Developing CV parsing is a journey towards perfection and everyday, our engineers strive to improve the quality of CV parsing. In this release (2015.2), the main focus was set on improving the English CV parsing model. You will find a list of these improvements below as well as other parsing model updates included in this new version of the software.
Improvements to existing CV parsing models
Improvements to the English model include:
- A more accurate segmentation and extraction of education items
- Improved detection and parsing of USA addresses
- Better extraction of language skills and proficiency
Improvements to other languages include:
- Better extraction of nationalities in German CVs
- Refined extraction of names in Czech CVs
Improved extraction of emails with a + sign for all languages
Candidates sometimes customise their email address in order to keep track of a certain correspondence and set up filters in their mailbox (firstname.lastname@example.org and email@example.com will both be functioning correctly since the information after the plus sign is automatically ignored by email clients). Textkernel’s CV parsers are now able to correctly extract such emails .
New: degree and institution type normalisation for Chinese
The Chinese CV parser is now able to provide normalised degree and institution types from education items as seen in the table below. By categorising education degrees and institution types, it becomes easier for recruiters to find candidates with a certain education level.For more information on this release or about Textkernel’s CV parsers, please contact Textkernel.