On page 16 of the JobCentral National Labor Exchange presentation you’ll find a graphic illustrating that jobs will be indexed daily and O*NET-coded for distribution to States.
The image at right is a slightly modified version (for visual clarity). It shows that States will have a rare advantage in this new National Labor Exchange – they’ll be receiving job ads with detailed occupational structure.
Assigning O*NET codes to jobs was an expensive proposition until R.M. Wilson Consulting, Inc. developed an accurate automated process.
Yes, O*NET coding software has enjoyed limited availability for several years; but none have delivered 90+% accuracy (to my knowledge) – until now.
So what’s the big deal?
The point of classifying jobs is to make it easier for the right candidates to find the right jobs.
Recall that the top search on Jobster is for the single word ‘sales’. The result, as is the case on almost all job sites, is a long unstructured list of jobs – a sales clerk job is just as likely to be followed by a sales manager job, as it is to be followed by a telemarketing job. Without structure, search uses a bag-of-words approach without context.
In contrast, look at the detailed occupation codes provided by O*NET for sales jobs.
If job ads are classified using this structure, then when a jobseeker searches for ‘sales’, they can be presented with the list of choices to help them refine their search.
As an example of the power of this approach, let’s assume that a jobseeker is looking for a position as a non-technical manufacturer’s sales rep.
This individual doesn’t want to miss any of the relevant opportunities, so the initial search is for the word ‘sales.’ Predictably, this generates a flood of poorly targeted ads.
But, if the occupational subcategories within sales are presented as a refinement option, then the jobseeker can choose ‘Sales Representatives, Wholesale and Manufacturing, Except Technical’ to see a precisely targeted list of ads that have not been artificially constrained by the choice of search keywords. The results are really stunning.
If you’d like to learn more, contact JobCentral, drop me a note, or check out my O*NET-SOC AutoCoder blog.
Comments