com.entitystream.identiza.entity.resolve.types

Class MatchBase

java.lang.Object

com.entitystream.identiza.entity.resolve.types.MatchBase

All Implemented Interfaces:

MatchProcInterface, Serializable

Direct Known Subclasses:

MatchAddress, MatchAge, MatchAgeRange, MatchBeatCode, MatchChinesePersonName, MatchCode, MatchColor, MatchCompanyName, MatchContent, MatchEditDistanceDate, MatchEditDistanceString, MatchEmail, MatchEndAddress, MatchFormula, MatchHeight, MatchList, MatchPersonName, MatchPersonNameSyllables, MatchPersonNameXDOB, MatchPhonetic, MatchPostCode, MatchSortedEditDistanceString, MatchString, MatchSuspect, MatchText, MatchTicker

public class MatchBase
extends Object
implements MatchProcInterface, Serializable

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected int	gradient
protected boolean	isMandatory
protected Logger	logger
protected int	maxScore
protected int	maxWidth
protected int	minWidth
protected boolean	partialKeys
protected String	ruleFunction
protected RuleSet	ruleSet
protected String	type

Fields inherited from interface com.entitystream.identiza.entity.resolve.types.MatchProcInterface

MATCH_CLOSE, MATCH_EXACT, MATCH_FAR, MATCH_MED, MATCH_VCLOSE, MATCH_VFAR, NARROW, STANDARD, WIDE

Constructor Summary

Constructors

Constructor and Description
MatchBase(String ruleFunction)
MatchBase(String ruleFunction, int _minwidth, int _maxwidth)
MatchBase(String ruleFunction, int minwidth, int maxwidth, String type)

Method Summary

Modifier and Type	Method and Description
protected ArrayList<String>	allCombs(String key, boolean partial, boolean alwaysUseFirst, int minWidth, int maxWidth)
double	calculateComparisonScore(Standardized stdBase, Standardized stdComp, boolean isSearch, boolean asContent)
protected ArrayList<String>	compressInitials(String[] in)
protected HashSet<String>	getAlternateWords(String word, WordList anon, WordList controlVocabulary)
String	getKey(String word)
WordList	getKeyAnon()
WordList	getKeyCv()
WordList	getKeyLookup()
Collection<String>	getKeys(HashMap<String,String> value, boolean batch)
Collection<String>	getKeys(Object value, boolean batch)
Collection<String>	getKeys(Object value, boolean batch, boolean isSearch)
int	getMaxWidth()
int	getMinWidth()
ArrayList<String>	getPurposeKeys(String in)
WordList	getRuleAnon()
WordList	getRuleCv()
WordList	getRuleLookup()
Class<Standardized>	getStandardClass()
String	getType()
void	init(String ruleFunction)
boolean	isMandatory()
boolean	isTextIndex()
protected HashSet<String>	processKey(ArrayList<HashSet<String>> alternates, int startpos, String keySoFar, boolean partialKeys, boolean isSearch, int searchWidth)
protected Collection<String>	processKeyInOrder(ArrayList<ArrayList<String>> alternates, int startpos, String keySoFar, boolean partialKeys)
void	setGradient(int gradientType)
void	setMandatory(boolean isMandatory)
void	setRuleFunction(String ruleFunction)
void	setRuleSet(RuleSet ruleset)
void	setType(String type)
String	sha256(String base)
Standardized	standardise(String originalText, String[] words)

Methods inherited from class java.lang.Object

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

Field Detail

ruleSet

protected RuleSet ruleSet

partialKeys

protected boolean partialKeys

type

protected String type

gradient

protected int gradient

isMandatory

protected boolean isMandatory

ruleFunction

protected String ruleFunction

maxScore

protected int maxScore

logger

protected Logger logger

minWidth

protected int minWidth

maxWidth

protected int maxWidth

Constructor Detail

MatchBase

public MatchBase(String ruleFunction,
                 int _minwidth,
                 int _maxwidth)
          throws Exception

Throws:

Exception

MatchBase

public MatchBase(String ruleFunction)

MatchBase

public MatchBase(String ruleFunction,
                 int minwidth,
                 int maxwidth,
                 String type)
          throws Exception

Throws:

Exception

Method Detail

compressInitials

protected ArrayList<String> compressInitials(String[] in)

init

public void init(String ruleFunction)

Specified by:

init in interface MatchProcInterface

getType

public String getType()

setRuleFunction

public void setRuleFunction(String ruleFunction)

Specified by:

setRuleFunction in interface MatchProcInterface

isMandatory

public boolean isMandatory()

setMandatory

public void setMandatory(boolean isMandatory)

setType

public void setType(String type)

setRuleSet

public void setRuleSet(RuleSet ruleset)

Specified by:

setRuleSet in interface MatchProcInterface

getKeyCv

public WordList getKeyCv()

Specified by:

getKeyCv in interface MatchProcInterface

getKeyAnon

public WordList getKeyAnon()

Specified by:

getKeyAnon in interface MatchProcInterface

getRuleAnon

public WordList getRuleAnon()

Specified by:

getRuleAnon in interface MatchProcInterface

getRuleCv

public WordList getRuleCv()

Specified by:

getRuleCv in interface MatchProcInterface

getKeyLookup

public WordList getKeyLookup()

getRuleLookup

public WordList getRuleLookup()

getAlternateWords

protected HashSet<String> getAlternateWords(String word,
                                            WordList anon,
                                            WordList controlVocabulary)

getKeys

public Collection<String> getKeys(Object value,
                                  boolean batch)

Specified by:

getKeys in interface MatchProcInterface

getKeys

public Collection<String> getKeys(Object value,
                                  boolean batch,
                                  boolean isSearch)

Specified by:

getKeys in interface MatchProcInterface

processKey

protected HashSet<String> processKey(ArrayList<HashSet<String>> alternates,
                                     int startpos,
                                     String keySoFar,
                                     boolean partialKeys,
                                     boolean isSearch,
                                     int searchWidth)

processKeyInOrder

protected Collection<String> processKeyInOrder(ArrayList<ArrayList<String>> alternates,
                                               int startpos,
                                               String keySoFar,
                                               boolean partialKeys)

allCombs

protected ArrayList<String> allCombs(String key,
                                     boolean partial,
                                     boolean alwaysUseFirst,
                                     int minWidth,
                                     int maxWidth)

sha256

public String sha256(String base)

Specified by:

sha256 in interface MatchProcInterface

getMinWidth

public int getMinWidth()

Specified by:

getMinWidth in interface MatchProcInterface

getMaxWidth

public int getMaxWidth()

Specified by:

getMaxWidth in interface MatchProcInterface

getKey

public String getKey(String word)

Specified by:

getKey in interface MatchProcInterface

isTextIndex

public boolean isTextIndex()

Specified by:

isTextIndex in interface MatchProcInterface

setGradient

public void setGradient(int gradientType)

Specified by:

setGradient in interface MatchProcInterface

getKeys

public Collection<String> getKeys(HashMap<String,String> value,
                                  boolean batch)

Specified by:

getKeys in interface MatchProcInterface

calculateComparisonScore

public double calculateComparisonScore(Standardized stdBase,
                                       Standardized stdComp,
                                       boolean isSearch,
                                       boolean asContent)

Specified by:

calculateComparisonScore in interface MatchProcInterface

standardise

public Standardized standardise(String originalText,
                                String[] words)

Specified by:

standardise in interface MatchProcInterface

getPurposeKeys

public ArrayList<String> getPurposeKeys(String in)

Specified by:

getPurposeKeys in interface MatchProcInterface

getStandardClass

public Class<Standardized> getStandardClass()

Specified by:

getStandardClass in interface MatchProcInterface