Smarter Search and Matching
This release delivers a set of powerful improvements to the Talent Catalog’s matching engine, helping users find the most relevant candidates faster and more accurately.
Search results are now sorted by closeness of match, keyword matches are highlighted directly in candidate CVs, and users can now generate searches directly from a job posting — with relevant skills automatically extracted and used to find matching candidates.
Behind the scenes, we’ve transitioned to using PostgreSQL’s full-text search to handle keyword matching, replacing the need for a separate Elasticsearch service. This simplifies system architecture, improves data consistency, and reduces operating costs.
Together, these enhancements make matching more intuitive, responsive, and scalable — setting the foundation for future capabilities like automated candidate outreach and intelligent job recommendations.
Prioritised Matching
Results of a search that contains keywords logic is now sorted by closeness of match.
Generate a Search Directly from a Job Description
Click on the search icon for any job…
The TC will scan all text related to the job - uploaded job description, job summary, job intake etc extracting skills from the text.
We have a database of around 30,000 skills extracted from the ESCO and O*NET collections.
The extracted skills are automatically added to a New Search which searches for candidates with those skills.
You can add or remove skills if you wish. You can also modify the Keyword search as usual to construct boolean expressions - for example requiring this skill AND that skill.
Global Skills Standards and AI-Powered Extraction
This release also marks the introduction of a new, structured skills framework based on two leading global standards:
- ESCO (European Skills, Competences, Qualifications and Occupations)
- O*NET (Occupational Information Network, USA)
Together, these sources contribute over 30,000 standardised skills used across industries and professions, now integrated into the Talent Catalog.
An AI-powered extraction service has been introduced to automatically identify and apply relevant skills from job descriptions. This in turn drives AI-generated candidate searches that are based on consistent, structured skill data — improving search relevance and will assist in skills comparability across job types.
💡 In future releases, we’ll expand both the skills base and the use of AI — including support for skill extraction from candidate-submitted experience, and smarter prompts to help candidates describe their skills more effectively.
Highlight Search Keyword Matches in Uploaded CVs
Keyword search matches are now not only shown highlighted in the entered Experience and Education data…
… but keyword matches are also shown in any uploaded Cvs
Improved Text Search Infrastructure
This release introduces the use of PostgreSQL’s full-text search extensions to power keyword search logic within the Talent Catalog.
Postgres now handles text indexing and relevance scoring which can be used for more efficient candidate profiles and uploaded CVs searches.
Previously, text search was handled through Elasticsearch. By consolidating this capability into Postgres, we can:
- Simplify the system architecture by removing a separate Elasticsearch service
- Reduce maintenance overhead and streamline data consistency
- Save approximately US $2,500 annually in licensing and hosting costs
This change is an important step toward fully retiring Elasticsearch in a future release, and handling all core search capabilities natively within Postgres instead.