- From: Parsa Mirhaji <Parsa.Mirhaji@uth.tmc.edu>
- Date: Fri, 9 Jul 2010 13:55:54 -0500
- To: W3C HCLSIG hcls <public-semweb-lifesci@w3.org>
- CC: Marco Roos <M.Roos1@uva.nl>, Steven Bedrick <bedricks@ohsu.edu>
Take a look at the following presentation from NCBO regarding SKOS representation of the UMLS for semantic applications that can potentially address these use cases: http://www.bioontology.org/umls-skos . the current incarnation of algorithm only maps English, but can be used to include Spanish, Dutch, French etc, basically most Indo-European and some east Asian languages as much as they are provided by and through UMLS. it can also be extended directly and out of the scope of UMLS, to account for those language differences, still maintaining knowledge and relationships according to UMLS... Parsa Mirhaji, MD. PhD. Assistant Professor The School of Biomedical informatics The University of Texas Health Science Center at Houston Tel: (713) 500-3157 Fax: (713) 500-0370 Assistance (Yuki Bryson): (713) 500-3937 On Jul 9, 2010, at 10:03 AM, Steven Bedrick wrote: I'd suggest looking at the US National Library of Medicine's Unified Medical Language System (UMLS): http://www.nlm.nih.gov/research/umls/ It is, among other things, a metathesaurus of biomedical language built from a couple dozen source vocabularies. I'm pretty sure that there's at least one dutch vocabulary in there, but I mostly use the English and Spanish vocabs. It can definitely be used to map terms from one language to another, especially if they're from the same vocabulary (e.g. SNOMED-CT/English and SNOMED-CT/Spanish), but for a variety of reasons it's not particularly useful for simple dictionary-style translation- each vocabulary has its own "view of the world", as it were, and while the editors do a good job of mapping terms from one to another, there's always a loss of resolution even within a single language, let alone across multiple languages (although English and Dutch are close enough cousins that it might not be so bad). The UMLS also has a semantic network mapping terms to one another (i.e., "Peanut Allergy" is a "Food Allergy" which is a "Disease or Condition", etc.), so once you've mapped your text to UMLS terms you've got a semantic model to work with. It's not directly exposed as RDF triples, but I know that various people have tried various things to expose UMLS data in a semantic-web-friendly way. For example, the US National Center for Biomedical Ontology has the BioPortal web service: http://bioportal.bioontology.org/ Let me know if you want more information... -SB On Jul 9, 2010, at 2:11 AM, Marco Roos wrote: > Hello all, > > Do you know of semantic models that contain translations of biomedical terms in different languages, preferably including Dutch? > > The context is a pilot project where we want to show the use of semantically disclosing medical surveys. The first resource we start working on is in Dutch, but I'm interested if something like this exists for other languages too. > > Many thanks! > Marco. > > -- > Marco Roos (PhD) > BioSemantics group, Human Genetics department, Leiden University Medical Centre > Albinusdreef 2 (Building 2, room S3-46) > 2333 ZA Leiden > Tel. +31 (0) 71 526 8642 > Adaptive Information Disclosure group, Informatics Institute, University of Amsterdam > Kruislaan 403 ( room N2.30) 1098 SJ Amsterdam > Tel. +31 (0) 20 525 7522 > Communities: > Professional: LinkedIn, myExperiment Personal: Hyves
Received on Saturday, 10 July 2010 12:57:24 UTC