There is one fundamental problem at the heart of every job search and every recruiting drive - the search for jobs and the search for candidates. Google tries to address the first part - with an improvement in the search for jobs. It has received good validation and is being tested by veritable recruiting vendors CareerBuilder, Dice and Jibe.
Not surprising for search giant Google to pick this challenge. Also not surprising it all starts with ontologies, Google uses two of them: A three tiered occupation ontology (based on the O*Net Standard Occupational Classification) and a skill ontology of over 50k hard and soft skills (how important they are see e.g. here - Workday acquired Identified). The magic then lies into mapping these two to each other, which Google does in a third step.
 |
| A visualization of the occupation ontology - from Google web site - here |
It's the third step where Machine Learning really unfolds its power. Google states that its vector for job titles is 100k dimensions big, based on an analysis of 17M job posts. Beyond human grasp and likely also beyond any single project based modelling by a single or team of data scientists.
MyPOV
Good to see Google using expertise and moving it towards enterprise applications. One more offering that helps Google move the conversation towards the enterprise, something the vendor desparately wants to foster, grow and ultimately monetize. Good to see uptake and pilots by CareerBuilder, Dice and Jibe. Ultimately also another avenue to get more load into Google Cloud, something the vendor is equally interested in. Finally it is likely the API can be applied also for the reverse search - finding the right candidate - but that's something we will have to wait for Google to provide. In that case some privacy questions will loom...
On the concern side it is one more component that may break a Talent Acquisition system. But probably Google can scale more and better than the data science teams at relatively smaller vendors, ultimately given the overall system more stability.
Finally it's an inflection point. Google has probably a lead when it comes to overall machine learning from both an algorithm and platform (with TensorFlow and GPU architectures) - but it has not forayed into enterprise software. Congrats on the first step and we will be watching.
More about Google:
- Event Report - Google I/O 2016 - Android N soon, Google assistant sooner and VR / AR later - read here
- First Take - Google Google I/O 2016 - Day #1 Keynote - Enterprise Takeaways - read here
- Event Preview - Google's Google I/O 2016 - read here
- Event Report – Google Google Cloud Platform Next – Key Offerings for (some of) the enterprise - read here
- First Take - Google Cloud Platform - Takeaways Day #1 Keynote - read here
- News Analysis - Google launches Cloud Dataproc - read here
- Musings - Google re-organizes - will it be about Alpha or Alphabet Soup? Read here
- Event Report - Google I/O - Google wants developers to first & foremost build more Android apps - read here
- First Take - Google I/O Day #1 Keynote - it is all about Android - read here
- News Analysis - Google does it again (lower prices for Google Cloud Platform), enterprises take notice - read here
- News Analyse - Google I/O Takeaways Value Propositions for the enterprise - read here
- Google gets serious about the cloud and it is different - read here
- A tale of two clouds - Google and HP - read here
- Why Google acquired Talaria - efficiency matters - read here
