VINCI

From NLG Systems Wiki

Jump to: navigation, search

Summary

  • System Name: VINCI
  • Short Description: VINCI is a natural language generation environment used since 1986 for language learning and testing, subject testing, and modelling of diverse linguistic phenomena ranging from word-formation, to narrative structure, to generation of verbal humour.
  • URL: http://www.cs.queensu.ca/CompLing/
  • System Builders: Lessard, Levison[1][2][3][4]
  • Development Period: 1986–ongoing
  • Languages: English, French
  • Domain: CALL

Description

VINCI is a natural language generation environment under development for the past 20-odd years. It has been used since 1986 for language learning and testing, subject testing, and modelling of diverse linguistic phenomena ranging from word-formation, to narrative structure, to generation of verbal humour. Most research has been done in French or English, but the system was designed to be multilingual and some work has been done in other languages as well.<P>The system includes a variety of features, including:

  • a semantic metalanguage which is not language-specific
  • lexical preselection mechanisms which allow initial choice of lexical items in some natural language and inheritance of traits from these items in subsequent syntactic development
  • a system of attributes which provides the 'glue' between semantics, syntax, morphology and lexicon; attributes may be partially ordered, compounded, and deconstructed
  • a syntactic mechanism based on a context-free grammar with attribute attachment, which allows inheritance, guarded syntax rules allowing for conditional development of child nodes, and transformations of syntax trees
  • lexical representation based on syntactic and semantic attributes, morphological traits, orthography, phonology, and frequency; also included are lexical pointer mechanisms to permit representation of relations such as hyponymy, synonymy, etc.
  • morphology rules based on attribute values, orthography, phonology, and preceding and following context
  • lexical transformation devices which permit the synthesis by rule of sets of derived or compound words, either statically or dynamically
  • error analysis mechanisms which take some user input and compare it to some set of rule-governed candidate parses at the phonological, phonographic, orthographic, morphological, lexical, syntactic and semantic levels.
<P>VINCI is embedded in an editing environment (IVI) which provides for easy modification of language specifications, including template-based editing of lexical records. IVI also permits the use of driver programs written in C or some other language. The program is written in C and runs under Solaris, Linux, and Windows (using Cygwin). It is freely available from the project website, which also contains documentation on the system, sample language descriptions, and references to the papers and articles to which the project has given rise.

References

  1. Levison, M., Lessard, G., Danielson, A. M., & Merven, D. 2001. From Symptoms to Diagnosis. Bib
  2. Lessard, G., & Levison, M. 1995. Le logiciel VINCI: lexigrammaire et génération automatique. Bib
  3. Levison, M., & Lessard, G. (1996). Using a Language Generation System for Second Language Learning. Computer-Assisted Language Learning, 9(2-3), 181. Bib
  4. Levison, M., & Lessard, G. (1992). A System for Natural Language Sentence Generation. Computers and the Humanities, 26, 43. Bib

Facts about VINCIRDF feed
DescriptionVINCI is a natural language generation environment used since 1986 for language learning and testing, subject testing, and modelling of diverse linguistic phenomena ranging from word-formation, to narrative structure, to generation of verbal humour.  +
DomainCALL  +
LanguageEnglish  +, and French  +
NameVINCI  +
Started1986  +
URLhttp://www.cs.queensu.ca/CompLing/  +
WorkerLessard  +, and Levison  +
Personal tools