This post was originally published on this site.
The promise of skills-based hiring is clear: A labor market where individuals are matched to jobs based on what they can do rather than where they went to school. Yet despite advances in AI and data-driven hiring tools, traditional job matching systems remain entrenched, relying on outdated credentials and rigid classifications. This blog explores how a hybrid approach—combining structured taxonomies with adaptive ontologies—can provide a scalable solution to these inefficiencies. By leveraging both frameworks, employers, policymakers, and educators, it is possible to create a more dynamic and inclusive job market. However, for this model to scale, stakeholders must invest in interoperable data systems, AI-driven updates, and policy reforms that incentivize competency-based hiring.
Picture a job seeker endlessly scrolling through job postings, unsure whether their experience matches what employers seek. Meanwhile, hiring managers skim through resumes, relying on intuition and outdated credentials as stand-ins for actual competence. This mismatch results in prolonged unemployment, wage disparities, and costly hiring mistakes—undermining both individual careers and overall economic productivity.
Now, imagine a different reality: A job seeker uploads their resume (only once!), employers enter job requirements, and AI-driven platforms instantly match candidates based on skills, credentials, and verified competencies—rather than keywords or degrees. This shift could transform hiring for millions of workers without degrees or traditional employment histories, including gig workers and self-taught professionals, allowing them to compete based on ability, not pedigree.1
However, despite its clear advantages, skills-based hiring remains aspirational. Adoption is limited due to inconsistent implementation, lack of standardization, and employer reliance on traditional hiring habits. The real challenge lies not just in adopting new technology but in rethinking how talent is recognized and matched.2
One critical component of this paradigm shift is missing: A shared skills language that seamlessly links talent supply with employer demand. Without it, even the most sophisticated hiring technologies risk reinforcing current inefficiencies.
What’s wrong with traditional job hiring?
For decades, job matching has relied on a flawed system: Employers post openings, job seekers apply based on guesswork, and hiring managers screen resumes with subjective filters—lacking reliable, standardized information. Absent clear information, job seekers rely on vague descriptions, while hiring managers use superficial criteria for selection such as resume formatting, job titles, or alma maters.
These biases don’t just waste time; they systematically exclude skilled workers without formal credentials—gig workers, freelancers, and self-taught professionals—despite having the right skills. Without a better system, businesses miss out on talent and millions of qualified candidates remain in low-paying jobs.
Taxonomies: A structured, rigid approach
To bridge the job matching gap, policymakers and researchers often use skill taxonomies—structured systems that organize competencies (and other job-relevant information such as education, credentials, or work experience) into hierarchical categories. A widely used example in the U.S. is O*NET, which maps skills from broad to specific levels. For example, basic skills (essential for various occupations) serve as a broad category, social skills (interpersonal effectiveness) form a mid-level subgroup, and active listening (paying attention and understanding others), speaking (conveying information), persuasion (convincing others), and negotiation (resolving differences) are specific competencies listed at the most detailed level.
In theory, taxonomies provide a shared hiring language. In practice, however, they are too rigid to fully capture the fluid nature of work. Job titles rarely map cleanly to fixed skill sets—a project manager in tech needs a vastly different skill mix than one in health care. And because taxonomies rely on predefined classifications, they struggle to keep up with rapidly evolving industries and emerging job roles. Moreover, maintaining an up-to-date taxonomy is resource intensive. O*NET, for instance, lists 1,016 occupation titles and codes (detailed data collection has been completed for 923 of them). Yet job descriptions change rapidly, making it difficult to keep these frameworks relevant.
Despite their limits, taxonomies remain useful for research and policy, informing studies on job polarization and technological change. However, they fail to fully capture job complexity and context.
Ontologies: A flexible, elusive target
As an alternative, researchers are turning to ontologies—dynamic frameworks that map how skills interact across jobs and industries. Unlike taxonomies, ontologies adapt to context, making them valuable for AI-driven hiring tools.
For example, rather than listing “data analysis” as a skill, an ontology maps its connections—to Python, statistical modeling, and industry applications like finance or health care. This adaptability helps ontologies evolve with shifting job market demands.
However, scalability remains a challenge. Manually mapping thousands of skills and industries is an enormous task, requiring continuous updates. This is where machine learning (ML) and natural language processing (NLP) come into play. AI-driven ontologies can extract skill relationships from job postings, resumes, and employer data, dynamically updating frameworks to reflect emerging roles and competencies.
A hybrid approach: The best of both worlds
Neither taxonomies nor ontologies alone provide a perfect solution. Taxonomies offer structure but lack flexibility, while ontologies are adaptive but can be difficult to standardize. A hybrid approach—layering ontologies onto taxonomies—leverages the strengths of both.
Taxonomies provide a structured foundation, categorizing skills and occupations. Ontologies add relational depth by mapping how skills evolve across industries. This improves interoperability, allowing AI-driven job matching to work across platforms. By blending structure with adaptability, hybrid models make skills-based hiring more effective and scalable.
ESCO, the European Commission’s skills framework, integrates both taxonomies and ontologies by categorizing skills hierarchically while mapping their relationships. For example, “Project Management” links to “Agile Methodology” in IT and “Supply Chain Coordination” in logistics. LinkedIn’s Economic Graph and Burning Glass Technologies use AI to map skill adjacencies and emerging job titles, enabling career transitions based on competencies rather than rigid titles. IBM’s SkillsBuild also applies a hybrid model, combining taxonomies with AI-driven ontologies for personalized learning and career pathways.
At the sub-national level, Arkansas’ Veterans SkillBridge Program applies a hybrid approach to help veterans transition into civilian employment. The program uses taxonomic classifications to structure job roles while incorporating ontological relationships to map military skills to civilian job requirements, providing targeted career recommendations and reskilling opportunities. A similar ongoing effort is Alabama’s Talent Triad, a state-led initiative mapping taxonomic job classifications with ontological career pathways.
By balancing structure with adaptability, hybrid systems benefit recruitment platforms, reskilling programs, and competency-based hiring. But their greatest impact is expanding opportunity: For workers without degrees, immigrants, and gig workers, these models ensure hiring decisions are based on skills, not credentials.
Why isn’t this the standard?
If hybrid models offer a better way to match talent with jobs, why haven’t they become the standard? Several entrenched barriers stand in the way:
- Outdated hiring norms: Many employers still default to degrees and job titles as shortcuts for assessing talent, making it difficult for skills-based approaches to gain traction.
- Fragmented data ecosystem: Job descriptions, resumes, and training credentials follow no common format, preventing seamless skill matching across platforms—a barrier also driven by profit incentives for proprietary data systems, as private vendors benefit from locking people into their platform (a governance failure, as noted below).
- High implementation costs: Building and maintaining AI-driven ontologies requires continuous updates, making adoption costly for companies and policymakers.
- Lack of policy mandates: Without regulatory or industry-wide incentives to make platforms accessible and interoperable, employers have little motivation to transition away from traditional hiring practices.
No single player can fix this alone—widespread adoption requires coordination across industries, government, and education.
Homework: Ensuring compatibility across hybrid efforts
For hybrid models to drive large-scale impact, they must be interoperable across industries, platforms, and regions. Without coordination, fragmented systems will only reinforce the very inefficiencies they aim to solve. A smart combination of taxonomies and ontologies is key to ensuring both structure and adaptability in skill-matching frameworks.
Key steps to achieving interoperability include:
- Aligning with standardized frameworks: Global competency models like ESCO, O*NET, and Credential Engine provide a structured foundation (taxonomy) while allowing for adaptability (ontology).
- Building API-based data exchanges: Seamless integration between job platforms, training providers, and employers is crucial for scaling skills-based hiring across different ecosystems.
- Leveraging AI for cross-system translation: NLP and semantic matching enable taxonomies and ontologies to interact dynamically, allowing skill classifications to evolve with market demands.
- Establishing multi-stakeholder governance: Public-private partnerships should oversee competency frameworks to ensure ongoing alignment between structured classifications and real-world labor market needs.
By integrating taxonomies for structure and ontologies for adaptability, hybrid models can provide a common skills language that scales effectively across industries. Prioritizing interoperability is not just a technical necessity—it’s the foundation for a more fluid, skills-based labor market where job seekers, employers, and educators operate within a shared, evolving skills ecosystem.
The path forward
Taxonomies and ontologies are not just classification tools—they are the backbone of a smarter, more inclusive labor market. When combined effectively, they provide both structure and adaptability, creating a common skills language that enables more precise, equitable, and scalable job matching.
Importantly, because AI-driven tools inevitably reflect the biases of their training data, architects must deliberately ensure that these tools are fair—in particular, that AI-driven matches do not simply emulate common matches in the past. Also, to the extent that matching algorithms make decisions based on digital footprints, strict privacy laws must give job seekers control over that footprint.
Realizing this vision requires action:
- Employers must integrate competency-based hiring into recruitment, pilot hybrid taxonomies-ontologies for job matching, and move beyond degree-based filtering.
- Policymakers must invest in open-source, standardized frameworks, incentivize skills-based hiring, and mandate interoperability between job platforms.
- Educators and training providers must align credentials with recognized taxonomies and ensure micro-credentials map onto in-demand skills.
- Technologists must embed AI-driven ontologies into unbiased hiring systems and improve cross-platform compatibility.
- Workforce organizations and researchers should advocate for policy change, conduct research on skills-based hiring, and push for industry-wide adoption of hybrid frameworks.
Without collective commitment and a fair use of new technologies, hiring inefficiencies will persist, reinforcing barriers to opportunity. The future of hiring must be driven by what you can do—not where you went to school or who you are familiar with.
For the millions of workers stuck below the traditional hiring radar, this shift isn’t just desirable—it’s urgent. Employers, policymakers, and educators must act now to build a labor market where skills—not credentials or networks—determine opportunity.