Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] annotator 'longest only' parameter loses annotations

Ray Fergerson ray.fergerson at stanford.edu
Tue Apr 19 14:55:17 PDT 2011


Joey,

This is a known problem. The longest only parameter will not work
correctly if you specify ontologies. It works fine if all ontologies are
included. This is a design problem and is hard for us to fix. The problem
is that the system that takes "longest only" is the external matching code
and doesn't know anything about which terms are in which ontologies. Then
ontologies are pruned after matching resulting in incorrect results, as
you have found. We need to fix the api logic to just fail this combination
of parameters. I will create a bug to do this.

Unfortunately to fix the underlying problem we would have to ignore the
built-in longest only logic in the term matcher and re-implement this
logic ourselves after matching. This project has never made it to the top
of our queue. The unpleasant workaround is for you to not pass "longest
only" and implement this logic yourself. Sorry.

Ray

> -----Original Message-----
> From: bioontology-support-bounces at lists.stanford.edu
> [mailto:bioontology-support-bounces at lists.stanford.edu] On Behalf Of
> Joey Geiger
> Sent: Tuesday, April 19, 2011 11:16 AM
> To: support at bioontology.org
> Subject: [bioontology-support] annotator 'longest only' parameter loses
> annotations
> 
> I'm trying to figure out why the 'longest only' parameter set to true
> is removing results.
> 
> With longest only off, I get 2 mapped results for muscle and get
> separate 'stomach' and 'stomach wall' annotations. (Some MAPPINGS were
> removed to shorten the results for inclusion)
> 
> ObaResultBean [
> ResultBean [
> 	resultID = OBA_RESULT_a998
> 	statistics = [(MAPPING, 16) , (CLOSURE, 0) , (MGREP, 2) ]
> 	parameters = [longestOnly = false, wholeWordOnly = true,
> filterNumber = true, withSynonyms = true, withContext = true,
> ontologiesToExpand = [], ontologiesToKeepInResult = [1000],
> isVirtualOntologyId = true, semanticTypes = [], levelMax = 0,
> mappingTypes = [], stopWords = [], withDefaultStopWords = false,
> isStopWordsCaseSenstive = false, text to annotate = skeletal muscle
> stomach wall]
> ]
> 	ontologies = [[Mouse adult gross anatomy, nbAnnotation: 18,
> score: 1658, (45065, 1.207, 1000)]]
> 	annotations = [AnnotationBean [
> 		score = 108
> 		concept = [localConceptId: 45065/MA:0000353, conceptId:
> 16754492, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0000353, preferredName: stomach,
> definitions: [], synonyms: [], semanticTypes: [[id: 20261123,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MAPPING(false), from = 17, to = 23,
> mappedConcept = [localConceptId: 45040/EHDA:5799, conceptId: 16938785,
> localOntologyId: 45040, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/EHDA#EHDA_5799, preferredName: stomach,
> definitions: [], synonyms: [], semanticTypes: [[id: 20445416,
> semanticType: T999, description: NCBO BioPortal concept]]], mappingType
> = Automatic, ]
> ], AnnotationBean [
> 		score = 108
> 		concept = [localConceptId: 45065/MA:0000353, conceptId:
> 16754492, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0000353, preferredName: stomach,
> definitions: [], synonyms: [], semanticTypes: [[id: 20261123,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MGREP(true), from = 17, to = 23, [name:
stomach,
> localConceptId: 45065/MA:0000353, isPreferred: true], ]
> ], AnnotationBean [
> 		score = 14
> 		concept = [localConceptId: 45065/MA:0000015, conceptId:
> 16754156, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0000015, preferredName: muscle organ,
> definitions: [], synonyms: [], semanticTypes: [[id: 20260787,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MAPPING(false), from = 10, to = 15,
> mappedConcept = [localConceptId: 42838/Muscle, conceptId: 15665055,
> localOntologyId: 42838, isTopLevel: 0, fullId:
> http://ncicb.nci.nih.gov/xml/owl/EVS/Thesaurus.owl#Muscle,
> preferredName: Muscle, definitions: [], synonyms: [Muscular],
> semanticTypes: [[id: 19137421, semanticType: T999, description: NCBO
> BioPortal concept]]], mappingType = Manual, ]
> ], AnnotationBean [
> 		score = 14
> 		concept = [localConceptId: 45065/MA:0000015, conceptId:
> 16754156, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0000015, preferredName: muscle organ,
> definitions: [], synonyms: [], semanticTypes: [[id: 20260787,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MAPPING(false), from = 10, to = 15,
> mappedConcept = [localConceptId: 45061/ZFA:0005145, conceptId:
> 16881603, localOntologyId: 45061, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/ZFA#ZFA_0005145, preferredName: muscle,
> definitions: [], synonyms: [muscles], semanticTypes: [[id: 20388234,
> semanticType: T999, description: NCBO BioPortal concept]]], mappingType
> = Automatic, ]
> ], AnnotationBean [
> 		score = 10
> 		concept = [localConceptId: 45065/MA:0002692, conceptId:
> 16756809, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0002692, preferredName: stomach wall,
> definitions: [], synonyms: [], semanticTypes: [[id: 20263440,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MGREP(true), from = 17, to = 28, [name: stomach
> wall, localConceptId: 45065/MA:0002692, isPreferred: true], ]
> ]]
> ]
> 
> When I set 'longest only' to true, I get no results for muscle and only
> a single (proper) result for 'stomach wall'.
> 
> ObaResultBean [
> ResultBean [
> 	resultID = OBA_RESULT_a9e7
> 	statistics = [(MAPPING, 0) , (CLOSURE, 0) , (MGREP, 1) ]
> 	parameters = [longestOnly = true, wholeWordOnly = true,
> filterNumber = true, withSynonyms = true, withContext = true,
> ontologiesToExpand = [], ontologiesToKeepInResult = [1000],
> isVirtualOntologyId = true, semanticTypes = [], levelMax = 0,
> mappingTypes = [], stopWords = [], withDefaultStopWords = false,
> isStopWordsCaseSenstive = false, text to annotate = skeletal muscle
> stomach wall]
> ]
> 	ontologies = [[Mouse adult gross anatomy, nbAnnotation: 1, score:
> 10, (45065, 1.207, 1000)]]
> 	annotations = [AnnotationBean [
> 		score = 10
> 		concept = [localConceptId: 45065/MA:0002692, conceptId:
> 16756809, localOntologyId: 45065, isTopLevel: 0, fullId:
> http://purl.org/obo/owl/MA#MA_0002692, preferredName: stomach wall,
> definitions: [], synonyms: [], semanticTypes: [[id: 20263440,
> semanticType: T999, description: NCBO BioPortal concept]]]
> 		context = [MGREP(true), from = 17, to = 28, [name: stomach
> wall, localConceptId: 45065/MA:0002692, isPreferred: true], ]
> ]]
> ]
> 
> Isn't the idea that it should match something no matter what, and if
> there's something 'longer' it should use that result? I know this
> probably has something to do with the mapping functionality, but from a
> use standpoint it's rather annoying. I have to keep it off all the time
> since it will miss annotations that clearly exist, but at the same time
> gives me multiple sub-annotations for a clearly more specific term.
> 
> Thanks.
> _______________________________________________
> bioontology-support mailing list
> bioontology-support at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/bioontology-support



More information about the bioontology-support mailing list