With few exceptions (e.g. Just-Posted and Itzbig), search technology has remained essentially unchanged for the past ten years – we still enter a few keywords, we still hope employers are using the same words, and we still sift through volumes of marginally related results – or worse, we stare at some variant of the message ‘no jobs match your search criteria.’
The problem is simply this – keyword-based search is a poor proxy for intention-based search.
Intention-based search differs dramatically in that each search has two equally important components:
¾ First, search terms/phrases entered by the user are interpreted by the system to identify the intent or theme of the search; and,
¾ Then, content matching the intent or theme is displayed … it doesn’t matter if the original search terms are included in the ad text, so long as the ad falls within users’ intent.
The image at right illustrates the following example of intention-based search from JobCentral’s new search engine:
First, I select the ‘Occupation Title’ search option.
Second, I enter words that describe the type of position I’m seeking … note that I’ve chosen to use ‘medical lab tech’, but I could have entered skills or duties.
Third, the system interprets my search terms and offers five occupational themes for me to choose from … in this case I’ve chosen ‘Medical and Clinical Laboratory Technicians’. As a side note, this is a standard ONET occupation for which significant labor market information exists, including employment growth projections and possible employers … more on this in a follow-up post.
Fourth, the JobCentral system then returns ads within this occupational area. In fact, no keywords were used to generate this list once I selected the occupational category. Cool!
A scan of the results looks pretty good, but not perfect … yet. This first release of JobCentral’s new search engine is just beginning to leverage their structured data; so accuracy should improve as the engine is refined and additional metadata elements are incorporated.
I think the same can be said regarding the interface … it’s a bit complicated, and I noticed a few quirks. For example, between steps three and four I had to manually remove my search terms from the keyword box to search for all ads within the occupational category.
No matter … JobCentral’s new intention-based search is a big step towards the goal of a more intuitive search process that returns highly relevant search results. Kudos JobCentral!
Note: JobCentral uses O*NET-SOC AutoCoder software from R.M. Wilson Consulting, Inc. to assign occupational metadata.