How We Classify Job Titles

The Occupation Standard Classification for Australia (OSCA) is the ABS’s official framework for classifying every job into a 5-level hierarchy: Major Group (1 digit), Sub-Major (2), Minor (3), Unit Group (4), and Occupation (6). The Talent Map maps every employment record on a candidate’s profile to an OSCA code — but how confident we are at each level varies by title.

1. Lookup cache: if we’ve classified the same title in the same industry before and we’re confident the title’s meaning is unambiguous, we reuse that result.
2. Official ABS alternative titles: the OSCA framework publishes alternative names for each occupation (e.g. “RN” → Registered Nurse). Direct matches take this path.
3. Title pattern matching: exact and fuzzy title matches against the OSCA reference.
4. Context-aware AI: for ambiguous titles, we send the title plus the candidate’s industry (ANZSIC division), employer description, job duties, education field, and other roles they’ve held to a language model trained on the OSCA rules. This is what handles “Site Manager”, “Field Technician”, “Specialist”, and similar context-dependent titles.

A correct partial code beats a confidently-wrong 6-digit code every time. If the evidence only supports the 4-digit Unit Group level, we stop there.

• Police Officer → 6-digit (441312). The OSCA Unit Group for Police only contains one occupation, so the 6-digit code is just the disambiguated form of the 4-digit code. There is no ambiguity to resolve.
• Social Worker → 4-digit (2725) is a confident answer; the 6-digit specialisation depends on context that may not be present in the record.
• Site Manager → must use full context. Could be a Construction Site Supervisor, a Quality Assurance Manager, an Environmental Site Manager, or several others depending on the employer’s industry and the candidate’s duties.

Some occupations are regulated, licensed, or named so specifically that the title is the occupation regardless of employer. Registered nurses, electricians, solicitors, paramedics, and pharmacists are examples. We cache these confidently.

Other titles are role-shaped words (“Manager”, “Officer”, “Coordinator”, “Specialist”, “Lead”, “Trainee”, “Engineer”) whose OSCA code genuinely depends on industry, duties and education. We never serve these from cache by title alone — every classification gets the full context-aware treatment.

“Board Member”, “Committee Member”, “Volunteer”, “Student Representative”, and bare “Consultant” without a specialty word are not occupations in the OSCA sense. We exclude these from our occupation counts so they don’t inflate or distort workforce planning numbers. They remain in the candidate’s history; they’re just not classified as paid work.

The ABS publishes a free job-title classification API that maps titles to 6-digit OSCA codes. The Talent Map uses that API as one signal among several. The ABS API has no employer context — it can’t tell you whether a “Field Technician” works in mining or telecommunications, whether a “Specialist” has a nursing or accounting background, or whether a “Site Manager” is in construction or quality assurance.

Our pipeline considers the candidate’s industry, employer description, job duties, education field, and other roles they’ve held before assigning a code. For unambiguous occupations the ABS API is fine; for the long tail of generic titles, the context-aware approach materially improves accuracy. Where the evidence doesn’t support a specific 6-digit code we honestly stop at the 4-digit unit or 3-digit minor group rather than guessing.

When you filter the Talent Map for “Project Officer”, you get candidates whose duties were actually those of a project officer in their industry — not everyone who happened to have those words in their job title. That precision is what makes feasibility scores, network strength badges, IPP compliance reports, and RAP workforce gap analyses defensible to your board, your tender authority, and your stakeholders.

The classification methodology evolves over time. We re-evaluate ambiguous titles as language models improve and as the underlying ABS reference data is updated. If you ever see a candidate count or classification that looks wrong, your Shine consultant can help interpret the edge case.