Package one.microstream.util.similarity

Interface MultiMatcher<E>

Type Parameters:

E - the type of the elements being matched.

All Known Implementing Classes:

MultiMatcher.Default

public interface MultiMatcher<E>

Logic for bidirectionally and exclusively linking all matching elements from two collections according to equality and/or sufficient similarity.

Exclusviely means each element in both collections can at most be linked with one element from the other collection. Bidirectionally means the link between two elements always has two directions. If element A is linked to element B, element B is inherently linked to element A as well.

Equality and similarity are defined by Equalator and Similator functions that can be passed at creation time. All values controlling the matching algorithm can be optionally configured in the factory class if the default configuration is not desired. Additionally, a callback function for deciding found matches with questionable similarity can be injected.

This is a powerful general purpose means of building associations of two sets of similar but not equal elements.
A very simple use case is the formal recognition of a changed table column structure (for which this class was originally developed).

For example given the following two hypothetical definitions (old and new) of column names:

Old:

• Name
• Firstname
• Age
• Address
• Freetext
• Email
• OtherAddress

and

New:

• firstname
• lastname
• age
• emailAddress
• postalAddress
• noteLink
• newColumn1
• someMiscAddress

When using a case insensitive modified Levenshtein Similator(see Levenshtein.substringSimilarity(java.lang.String, java.lang.String)) the algorithm produces the following associations:

 firstname       -1.00- Firstname
 lastname        -0.75- Name
 age             -1.00- Age
 emailAddress    -0.71- Email
 postalAddress   -0.77- Address
 noteLink        [new]
 newColumn1      [new]
 someMiscAddress -0.56- OtherAddress
 X Freetext
 

Nested Class Summary

Nested Classes

Modifier and Type	Interface	Description
static class	MultiMatcher.Default<E>

Method Summary

Modifier and Type	Method	Description
static double	defaultNoiseFactor()
static double	defaultSimilarityThreshold()
static double	defaultSingletonPrecedenceBonus()
static double	defaultSingletonPrecedenceThreshold()
Equalator<? super E>	equalator()
MultiMatch<E>	match(XGettingCollection<? extends E> source, XGettingCollection<? extends E> target)
static <E> MultiMatcher<E>	New()
double	noiseFactor()
MultiMatcher<E>	setEqualator(Equalator<? super E> equalator)
MultiMatcher<E>	setNoisefactor(double noiseFactor)
MultiMatcher<E>	setSimilarityThreshold(double similarityThreshold)
MultiMatcher<E>	setSimilator(Similator<? super E> similator)
MultiMatcher<E>	setSingletonPrecedenceBonus(double singletonPrecedenceBonus)
MultiMatcher<E>	setSingletonPrecedenceThreshold(double singletonPrecedenceThreshold)
MultiMatcher<E>	setValidator(MatchValidator<? super E> validator)
double	similarityThreshold()
Similator<? super E>	similator()
double	singletonPrecedenceBonus()
double	singletonPrecedenceThreshold()	This is a measure of how "eager" the algorithm is to find as many matches as possible.
MatchValidator<? super E>	validator()

Method Detail

similarityThreshold

double similarityThreshold()

singletonPrecedenceThreshold

double singletonPrecedenceThreshold()

This is a measure of how "eager" the algorithm is to find as many matches as possible. The lower this threshold is, the more "single potential match" items will be preferred over actually better matching pairs just to not leave them unmatched. To deactivate this special casing, set the threshold to 1.0, meaning only items that fit perfectly anyway take precedence over others.

singletonPrecedenceBonus

double singletonPrecedenceBonus()

noiseFactor

double noiseFactor()

equalator

Equalator<? super E> equalator()

similator

Similator<? super E> similator()

validator

MatchValidator<? super E> validator()

setSimilarityThreshold

MultiMatcher<E> setSimilarityThreshold(double similarityThreshold)

setSingletonPrecedenceThreshold

MultiMatcher<E> setSingletonPrecedenceThreshold(double singletonPrecedenceThreshold)

setSingletonPrecedenceBonus

MultiMatcher<E> setSingletonPrecedenceBonus(double singletonPrecedenceBonus)

setNoisefactor

MultiMatcher<E> setNoisefactor(double noiseFactor)

setSimilator

MultiMatcher<E> setSimilator(Similator<? super E> similator)

setEqualator

MultiMatcher<E> setEqualator(Equalator<? super E> equalator)

setValidator

MultiMatcher<E> setValidator(MatchValidator<? super E> validator)

match

MultiMatch<E> match(XGettingCollection<? extends E> source,
                    XGettingCollection<? extends E> target)

defaultSimilarityThreshold

static double defaultSimilarityThreshold()

defaultSingletonPrecedenceThreshold

static double defaultSingletonPrecedenceThreshold()

defaultSingletonPrecedenceBonus

static double defaultSingletonPrecedenceBonus()

defaultNoiseFactor

static double defaultNoiseFactor()

New

static <E> MultiMatcher<E> New()