com.exalead.mot.components

Class OntologyMatcher

java.lang.Object

com.exalead.gni.GNIObject

com.exalead.mot.core.Processor

com.exalead.mot.components.OntologyMatcher

public class OntologyMatcher
extends Processor

Ontology matcher, matches expressions from an ontology

Field Summary

Fields

Modifier and Type	Field and Description
boolean	addReferencesForIgnoredTags add references for ignored tags
java.lang.String	annotationPrefix prefix to prepend to annotations' tag
boolean	enableApproxMatching When set to true, we enable approximative matching in ontology Approximative matching use Damerau-Levenshtein edit distance
boolean	ignoreSpaces if your ontology was compiled with matchOnSeparators = false - this allows 'lemonde' to retrieve 'le monde' entry or 'le monde' to retrieves 'lemonde' if your ontology was compiled with matchOnSeparators = true - this allows 'le monde' to retrieve 'le monde'
java.lang.String	matchAgainst annotation the ontology terms are matched against
int	minWordSizeForDist1 Minimum number of chars in token to enable damerau-levenshtein distance of 1
int	minWordSizeForDist2 Minimum number of chars in token to enable damerau-levenshtein distance of 2
boolean	packageLevelMatchDedup When deduping results, try to keep only one match per package, i.e.
java.lang.String	resource The processor resource
boolean	restrictLanguage when set to true, keep only expression added with language == Language.XX or with document language for exemple if Ontology contains an expression added with language=En, if will be extracted only for english document with restrictLanguage is set to true
boolean	splitHandling Match a term even if there are no blanks between alphanums tokens in the text (Spellchecker may generates two tokens from one by splitting it up but no blank is inserted) When enabled, the text [air][france] will match the entry [air][ ][france]
boolean	suffixApproxMatching When set to true, all approximative match have a tag suffixed by '.approx' If you disable this suffix, the trustLevel of exact entry will all by set to 100 (this override the trustLevel in original xml file)
java.util.List<java.lang.String>	tagsToIgnore set the list of tag to be ignored this feature allows to define a list of words/expressions to ignore in the recognotion of this ontology For example if you add expression "of" and "the" with the tag "toIgnore" in the ontology A, then you add the expression "website embassy" in ontology B with tagsToIgnore=["toIgnore"], you will be able to match "website of the embassy", "website of embassy" and "website ambassy" WARNING: for the moment this option is not compatible when ontology was compiled without matchOnSeparators=false
boolean	trustLevelBasedDedup Keeps only the annotations with the highest trust level when several overlap.

Fields inherited from class com.exalead.mot.core.Processor

fields, name

Constructor Summary

Constructors

Constructor and Description
OntologyMatcher(java.lang.String name, java.lang.String resource, java.lang.String fields)
OntologyMatcher(java.lang.String name, java.lang.String resource, java.lang.String matchAgainstAnnotation, java.lang.String fields, boolean restrictToLang, boolean addReferencesForIgnoredTags, boolean enableApproxMatching, boolean suffixApproxMatching, int minWordSizeForDist1, int minWordSizeForDist2, boolean ignoreSpaces, java.lang.String annotationPrefix) Initialize a features extractor

Method Summary

Modifier and Type	Method and Description
protected void	addTagToIgnore(java.lang.String tag) Add a tag to ignore
protected void	finalize()
static java.lang.String	getApproxTagSuffix()
void	init(java.lang.String name, java.lang.String[] fields) Initialize the processor
void	init(java.lang.String name, java.lang.String resource, java.lang.String matchAgainstAnnotation, java.lang.String[] fields, boolean restrictToLang, boolean addReferencesForIgnoredTags, boolean enableApproxMatching, boolean suffixApproxMatching, int minWordSizeForDist1, int minWordSizeForDist2, boolean ignoreSpaces, java.lang.String annotationPrefix, boolean trustLevelDedup, boolean splitHandling, boolean packageLevelMatchDedup)
protected void	initNative(java.lang.String name, java.lang.String resource, java.lang.String matchAgainstAnnotation, java.lang.String[] fields, boolean restrictToLang, boolean addReferencesForIgnoredTags, boolean enableApproxMatching, boolean suffixApproxMatching, int minWordSizeForDist1, int minWordSizeForDist2, boolean ignoreSpaces, java.lang.String annotationPrefix, boolean trustLevelDedup, boolean splitHandling, boolean packageLevelMatchDedup)

Methods inherited from class com.exalead.mot.core.Processor

checkResource, destroy, getName, init

Methods inherited from class com.exalead.gni.GNIObject

printInternalCppPointer

Methods inherited from class java.lang.Object

clone, equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

resource

public java.lang.String resource

The processor resource

enableApproxMatching

public boolean enableApproxMatching

When set to true, we enable approximative matching in ontology Approximative matching use Damerau-Levenshtein edit distance

suffixApproxMatching

public boolean suffixApproxMatching

When set to true, all approximative match have a tag suffixed by '.approx' If you disable this suffix, the trustLevel of exact entry will all by set to 100 (this override the trustLevel in original xml file)

minWordSizeForDist1

public int minWordSizeForDist1

Minimum number of chars in token to enable damerau-levenshtein distance of 1

minWordSizeForDist2

public int minWordSizeForDist2

Minimum number of chars in token to enable damerau-levenshtein distance of 2

ignoreSpaces

public boolean ignoreSpaces

if your ontology was compiled with matchOnSeparators = false - this allows 'lemonde' to retrieve 'le monde' entry or 'le monde' to retrieves 'lemonde' if your ontology was compiled with matchOnSeparators = true - this allows 'le monde' to retrieve 'le monde'

splitHandling

public boolean splitHandling

Match a term even if there are no blanks between alphanums tokens in the text (Spellchecker may generates two tokens from one by splitting it up but no blank is inserted) When enabled, the text [air][france] will match the entry [air][ ][france]

packageLevelMatchDedup

public boolean packageLevelMatchDedup

When deduping results, try to keep only one match per package, i.e. consider only annotation name and not annotation value when comparing results for equivalence

restrictLanguage

public boolean restrictLanguage

when set to true, keep only expression added with language == Language.XX or with document language for exemple if Ontology contains an expression added with language=En, if will be extracted only for english document with restrictLanguage is set to true

addReferencesForIgnoredTags

public boolean addReferencesForIgnoredTags

add references for ignored tags

matchAgainst

public java.lang.String matchAgainst

annotation the ontology terms are matched against

annotationPrefix

public java.lang.String annotationPrefix

prefix to prepend to annotations' tag

tagsToIgnore

public java.util.List<java.lang.String> tagsToIgnore

set the list of tag to be ignored this feature allows to define a list of words/expressions to ignore in the recognotion of this ontology For example if you add expression "of" and "the" with the tag "toIgnore" in the ontology A, then you add the expression "website embassy" in ontology B with tagsToIgnore=["toIgnore"], you will be able to match "website of the embassy", "website of embassy" and "website ambassy" WARNING: for the moment this option is not compatible when ontology was compiled without matchOnSeparators=false

trustLevelBasedDedup

public boolean trustLevelBasedDedup

Keeps only the annotations with the highest trust level when several overlap.

Constructor Detail

OntologyMatcher

public OntologyMatcher(java.lang.String name,
                       java.lang.String resource,
                       java.lang.String matchAgainstAnnotation,
                       java.lang.String fields,
                       boolean restrictToLang,
                       boolean addReferencesForIgnoredTags,
                       boolean enableApproxMatching,
                       boolean suffixApproxMatching,
                       int minWordSizeForDist1,
                       int minWordSizeForDist2,
                       boolean ignoreSpaces,
                       java.lang.String annotationPrefix)

Initialize a features extractor

Parameters:

name - Its name

resource - The associated resource name

matchAgainstAnnotation - annotation the ontology terms are matched against

fields - The list of fields on which it's active

restrictToLang - Keep only expression added with language == Language.XX or with document language

addReferencesForIgnoredTags - Add references for ignored tags

enableApproxMatching - Approximative matching using Damerau-Levenshtein edit distance

suffixApproxMatching - Approximative match have a tag suffixed by '.approx'

minWordSizeForDist1 - Minimum number of chars in token to enable damerau-levenshtein distance of 1

minWordSizeForDist2 - Minimum number of chars in token to enable damerau-levenshtein distance of 2

ignoreSpaces - Ignore spaces when matching

annotationPrefix - Prefix to add to annotations' tag

OntologyMatcher

public OntologyMatcher(java.lang.String name,
                       java.lang.String resource,
                       java.lang.String fields)

Method Detail

init

public void init(java.lang.String name,
                 java.lang.String resource,
                 java.lang.String matchAgainstAnnotation,
                 java.lang.String[] fields,
                 boolean restrictToLang,
                 boolean addReferencesForIgnoredTags,
                 boolean enableApproxMatching,
                 boolean suffixApproxMatching,
                 int minWordSizeForDist1,
                 int minWordSizeForDist2,
                 boolean ignoreSpaces,
                 java.lang.String annotationPrefix,
                 boolean trustLevelDedup,
                 boolean splitHandling,
                 boolean packageLevelMatchDedup)

init

public void init(java.lang.String name,
                 java.lang.String[] fields)

Initialize the processor

Specified by:

init in class Processor

Parameters:

name - Its name

fields - The list of fields on which it's active

getApproxTagSuffix

public static java.lang.String getApproxTagSuffix()

initNative

protected void initNative(java.lang.String name,
                          java.lang.String resource,
                          java.lang.String matchAgainstAnnotation,
                          java.lang.String[] fields,
                          boolean restrictToLang,
                          boolean addReferencesForIgnoredTags,
                          boolean enableApproxMatching,
                          boolean suffixApproxMatching,
                          int minWordSizeForDist1,
                          int minWordSizeForDist2,
                          boolean ignoreSpaces,
                          java.lang.String annotationPrefix,
                          boolean trustLevelDedup,
                          boolean splitHandling,
                          boolean packageLevelMatchDedup)

addTagToIgnore

protected void addTagToIgnore(java.lang.String tag)

Add a tag to ignore

finalize

protected void finalize()

Overrides:

finalize in class java.lang.Object