It seems there are some obvious USP’s that make Actonomy’s xMP best of breed in the searching technology business. In a series of blogposts, Actonomy’s CEO Filip De Geijter explains the details of these USP’s. After his posts on flexibility, completeness and transparency, De Geijter now focuses on accuracy.
Correctly understanding context and meaning
For searching and matching technology to be top notch, flexibility and transparancy are elementary: technology is useless when not flexible in its capabilities or when not transparent about the way it creates matches. But another quintessential element is accuracy. To put it simple: when a recruiter is searching a software developer with knowledge of Java, he/she is not primarily interested in a candidate who has traveled to an Indonesian island. ‘Knowledge of Oracle’ does not mean ‘knowledge about the company that once was started by a sailor called Larry Ellison.’
In order to be accurate, searching and matching technology, such as our xMP-technology at Actonomy, needs to fulfill two criteria. First, it should understand the meaning of words in a given context. Java and Oracle get their specific meaning due to the recruiting context. If you understand the exact meaning of words, you can more accurately match both job profiles and candidates. Unless you want to recruit a secretary who has been working at the Oracle headquarters, that is.
Second, in large ontologies you should not normalize too much. Rather than normalizing ‘sales manager’ and ‘area sales manager’, good technology will see and understand the difference between these two job titles. The more distinctions you can understand, the more accurate your matches will be.
One of the more outspoken advantages of accuracy is the gain of efficiency. Automatic matching may save time, but it is costly when the outcome is of average quality. Thanks to accuracy, automatic matching provides most efficiency, as the resulting matches are of high quality.
However, it remains difficult to measure how ‘accurate’ searching and matching technology is. We at Actonomy use two methods to judge accuracy. On the one hand, we compare the results of our xMP-technology to human recruiters: our 10 best candidates should be very much comparable to the 10 best matches delivered by human recruiters. And what is more, our technology does provide matches that human recruiters do not even find. A human recruiter looking for an online marketer may not know what SEO is and – as a consequence – not see the link with online marketing. Actonomy’s fine-grained, huge ontology does know what SEO is and will therefore find potential matches.
Continuously increasing accuracy
One last thing should be mentioned when zooming in on accuracy. And it is good news: as you could read in an earlier blogpost on Actonomy’s deal with Vivaldis, our ontology gets wiser and wiser. Thanks to machine learning, it continuously increases its accuracy. The more our xMP-technology is used, the more accurate it will get, and the better the matches between candidates and job profiles. We are good, but will get even better: sorry for sounding a bit impertinent…
In our fourth and last USP-blog, we will focus on completeness.