Interview with SkyCat developer

An Ad

Integrate business news on your website free of charge and offer your visitors up-to-date content! Click here for more information.

As reported earlier this week, Hungary-based Sky CAT will soon release a new computer-assisted translation suite of the same name. Since the company website is still under construction, babelport.com contacted Sky CAT to find out more about this tool.

Below you will find the interview with István Lengyel of Sky CAT and SiteFounder Christian Hansel. A product review will follow as soon as we have tested the demo version provided by Sky CAT.

CH: What versions of Sky CAT will be offered?

IL: There will be four versions, one being the unpriced Corporate version. The Server version (1499 Euro) will include a server and five networked "mobile" licences (i.e. you can activate and deactivate them, so it's limited by the concurrent number of installations, not the total number), and that is our "unbeatably priced product". The Standalone version will be available for 399 Euro. Furthermore, there will be a Freeware version.

CH: When do you expect the product to be finished / marketed?

IL: The product will be launched at Localization World [October 25-27, 2005 in Seattle], marketing starts end of November / early December.

CH: How do you plan to market the software?

IL: We will set up a separate website for Sky CAT and it will also be marketed in a one-stop CAT shop. The freeware version will actually contain two licences: a 30-day unlimited licence, and once it expires, an infinite freeware licence.

CH: In what ways has machine translation experience influenced the development?

IL: Work on a complex rule-based machine translation system has brought us two fundamental experiences.

First, we have gained the IT know-how to develop robust and efficient natural language text manipulation software. One of the designers of Sky CAT developed the core linguistic module of MorphoLogic's MT system, which has been used as an on-line service for the past year without interruption. (Moose: a robust high performance parser and generator. EAMT 2004, Malta)

Second, our insight into computer linguistics theory has allowed us to create a unique system of indexing for fuzzy lookup which we believe yields hits that are closer to what humans perceive as a "similar sentence." While our present solution does not contain any natural language processing, we will move on to enhance our indexing method with language-aware features that will yield even better results - less annoying rubbish, and more truly relevant hits.

CH: Are any MT-features to be integrated in the CAT tool?

IL: On one hand, both SMT (statistical MT) and rule-based MT will be integrated with Sky CAT; on the other hand, we will try to come up with a new rule-based MT approach to correct TM hits and retrieve more value out of previous translations.

We want to set up a worldwide network of natural language processing specialists to get the most out of languages - and not just plain meaningless statistics. We will open up our architecture for add-ins that create value and increase productivity.

CH: What's the size of the development team?

IL: Sky CAT was developed by Kilgray, and the development team consists of ca. 10 people from the Budapest University of Technology, one of the most reputable Microsoft development centres in Central-Europe. We will immediately continue development after the release of the first version.

CH: What do think distinguishes Sky CAT from other translation memory systems like TRADOS / SDLX / across / wordfast?

IL: Its ease of use, the fact that you can learn it very quickly (in ca. 1 hour everyone can get access to 95% of its functionality), its powerful aligner and configuration, and its open architecture.

Related articles: New CAT Tool Sky CAT to be released soon

Related news: SkyCAT beta testers sought