Rosette Cloud 1.12: New Languages for Entity Linking, Better Korean Named Entity Extraction, and Delivery via Docker!
We’re thrilled to announce the latest version of Rosette (1.12). This release features many exciting updates to our text analytics platform, including expanded language coverage, better accuracy, as well as new options for software delivery.
Entities: Linking expanded to more languages and better Korean
We’ve devoted a lot of focus to improving our support for named entity extraction and linking over the last few months and are proud to announce that we now support entity linking in 16 new languages. That means that if a language is supported for entity extraction, we now also support it for entity linking. That’s not all! It also means that we support recognition and linking of the 700+ DBpedia types we introduced in Rosette 1.11 in all the same 20+ languages. New to named entity recognition/linking? Click here for more information.
We’ve improved the accuracy of our Korean named entity extraction, largely through better handling of Josa (postpositions) and compound words. Korean is a morphologically complex language that often behaves very differently from English. For example, titles are often written as part of the same word that contains the entity they refer to, like 백설공주, which means “Princess Snow White”. Rosette now does a better job of recognizing these compound words and splitting them up into their proper component entities.
Morphological Analysis: Disambiguation for Hebrew and lemmatization support for Persian
We’ve improved analysis in Hebrew by adding disambiguation, a mechanism for more accurately choosing which of several candidate analyses is provided in the response. Woah, what does that mean? Basically, sometimes a word can have more than one potential “analysis” ie, it could be either a noun or a verb. In English, think of the word “spoke” for example. Disambiguation helps Rosette to smartly decide whether we’re talking about part of a wheel or the act of speaking in the past tense. We’ve also added lemmatization support for Persian.
Rosette Enterprise (On-Premise Users Only)
Rosette Enterprise is now available as a Docker container! Docker is an Enterprise-grade container service, which provides operating system-level virtualization. Basically, that means that if we ship you Rosette in a Docker container, you can run it any environment you want (as long you have Docker installed)! The Rosette Docker images are available on Docker Hub. Note that a Basis shipment, containing a license file and a docker-compose file customized to your licensed endpoints, is still required.
Check out the release notes for all the details. We look forward to your feedback! As always you can reach us at firstname.lastname@example.org!