Classify

Overview

The classify tool converts raw user text into canonical taxonomy IDs for classifiable dimensions. It helps agents convert terms such as occupations, skills, companies, locations, industries, schools, benefits, and education programs into IDs that analysis tools can use.

Common Use Cases

Use classify tool to answer questions such as:

Use Case	Example
Occupation normalization	What is the canonical SOC ID for software developers?
Skill normalization	What is the canonical skill ID for Python?
Location normalization	What is the canonical city ID for Seattle?
Agent normalization	Convert user language into canonical IDs before analysis.
Filter construction	Build valid filters for classifiable dimensions.
Ranking setup	Select the correct taxonomy level for `top` rank dimensions.

Supported Datasets

classify returns results for the specified dimension. Supported concepts and taxonomy values differ across dimensions.

Name	Dimensions
Job Posting	`benefit`, `occupation_lot`, `postings_education_level`
Profile	`education_program_cip`, `education_school`, `industry_naics`, `location_us_area`, `occupation_onet`, `occupation_soc`, `profiles_education_level`, `skill`
Job Posting & Profile	`company`, `location_us_city`

The identifiers returned by classify can be used in supported filters and ranking requests across datasets, depending on the selected dimension and tool.

Inputs

The tool accepts the following objects.

Input	Description
`dimension`	Classifiable dimension name.
`level`	Numeric taxonomy level for the selected dimension.
`value`	Search string to classify.

Use dimension, level, and value together to find the taxonomy identifier for a specific concept. For detailed information on taxonomy levels, see Taxonomies.

Example Request

{
  "dimension": "occupation_soc",
  "level": 4,
  "value": "software developers"
}

Outputs

The response returns one or more matching taxonomy concepts.

Output	Description
`results.id`	Canonical taxonomy ID.
`results.name`	Canonical concept name.
`results.level`	Taxonomy level.
`results.confidence`	Optional confidence value where available.

The returned identifier can be used in supported filters and ranking dimensions.

Example Response

{
  "results": [
    {
      "id": "231715",
      "name": "Software Developers",
      "level": 4,
      "confidence": 0.98
    }
  ]
}