Interview donnée à TV4IT (8 min) sur les enjeux du Web Sémantique:

01 Informatique spécial DSI du 3 juillet 2008

The Antelope linguistic framework includes :

• Taggers and chunkers.
• Dependency parsers for English (Link Grammar 4.1, Stanford Parser 1.6).
• Both parsers can return multiple analyses (multiple best parse trees).
• Parsing support for French (TagParser).
• A full semantic lexicon, with rich relations, based on WordNet 3.0 data.
• A collocation (multi-word expression) collapser.
• Two similarity measures between lemmas (using gloss overlapping or the noun/verb hierarchy).
• A versatile user-defined lexicon facility, and two lexicons are provided:
  • A “French WordNet” containing 44,200 new French lemmas (French.Wolf.Lexicon.xml.gz, based on WOLF).
  • 300,000 new synsets with a mapping to the English Wikipedia (Wikipedia.Lexicon.xml.gz, based on DBPedia data).
• An anaphora resolver.
• A space and time features detector.
• Word Sense Disambiguation.
• A context extraction module.
• Deep syntax extraction.
• Semantic Role Labelling (a syntax / semantic layer, based on VerbNet 1.5).
• A sentence splitter for plain text or HTML text.
• A HTML document can be processed using its structure (interface IToken), and rewritten without any loss of information.
• A PROLOG interpreter for .NET.
• Paraphrases extraction (very experimental).
• A powerful multi-threaded semantic WebSearch utility.
• Documentation with both C# and Visual Basic.NET samples (managed C++ available on demand).
• Tested under Windows XP, Vista 32 bits and Vista 64 bits (in x86 mode).
• Partially tested under Linux (under developement), thanks to MONO.
• Multi-threading support, can be used on a Web server.
• A ribbon-based Graphical User Interface.