Textkernel is constantly working on improving the existing resume parsing models for both CVs and vacancies in multiple languages. We use different state-of-the-art machine learning techniques to refine the status-quo. You can find out more regarding our research and development on our Textkernel blog.
The new Extract! release contains the following improvements:
CV parsing improvements:
- All languages:
Textkernel’s CV parser can automatically count the years of experience. This can be very useful when you search for candidates with a minimum of e.g. 2 years of work experience. Within our new release we can also detect ‘gaps’ in the cvs and subtract years of unemployment.
- English: Textkernel can now also parse Indeed PDF CVs and LinkedIn HTML formatted CVs
In addition to the existing supported document formats, Textkernel can now provide support for parsing of Indeed PDF CVs and LinkedIn HTML CVs
- Dutch: improved title detection in Dutch names and improved extraction of postal codes
- French: improvements of the parsing results for Canadian CVs (education normalization, address parsing, phone numbers)
- Spanish: improvement of the extraction of addresses, dates and IT skill levels as well as the classification of education levels
- Czech: added the extraction of salaries
- Swedish: improvement of the extraction result for personal fields (such as names, addresses and phone numbers) as well as the classification of education levels
- Portuguese: improvement of the classification of education levels
Vacancy parsing improvements:
- All languages: improved classification of contract types
In order to filter and search on contract types (permanent, temporary, internship etc.) data out of vacancies needs to be mapped to the right contract type. This has been improved within the release
- German: improved extraction of locations in Germany and Austria
- English: improved extraction of phone numbers and US date formats
- English and French vacancy texts: improvement for the extraction of salaries
All our SaaS customers will automatically benefit from these improvements.
Do you questions regarding this new release? Contact us directly!