org.apache.ctakes.ytex.kernel.metric

Interface ConceptSimilarityService

All Known Implementing Classes:

ConceptSimilarityServiceImpl

public interface ConceptSimilarityService

Nested Class Summary

Nested Classes

Modifier and Type	Interface and Description
static class	ConceptSimilarityService.SimilarityMetricEnum

Method Summary

Modifier and Type	Method and Description
Object[]	getBestLCS(Set<String> lcses, boolean intrinsicIC, Map<String,Double> conceptFilter) get the best lcs
ConceptGraph	getConceptGraph()
String	getConceptGraphName()
Map<String,BitSet>	getCuiTuiMap() cui - tui map.
int	getDepth(String concept)
double	getIC(String concept, boolean intrinsicICMap)
int	getLCS(String concept1, String concept2, Set<String> lcses, List<LCSPath> lcsPaths) get the lcs(s) for the specified concepts
List<String>	getTuiList() list of tuis that corresponds to bitset indices
int	lcs(String concept1, String concept2, List<LCSPath> lcsPath)
double	loadConceptFilter(String label, int rankCutoff, Map<String,Double> conceptFilter) For the given label and cutoff, get the corresponding concepts whose propagated ig meets the threshold.
List<ConceptPairSimilarity>	similarity(List<ConceptPair> conceptPairs, List<ConceptSimilarityService.SimilarityMetricEnum> metrics, Map<String,Double> conceptFilter, boolean lcs) compute similarity for a list of concept pairs
ConceptPairSimilarity	similarity(List<ConceptSimilarityService.SimilarityMetricEnum> metrics, String concept1, String concept2, Map<String,Double> conceptFilter, boolean lcs) compute similarity for a pair of concepts

Method Detail

getConceptGraphName

String getConceptGraphName()

lcs

int lcs(String concept1,
        String concept2,
        List<LCSPath> lcsPath)

getConceptGraph

ConceptGraph getConceptGraph()

getCuiTuiMap

Map<String,BitSet> getCuiTuiMap()

cui - tui map. tuis are bitsets, indices correspond to tuis in getTuiList()

Returns:

getTuiList

List<String> getTuiList()

list of tuis that corresponds to bitset indices

Returns:

loadConceptFilter

double loadConceptFilter(String label,
                         int rankCutoff,
                         Map<String,Double> conceptFilter)

For the given label and cutoff, get the corresponding concepts whose propagated ig meets the threshold. Used by lin kernel to find concepts that actually have a non-trivial similarity

Parameters:

label - label

rankCutoff - cutoff

conceptFilter - set to fill with concepts

Returns:

double minimum evaluation

getLCS

int getLCS(String concept1,
           String concept2,
           Set<String> lcses,
           List<LCSPath> lcsPaths)

get the lcs(s) for the specified concepts

Parameters:

concept1 - required

concept2 - required

lcses - required - will be filled with the lcs(s).

lcsPathMap - optional - will be filled with lcs and paths through the lcses.

Returns:

distance of path through lcs

getBestLCS

Object[] getBestLCS(Set<String> lcses,
                    boolean intrinsicIC,
                    Map<String,Double> conceptFilter)

get the best lcs

Parameters:

lcses - set of lcses

intrinsicIC - should the intrinsic ic be used? false - use corpus-based ic. For multiple lcses not using concept filter, use the lcs with the lowest infocontent

conceptFilter - limit to lcses in the concept filter. The lcs with the highest value will be used.

Returns:

array with 2 entries. Entry 1 - lcs (String). Entry 2 - infocontent (double). Null if no lcses are in the concept filter.

getIC

double getIC(String concept,
             boolean intrinsicICMap)

similarity

ConceptPairSimilarity similarity(List<ConceptSimilarityService.SimilarityMetricEnum> metrics,
                                 String concept1,
                                 String concept2,
                                 Map<String,Double> conceptFilter,
                                 boolean lcs)

compute similarity for a pair of concepts

Parameters:

metrics - required, similarity metrics to compute

concept1 - required

concept2 - required

conceptFilter - optional - only lcs's in this set will be used.

simInfo - optional - pass this to get information on lcs. Instantiate the lcsPathMap to get paths through lcs

Returns:

similarities

similarity

List<ConceptPairSimilarity> similarity(List<ConceptPair> conceptPairs,
                                       List<ConceptSimilarityService.SimilarityMetricEnum> metrics,
                                       Map<String,Double> conceptFilter,
                                       boolean lcs)

compute similarity for a list of concept pairs

Parameters:

conceptPairs - required, concept pairs for which similarity should be computed

metrics - required, similarity metrics to compute

conceptFilter - optional - only lcs's in this set will be used.

simInfos - optional - if provided, this list will be filled with the similarity info for each concept pair.

Returns:

similarities

getDepth

int getDepth(String concept)