SPARKLE is a Language Engineering research project. It follows the standard project lifecycle of the Telematics Applications Programme and starts from phase 1. Conceptually, and to a significant extent temporally, the lifecycle of the project can be divided into the following phases:
Two generic software systems will be developed, a robust phrasal-level parser and a lexical acquisition system for acquiring the subcategorisation, argument structure and selectional preferences of predicates. Parsers will be developed for the four target languages using approaches for which infrastructure and experience exist at the responsible sites. The generic systems will be incorporated in demonstrators involving machine-aided translation tools for information retrieval, multilingual information retrieval, and lexical acquisition for speech dialogue systems.
The work plan incorporates an evolutionary development cycle, in which an initial version of the parser prototype is delivered at year one. This parser will achieve the basic functionality of phrase-level syntactic analysis for free text. In the course of the project, the parser will be refined and improved based on experience and the incorporation of information derived automatically by the second software component, the lexical acquisition system. This software system will use a parser, text corpus, and user-specified information to semi-automatically acquire a lexicon annotated with subcategorization and selectional information. This prototype will be delivered at two years. Final versions of both systems will be delivered at the end of the project.