CVlizer – optimal Optimizing

CVlizer – optimal Optimizing

Improvements in quality needs time

When you’ve developed a product like CVlizer and you want to improve the quality further, you can only do one thing: Optimize, optimize, optimize. But how does it work? What does it take? And especially, why does it take so long until a flaw in the extraction of a specific CV is corrected?

Parser Optimisation

It is no secret, no curriculum vitae is like the other. (Almost) Every CV is different, f.e. with dates: it might be written as “01.02.1934” or as “1934-02” or even “01.02.1934”. I won’t even talk about the complexity in detailing the skills and profession phases. To analyze these in millions of different CVs perfectly with all their individualities, that’s the challenge we’re confronted with every day.

Countless millions of CVs run through the CVlizer. It’s absolutely necessary to not focus on one individual CV, but to keep an eye on all of them, because a 100% optimization of one CV without considering them all leads to a decline of the whole extraction quality. This is abundantly clear in the example given earlier: If the format of dates is “01.02.1934” in one CV, it doesn’t necessarily lead to the conclusion that all the dates are always written this way.

The optimizing process

The basis of the optimizing process is formed by CVs which were provided by our customers. These CVs are automatically extracted. Subsequently one of our employees rectifies how accurate the extraction per CV was - this is very time consuming. A meticulous analysis of a complex CV can take up to an hour and in our test sets we work with thousands of CVs. Machine learning always requires a manual correction since the algorithms only learn from their mistakes and our corrections. To add insult to injury, our test-CVs have do be deleted and replaced due to deletion deadlines set by the data privacy regulations.

Why are submitted CVs still not identified perfectly after analysis?

This one CV of this one client of ours is only one of thousands in our sets for the testing. If an abnormality is detected, this one CV is being marked, being analyzed according to that abnormality and processed. The goal cannot be that the improvement of one CV causes millions of other application documents to not be analysed as good as before. Therefore intelligent solutions and the involvement of the latest technologies, like the machine learning, are necessary. And it need time to secure the optimization of the quality improvement for all CVs.

Our aspiration stays the same: 100% accuracy for ALL CVs.


Contact us

Our Office

Wehrgasse 28 / Top 3+4

1050 Vienna

contact@joinvision.com

+43 (0)1 505 80 70

+43 (0)1 505 80 70 60

Drop us a line









JoinVision is a leading provider of multilingual semantic recruiting technology. With the two parsers CVlizer and JOBolizer, applicant documents and job advertisements are automatically recorded, analyzed and coded. Modules, such as HRclassifier, HRcapture and HRmerger, expand the possibilities to have all information immediately available as a standardized, structured candidate or job profile in XML format.

Connect with us

Latest Tweets


  • Das Thema aktuelle Trends im #recruiting ist natürlich unerschöpflich und lässt jede Menge Antworten zu:… https://t.co/cKxokfx00k
    Wed Jul 24 14:34:31 +0000 2019

  • Digitalisierung des Bewerbungsprozesses! Das BerufsInfoZentrum (BIZ) des Arbeitsmarktservices (AMS) hat sich ausfüh… https://t.co/7ipShbfRqk
    Thu Jul 18 07:03:21 +0000 2019