These release notes are for the self-hosted offering. If you're looking for Sovren's SaaS release notes go here.
July 18, 2019
Accuracy improvements in all languages.
More accurate contact info parsing. Better phone number type classification.
More accurate LinkedIn parsing.
More accurate Employment parsing.
More accurate Skills parsing with new terms.
More accurate Education parsing. Far more accurate degree types in every language.
Added support for Resume Quality codes 121-124 for resumes from UK and NZ
Better parsing of Colombian national identity numbers.
Approximately 3%-14% faster than previous release.
Parsing throughput has increased over 50%.
Better detection and correction of reversed text.
More useful conversion of LinkedIn pdf profiles in the varied two-column formats.
Faster conversion of PDF to HTML.
Better trimming of leading whitespace preceding first non-whitespace character in converted text.
When documents take too long to convert, we now return converted text in many more instances, whereas before we returned no text.
Improved messaging for possible conversion errors.
New Output Validity: ovLinesSeemTooShort.
Added messages for output validity warnings and errors.
Upgraded third party DLL versions:
- Aspose Words 19.4
- Aspose PDF 19.4
- dtSearch 7.93.8596.18093
Improved non-English Sovren-generated candidate summaries.
Fixed bug that caused some jobs to be duplicated when reported on some German CVs.
Corrected situation where some non-binary data was being detected as binary data and being reported as ovProbableGarbageInText.