Here at PI, we like to humanize the search process for lots of different categories of knowledge. Two projects we worked on last year epitomized the need that many information providers have to make sure that their search keywords and taxonomic structures are complete in the sense that they have enough of the right sorts of terms to get the users to the information they want, and to the providers (of jobs, services, products) who want to serve them. Today, I’ll cover the first of these projects where we contributed to a major expansion of a jobs listing taxonomy.
What’s in a Job Title?
The first project we worked on last summer turned out to be a fascinatingly complex taxonomy expansion for a job directory. The client was looking to expand their business to other English-speaking countries outside the USA and bumped up against the almost infinite variety in human labeling and categorizing capabilities. I was asked to assist a seriously brilliant semantic architect (new term for me; I like it!) to map American job titles to their European counterparts, which turned out not to be easily intuitive, thanks to the way humans have organized industries and trades since the nineteenth century.
The main challenge was to figure out what level the different job titles related to, as in what educational prerequisites were involved for a given title in different countries (e.g., engineer or engineering technician? It depends…). Oh, my! We had to build a complex hierarchy to include everything from nurses to welders to shepherds and decide on the level of match between U.S. and European counterparts. This part reminded me of book indexing because of all the judgment calls involved based on an understanding of the facets of meaning for given job titles. Amazing learning process for me personally, and a great opportunity to use my inner thesaurus to make appropriate connections and add to the depth and breadth of information to match job searchers with providers.
I think that even Watson, the venerable artificial intelligence system that played Jeopardy, would have had trouble understanding the partial semantic overlaps in the totality of occupations that humans find themselves in. And having not just one human, but several, sharing background information and researching the nuances using their complex brain power was crucial to the success of this project.
PI Pick of the Week
A perusal of LinkedIn’s news feed picked up this cool article by Marcia Reifer Johnston (with thanks to colleague Dave Ream at Leverage Technologies for reposting) on the meaning of “automatically discoverable” and the importance of intelligent content. Check it out here.