Knowledge Sources Unified Medical Language System

1 knowledge sources

1.1 metathesaurus
1.2 semantic network
1.3 specialist lexicon

knowledge sources
metathesaurus

the metathesaurus forms base of umls , comprises on 1 million biomedical concepts , 5 million concept names, of stem on 100 incorporated controlled vocabularies , classification systems. examples of incorporated controlled vocabularies icd-10, mesh, snomed ct, dsm-iv, loinc, adverse drug reaction terminology, uk clinical terms, rxnorm, gene ontology, , omim (see full list).

the metathesaurus organized concept, , each concept has specific attributes defining meaning , linked corresponding concept names in various source vocabularies. numerous relationships between concepts represented, instance hierarchical ones such isa subclasses , part of subunits, , associative ones such caused or in literature occurs close (the latter being derived medline).

the scope of metathesaurus determined scope of source vocabularies. if different vocabularies use different names same concept, or if use same name different concepts, faithfully represented in metathesaurus. hierarchical information source vocabularies retained in metathesaurus. metathesaurus concepts can link resources outside of database, instance gene sequence databases.

semantic network

each concept in metathesaurus assigned 1 or more semantic types (categories), linked 1 through semantic relationships. semantic network catalog of these semantic types , relationships. rather broad classification; there 127 semantic types , 54 relationships in total.

the major semantic types organisms, anatomical structures, biologic function, chemicals, events, physical objects, , concepts or ideas. links among semantic types define structure of network , show important relationships between groupings , concepts. primary link between semantic types isa link, establishing hierarchy of types. network has 5 major categories of non-hierarchical (or associative) relationships, constitute remaining 53 relationship types. these physically related , spatially related , temporally related , functionally related , conceptually related .

the information semantic type includes identifier, definition, examples, hierarchical information encompassing semantic type(s), , associative relationships. associative relationships within semantic network weak. capture @ some-some relationships, i.e. capture fact instance of first type may connected salient relationship instance of second type. phrased differently, capture fact corresponding relational assertion meaningful (though need not true in cases).

an example of associative relationship may-cause , applied terms (smoking, lung cancer) yield: smoking may-cause lung cancer.

specialist lexicon

the specialist lexicon contains information common english vocabulary, biomedical terms, terms found in medline , terms found in umls metathesaurus. each entry contains syntactic (how words put create meaning), morphological (form , structure) , orthographic (spelling) information. set of java programs use lexicon work through variations in biomedical texts relating words parts of speech, can helpful in web searches or searches through electronic medical record.

entries may one-word or multiple-word terms. records contain 4 parts: base form (i.e. run running ); parts of speech (of specialist recognizes eleven); unique identifier; , available spelling variants. example, query anesthetic return following:

{ base=anaesthetic
spelling_variant=anesthetic
entry=e0008769
cat=noun
variants=reg
}
{ base=anaesthetic
spelling_variant=anesthetic
entry=e0008770
cat=adj
variants=inv
position=attrib(3)
}

the specialist lexicon available in 2 formats. unit record format can seen above, , comprises slots , fillers. slot element (i.e. base= or spelling variant= ) , fillers values attributable slot entry. relational table format not yet normalized , contain great deal of redundant data in files.