Posted on November 22, 2017

Extract! 4.0 release: Deep Learning update and much more!

We are proud to announce an important technical breakthrough in the field of CV Parsing. Extract! 4.0, the first fully Deep Learning powered CV parsing solution, has just been released! This release is an important milestone since it marks the beginning of upgrading all major Textkernel products with our new Deep Learning technology. This release also includes several improvements to our English model.

Extract! 4.0: enter a new era of CV parsing

The following improvements have been made to the English parsing model:

  • Overall decrease of extraction errors by 15-30% thanks to Deep Learning
    • Improved extraction of personal experience and education sections.
    • Improved extraction of name, address, e-mail address, phone number and date of birth.
    • Improved extraction of job title, location and organization fields as well as experience items.
    • Improved extraction of education level, degree direction, institution and degree dates as well as education items.
  • Extract! 4.0 recognizes rare sections better, which were beforehand sometimes extracted as an education or experience section. Publications, conferences etc. will now no longer be extracted as an experience item.
  • Extract! 4.0 is now able to extract the major of an education item.

The following improvements are available for all languages:

  • The handling of CVs with column layout (for all languages) has been greatly improved.
  • Github, soundcloud and dribble have been added to extracted social media profile links.
  • Personal website URLs can now be extracted.

Extract! 4.0 is available as of today! All major languages supported by Textkernel will be updated with Deep Learning in the near future.
Please contact us should you have any questions concerning Extract! 4.0.

.