Evaluating Bilingual Medical Terminologies with Word Alignment Methods
Mikael Nyström, Magnus Merkel, Håkan Petersson, Hans Åhlfeldt · Medinfo 2007: Proceedings of the 12th World Congress on Health (Medical) Informatics; Building Sustainable Health Systems · 2007
Automatic word alignment in parallel texts can be used to generate lexicons, but the lexicons' quality is depending on the quality and type of the input resources. In this study the English and Swedish versions of the medical terminology systems ICD-10, ICF, NCSP, KSH97-P and MeSH are automatically word aligned and the influence the different systems have on the quality is evaluated. The most generally useful resources were generated from ICD-10 and the size of the partitions used to generate the resources only partly affects the quality.