Extraction of Dutch definitory contexts for eLearning purposes

Abstract

The aim of the Language Technology for eLearning project is to facilitate the retrieval, management and distribution of learning material within a Learning Management System by exploiting Natural Language Processing techniques as well as semantic knowledge. One of the functionalities provided by the project is the possibility to create glossaries semiautomatically. Glossaries are derived from the learning objects in order to capture the exact definition which the author of these documents uses. A rule-based approach is employed to identify the relevant lexical and linguistic patterns which underlie the definition. In this paper, we discuss the grammar developed to identify the definitory contexts in the Dutch learning objects and we present the results of the quantitative evaluation.