Double challenge: accuracy and speed

 

The ICE technology performs analysis and versatile categorization of text input, regardless of origin and format. Detected categories cover all parental control -related topics and semantic areas.

ICE delivers over 98% categorization accuracy* in five major western languages (English, Spanish, French, German and Italian) into 23 different categories from porn to gambling and social networks, with over 99.9% protection via language detection (30+ languages covering all continents).

Designed with speed as a primary requirement, ICE -based software typically spend a couple of dozen milliseconds per Web page or e-mail, making for a true real-time filtering solution in content intensive traffic and heavy-duty environments.

 

 

Award-winning technology

 

The flagship of the ICE technology is Profil Parental Filter for Windows®, a software solution regularly taking part in all-encompassing filtering quality benchmarks.

Thanks to ICE, Profil Parental Filter has secured either gold or silver medals in such competitions for years in a row. For details, please see independent test : www.filtra.info (ranking pages available in French and German).

Up to the shifting paradigms of web 2.0 and beyond

 

The early ages of the World Wide Web brought along a few, easy to grasp content patterns, with invariants that stayed around for more or less a decade: each document used to have a home address, with all documents in a neighbourhood sharing some content significance and reflecting upon the image of the neighbourhood as a whole.

In those pre-blogging, un-aggregated Web content years, URL lists were by and large an adequate response to parental control requirements.
Not anymore.

There is a new website going online every second or so. With everyone able to start up a new blog in a matter of minutes, content providers themselves have long given up ex-ante moderation (leaving behind a fig-leaf reading: «Click here to report inappropriate content»). Trolls twist initial topics in every possible way, adding semantic hazard to blogs and newsgroups by the minute.

To make matters worse, Web 2.0 added dynamic aggregation of content, RSS and online profiling to the picture, a challenge to the very meaning of an URL.

The one and only relative invariant in this moving landscape is language itself. This is the starting point of ICE as a technology, first and foremost designed to deal with one task: «Here is a bunch of content, a split second ahead of its actual displaying before reader’s eyes – What does it talk about?». Arguably, this should still work in Web 3.0 and beyond.