Browse Prior Art Database

Language Independent Document Representation System

IP.com Disclosure Number: IPCOM000013382D
Original Publication Date: 2001-Apr-13
Included in the Prior Art Database: 2003-Jun-18

Publishing Venue

IBM

Abstract

A known Language Translation System for the Web is based on his English Slot Grammar (ESG). At the University of Brightona system was built to capture documents in language-independent representation for manuals. The system is described in "The Multilingual Generation Game: authoring fluent texts in unfamiliar languages" by Donia R Scott, Proceedings of the IJCAI 99 Vol. 2, page 1407-1411. If one wants to generate a document (e.g. a Web page or a manual) in several different languages, then this is usually done by creating the document in one language, and then using machine translation tools to translate it into the desired languages. This method has several problems. One problem is that the machine translation tools don't always produce good quality translations and may make the result unusable. Another problem is that meaning depends on the socio-cultural context, or, in other words, something that can be expressed in one language may not be expressable in another language. These problems are addressed and solved by a Language Independent Document Representation System where all kinds of documents can first be generated in some language-independent way and then be transformed into several different languages, where the resulting document is of high quality and captures all the intended meanings. The proposed system is based on a document representation in Turtle Language, i.e. the document is generated as a fractal network of n-th kind. A graphical editor can be used to generate the document in Turtle Language. The resulting document is then classified on all scales where this can be done either automatically by the Classification Jani or manually.