Ontology Engineering and Knowledge Extraction for Cross-Lingual Retrieval

International Conference RANLP 2009 - Borovets, Bulgaria, pages 455–459 Ontology engineering and knowledge extraction for crosslingual retrieval Jantine Trapman Utrecht University jeandix23@yahoo.com Paola Monachesi Utrecht University P.Monachesi@uu.nl Abstract In this paper, we show that by integrating existing NLP techniques and Semantic Web tools in a novel way, we can provide a valuable contribution to the solution of the knowledge acquisition bottleneck problem. NLP tech- niques to create a domain ontology on the ba- sis of an open domain corpus have been com- bined with Semantic Web tools. More specif- ically, Watson and Prompt have been em- ployed to enhance the kick-off ontology while Cornetto, a lexical database for Dutch, has been adopted to establish a link between the concepts and their Dutch lexicalization. The lexicalized ontology constitutes the basis for the cross-language retrieval of learning objects within the LT4eL eLearning project. Keywords Ontology learning, eLearning, NLP, Semantic Web, cross-lingual retrieval 1 Introduction The aim of the Language Technology for eLearning (LT4eL)1 project is to employ Language Technology and Semantic Web resources and tools to enhance eLearning in order to develop innovative applications for education and training [8]. One important ob- jective is to enhance the management, distribution, search and reuse of multilingual learning material [7] . Ontologies play a relevant role in the realization of this objective. More specifically, in the LT4eL project, the ontology mediates between the user and the learn- ing material. The relevant concepts which are attested in the learning objects constitute the backbone of the ontology. Thus, a link is created between the learning material and its conceptualization which is represented by means of the ontology allowing for the creation of individualized learning paths. However, the most im- portant contribution of the ontology is its role as in- terlingua. It facilitates access to documents in various languages since it allows for cross-lingual retrieval by mediating at the conceptual level among language spe- cific textual realizations of the concepts. The LT4eL project has provided a prototype which has shown the feasibility of the approach that has been validated within an eLearning context. However, in order to develop a real life application, the knowledge 1 http://www.lt4el.eu needs to be extracted and modeled semi-automatically. Several approaches have been proposed to this end in the Natural Language Processing as well as in the Se- mantic Web literature, providing a valuable contribu- tion to the solution of this problem. Our goal is to rely on previous results and inte- grate, in a novel way, existing Natural Language Pro- cessing techniques such as that proposed in [2] for ontology learning from text with recent approaches emerging from the Semantic Web community. An ex- ample of which is [12], that uses dynamically selected ontologies as background knowledge to enrich existing ontologies with new concepts. Our aim is to extend (semi-automatically) the ontology developed within the LT4eL project to new domains enabling thus the cross-lingual retrieval of new learning objects. To this end, a mapping has been carried out between the on- tology and …

Content truncated to 500 words