors at each level so the levels mean the same thing across the organization.

Evidence and inference

Skills aren't just self-declared. They are inferred from project history, peer endorsements, certifications, learning completions, and outcomes. The ontology supports linking each claimed skill to evidence so it can be trusted by downstream decisions.

Why it matters for enterprise

The shift from job-title mobility to skills-based mobility is one of the defining trends in modern HR. Job titles are coarse, slow-changing, and inconsistent across companies. Skills are fine-grained, faster-evolving, and translatable across roles and industries. A skills ontology is the precondition for moving HR systems from title-thinking to skills-thinking.

Deloitte's 2024 Skills-Based Organization research consistently shows skills-based organizations outperforming title-based peers on internal-mobility rates, time-to-fill, and engagement — but only when the underlying skills data is structured and trustworthy. A messy skills "list" without ontology gives the appearance of skills-based work while delivering none of the benefits.

Public ontologies like the European Skills, Competences, Qualifications and Occupations (ESCO) and O*NET provide reusable starting points; most enterprises tailor a layer on top for industry-specific and proprietary skills.

Common use cases

  • Internal mobility — surfacing employees who match an open role even if their job title doesn't match.
  • Hiring — writing job descriptions in skills terms and screening candidates against skill profiles. See skills-based hiring.
  • Learning recommendations — recommending courses and assignments based on skill gaps relative to an employee's career path.
  • Workforce planning — forecasting future skill demand against current skill inventory to plan reskilling or hiring. See workforce intelligence.
  • Project staffing — matching consultants and project resources to engagement requirements based on skill rather than availability alone.
  • Acquisition integration — mapping the skills of an acquired workforce to the acquirer's organization for retention and integration planning.

Related concepts

For the platform view that puts a skills ontology at the center of an HR intelligence stack, see the HR intelligence platform pillar (UC-2).

Frequently asked questions

Should we build our own ontology or use a standard?

Hybrid is the right answer for most enterprises. Start from a standard like ESCO or O*NET as a base; layer industry- and company-specific extensions on top. Building from scratch is rarely worth it; relying purely on standards misses meaningful proprietary skills.

How is skill data validated?

A combination of self-declaration, manager validation, peer endorsement, certification linking, and inferred signals from project work and learning. The most trustworthy systems triangulate across at least three of these.

Does skills ontology work for non-technical roles?

Yes — and arguably better than title-based systems. Skills like "stakeholder management," "policy interpretation," and "facilitation" are central to many non-technical roles and are poorly captured by job titles. A skills ontology makes them explicit and comparable.

How does it interact with AI candidate matching?

It is the foundation. AI candidate matching without a skills ontology relies on title and keyword matching, which is brittle and biased. With a skills ontology, candidate-to-role matching becomes a graph-traversal problem with explicit reasoning.

How is the ontology kept current?

Continuously, not annually. Modern systems use AI to extract emerging skills from job postings, project descriptions, and external sources, then validate proposed additions through human review. Rapid skill turnover (especially in technology) makes annual review insufficient.