LogoVista E to J

REENGINEERING LINGUISTIC RESOURCES FOR MACHINE TRANSLATION IN MEDICAL APPLICATIONS G. Deville a, E. Herbigniaux a, P. Mousel b, G. Thienpont b a Ecole de Langues Vivantes - Facultés Universitaires de Namur 61 rue de Bruxelles - B-5000 Namur - Belgium b Centre de Recherche Public - Centre Universitaire 162a avenue de la Faiencerie - L-1511 Luxembourg - Luxembourg Abstract We discuss some key methodological and operational aspects related to the design and development of a machine translation (MT) prototype which can be integrated in healthcare information systems. We first describe the approach adopted for collecting, formating, sampling and analyzing multilingual corpora of diagnostic expressions. The resulting generic language representation model is then presented. Finally, the prototype's architecture, its application programming interface as well as its internal structures are outlined. For economical reasons, the use of existing linguistic resources and MT technology, reengineered for an original application, was an essential requirement in this project. 1. INTRODUCTION The present paper describes ongoing efforts in the framework of the LRE project ANTHEM1. The objective of ANTHEM is to develop a portable prototype of a multilingual natural language interface that allows users of healthcare information systems to enter diagnostic expressions using their own natural language and to have this input translated in whatever formal, structured or natural language. The project results from the collaboration of researchers from various disciplines: medicine, computer science and linguistics 2. The realization of the ANTHEM prototype necessitated an extensive linguistic study of large representative corpora of medical diagnostic expressions in order to model the 1 The ANTHEM project: "Advanced Natural Language Interface for Multilingual Text Generation in Healthcare" (LRE 62-007) is co-financed by the European Union within the "Linguistic Research and Engineering" program. 2 The ANTHEM consortium is coordinated by W. Ceusters of RAMIT vzw (Ghent University Hospital) and further consists of the Institute of Modern Languages of the University of Namur (G. Deville), the IAI of the University of Saarbrücken (O. Streiter), the CRP-CU of Luxembourg (P. Mousel), the University of Liege (C. Gérardy), Datasoft Management nv — Oostende (J. Devlies) and the Military Hospital in Brussels (D. Penson). 24 application sublanguage and to develop the appropriate lingwares (multi-lingual lexicons and grammars). In the next section, we first discuss the medical corpus collection, formating, sampling and analysis in the prospect of the ANTHEM sublanguage modeling. Then, we briefly present the semantic representation model developed in the context of the project. In a third section, we will focus on some operational aspects of the ANTHEM project. After a short description of the prototype's architecture, the specification of its application programming interface (API) will be presented. Finally, we will briefly sketch the prototype's internal structures. To conclude, options for further developments will be outlined in terms of an augmentation of the lingwares and of the distribution of the prototype components. 2. A CORPUS BASED SUBLANGUAGE MODELING In order to comply with the requirements of multidisciplinarity in the ANTHEM project, a methodology has been set up to develop and validate a …

Content truncated to 500 words