Sovren Resume Parser

First developed in 1996, the Sovren Resume Parser is the world’s most popular candidate parser.

Why it’s so popular:

  • Configurability. Parsing is configured per transaction, meaning that every single parsing transaction can be individually customized to do what you need, with the data you want to use, and passing back the results in the format you desire. One parser instance handles every language, every locale, and every type of configuration.
  • Coverage. Extracts more kinds of data than any other product.
  • Metadata analysis. Sovren was the originator of candidate characterization, and the Parser produces the most comprehensive set of data (machine readable and also human readable) to help you understand and classify candidates at any level of granularity.
  • Flexible deployment. The Parser can be deployed as installed components or via web services or licensed as SaaS, and can therefore be used from any OS/programming environment.
  • Scalability. Sovren powers the largest online and offline resume processors in the world. Our speed is typically 2-10x faster than our competitors’ products, and the Parser does not crash, leak memory, or require monitoring software of any kind.
  • Simplicity. One instance can handle every language and locale, with automatic detection and self-configuration, saving you resources, time, and aggravation.
  • Value. As we add languages and locales and features to the Parser, you don’t have to pay extra. The Parser started as with one language and one local, and now supports dozens of languages, dialects, and locales — and no Sovren customer has ever had to “pay extra” for these new capabilities.
  • Accuracy. All vendors claim to be the most accurate. Don’t EVER believe a vendor’s claims without testing them yourself. We have every confidence that once you test, you will see what all our customers have seen: that Sovren is truly the most accurate.
  • Accelerators. Lots of sample apps and web clients are available in many programming languages, with full source code.
  • Universality. The Parser’s built-in Sovren Document Converter allows the Parser to process documents in every commercially used format.

The Truth about Accuracy

When it comes to the accuracy of our Parser, the best verification is that the largest online and offline resume/CV processing organizations in the world use Sovren software.

Other parsing vendors trumpet absurd and unfounded claims about accuracy. All vendor claims about accuracy are self-serving and indefensible. Measured by whom? Measured how? Measured on what documents? Measured in what language? Measured on which version?

One minor vendor claims 100% accuracy. Read the fine print and you’ll learn that their parser is 100% accurate when the candidate corrects the errors. Well, of course.

If you are interested in the Sovren parsing technology, we recommend that you take a real life sample of the type of resumes you will be parsing and test them on your own. You will see firsthand that the Sovren parsing technology is the most accurate parser.


As with accuracy, other vendors make completely absurd claims. One vendor claims that their hosted software handles “tens of thousands of resumes at one time in under a second.” Just to be clear, that would be equivalent to processing at least 2,000,000,000 (yes, two billion) resumes per day, or trillions per year. (For the record, this is the same vendor that claims 99.9% accuracy).

The truth is that only the Sovren Resume Parser has been proven to handle the largest real-world loads. The largest online resume site and the largest offline resume processor both use Sovren.

Languages and Countries

The Parser presently supports many languages, all within the same version of the product. Several languages are being added each year. Full postal address parsing is supported in many countries, as well as local cultural conventions, companies, schools, etc. Name, phone number and email parsing are supported for all locales.

Chinese (Simplified)
English, all dialects
French, all dialects, including Canadian French
German, all dialects including Switzerland, Lichtenstein and Austria
Italian, all dialects
Portuguese, all dialects
Russian, including Belarusian
Spanish, also Catalan, Galician, Basque

Region support for Luxembourg, all of South America, Mexico, Poland, and Romania.  Language and region support for Danish, Polish, Romanian, and Flemish.

Supported Countries
Supported countries include Argentina, Australia, Austria, Brazil, Belgium, Canada, Chile, China, Czech Republic, Denmark, Finland, France, Germany, Greece, Hong Kong, Hungary, India, Ireland, Italy, Lichtenstein, Netherlands, New Zealand, Norway, Poland, Portugal, Russia, Singapore, Spain, South Africa, Sweden, Switzerland, United Kingdom, and United States.  Additionally the parser recognizes major international cities in other countries.

Learn More...

Sovren Semantic Matching Engine

Simply the most configurable and powerful matching engine available anywhere.

First developed in 1996, the Sovren Semantic Matching Engine (“SSME”) is the world’s most sophisticated semantic profile searching and matching engine.

The SSME is both a search engine and a matching engine. It includes all the full text and Boolean capabilities of traditional search engines, while adding semantic capabilities. Thus, the SSME does not require you to make compromises: you keep all the familiar capabilities of traditional engines, while adding the incredible power of semantic profile searching and matching.

Matches profiles, not words

The Sovren Semantic Matching Engine (“SSME”) understands candidates as people with specific career profiles, and understands jobs the same way. Unlike traditional engines that just look for words, the SSME looks for profiles. Instead of finding “Java” and saying “We have a match!!!”, the SSME understands that the job posting is looking for a midrange Java programmer who is bilingual in English and Spanish, and then looks for candidates who exactly match that profile. In addition through the skills hierarchy that ships the resume parser Taxonomies and Skills can be configured in any way that you would like enabling the configurable and powerful matching engine anywhere.

Eliminates over 90% of false matches

Because the Sovren Semantic Matching Engine matches profiles and not words, it dramatically reduces false positives. Instead of finding “Java” and saying “We have a match!!!”, the SSME understands that the job posting is looking for a midrange Java programmer who is currently a using Java as a skill indentified in current work experience. Want to add who is bilingual in English and Spanish, the matching engine will then filter and then looks for candidates who exactly match that profile. 90+% of Java programmers do NOT match that profile, yet ALL traditional engines would show ALL of those Java programmers as matches. But they are NOT valid matches, the Sovren Semantic Matching Engine knows that and will not return those false positive matches.

Captures the most qualified applicants

When the Sovren Semantic Matching Engine finds a search term, it evaluates all of the relevant criteria. It investigates the context, values and meaning of the search term to evaluate whether it is an actual match or just a false positive. The more criteria that a candidate matches, and the better the candidate is on those criteria, the higher the candidate will be ranked by the SSME.

Built for recruitment, not adapted from something else

Our engine was built specifically for resumes and jobs, not for some other industry and later adapted for the recruitment industry.

Allows weighting of candidate criteria

The SSME allows recruiters and candidates to set their own weightings for each criterion, ensuring that the matching engine “thinks” like the human user thinks when finding the best matches. Is this a job where having a specific educational degree is an absolute prerequisite? Then set the Education criterion to be the heaviest weighting, and the SSME will act accordingly, treating Education as the most important part of the matching process.

Let’s say you need someone who is indeed currently a Senior Accountant as compared to someone who used to be or wants to be a Senior Account. There is a high likelihood that a current CFO was a Senior accountant at one point in her career, but you wouldn’t want to contact her now with that position.

Like the Sovren parsing technology, the Sovren Semantic Matching Engine is the most configurable matching engine on the market. It enables searches to be re-weighted and re-run in real time to maximize the scope of the search or make it pinpoint specific.

Automatically matches in real time

The Sovren Semantic Matching Engine enables jobs and candidates to be automatically matched in real time on 16 different dimensions of “fit,” with no human intervention needed.

No other engine in the world can operate on even half that many dimensions and most are closed boxes, built to do “magic” that no on can explain.

Clones candidates for the fastest matches

Using the Sovren Semantic Matching Engine, recruiters can “clone” candidates by finding candidates who are similar to other candidates. This technique is easier and faster than creating job descriptions to use for matching.

Clones jobs, too

Find a job that is a good match? Great! Now let the SSME find other jobs that are similar, and show you the percentage of similarity, too. For example, a candidate builds a profile and you recommend jobs to that candidate. They apply to a job but maybe it was not the first one the engine recommend it was the 9th one out of ten. You can now change your approach and use the job they applied to and match against it. “Based on the job you just applied to here are jobs that are like that job”. Nothing is more impressive to a candidate when they get a sense that the engine know who they are and what they are looking for. You can build a search agent, the possibilities are endless.

Stability and Security like no other

Sovren software does not crash, hang, leak memory, or otherwise stop working or require special monitoring.

Learn More...

Sovren Job Order Parser

The Sovren Job Order Parser (“SJOP”) is included as an integral part of the SSME, but is also available as a separately licensed product delivered via SaaS.

The SJOP extracts relevant data from a job order to build a job order profile for use by the Sovren Semantic Matching Engine.

The job order profile will contain information to describe job titles, required skills, optional skills, years of experience desired, years of management experience desired, management level, educational requirements, certifications, foreign languages, etc. — 16 categories in all.

The precise specifications in the job order profile can now be used to find candidates whose profiles match the job order profile, or to find similar jobs.

The parsed job order can also be used as the source document or “Gold Standard” to match against other jobs that the candidate has applied to OR to create a search agent to recommend jobs to a candidate based on jobs they have applied for.

Sovren Data Normalizer

The Sovren Data Normalizer provides built-in and user-customizable standardization of parsed resumes.

The SDN normalizes locations, company names, job titles, school names, and degrees. Although the SDN includes built-in functionality, the user is in complete control and can specify custom behavior. For instance, a user may wish to normalize any company name that includes “Amazon” and that is not a restaurant, bar, etc., to “AMZN”.

The SDN normalizes data into HR-XML user areas, meaning that the original data AND the normalized data are both returned in the HR-XML, so that the original data is not lost.

Facebook IconTwitter IconVisit Our LinkedIn Page