net.seninp.gi.clusterrule

Class RuleOrganizer

java.lang.Object

net.seninp.gi.clusterrule.RuleOrganizer

public class RuleOrganizer
extends Object

Constructor Summary

Constructors

Constructor and Description
RuleOrganizer()

Method Summary

Modifier and Type	Method and Description
ArrayList<SameLengthMotifs>	classifyMotifs(double lengthThreshold, GrammarRules grammarRules) Classify the motifs based on their length.
protected SAXPointsNumber[]	countPointNumber(GrammarRules grammarRules, double[] ts) This method counts how many times each data point is used in ANY sequitur rule (i.e.
SAXPointsNumber[]	countPointNumberAfterRemoving(double[] ts, ArrayList<SameLengthMotifs> allClassifiedMotifs) This method counts how many times each data point is used in REDUCED sequitur rule (i.e.
protected boolean	decideRemove(SAXMotif motif1, SAXMotif motif2, SAXPointsNumber[] pointsNumberRemoveStrategy) Decide which one from overlapping subsequences should be removed.
protected ArrayList<SAXMotif>	getAllMotifs(GrammarRules grammarRules) Stores all the sub-sequences that generated by Sequitur rules into an array list sorted by sub-sequence length in ascending order.
protected ArrayList<SameLengthMotifs>	refinePatternsByClustering(GrammarRules grammarRules, double[] ts, ArrayList<SameLengthMotifs> allClassifiedMotifs, double fractionTopDist) Refines patterns by clustering.
protected ArrayList<SameLengthMotifs>	removeOverlappingInSimiliar(ArrayList<SameLengthMotifs> allClassifiedMotifs, GrammarRules grammarRules, double[] ts, double thresouldCom) Removes overlapping rules in similar rule set.

Methods inherited from class java.lang.Object

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

Constructor Detail

RuleOrganizer

public RuleOrganizer()

Method Detail

classifyMotifs

public ArrayList<SameLengthMotifs> classifyMotifs(double lengthThreshold,
                                                  GrammarRules grammarRules)

Classify the motifs based on their length. It calls "getAllMotifs()" to get all the sub-sequences that were generated by Sequitur rules in ascending order. Then bins all the sub-sequences by length based on the length of the first sub-sequence in each class, that is, the shortest sub-sequence in each class.

Parameters:

lengthThreshold - the motif length threshold.

grammarRules - the rules set.

Returns:

the same length motifs.

getAllMotifs

protected ArrayList<SAXMotif> getAllMotifs(GrammarRules grammarRules)

Stores all the sub-sequences that generated by Sequitur rules into an array list sorted by sub-sequence length in ascending order.

Parameters:

grammarRules - the set of grammar rules.

Returns:

the list of all sub-sequences sorted by length in ascending order.

removeOverlappingInSimiliar

protected ArrayList<SameLengthMotifs> removeOverlappingInSimiliar(ArrayList<SameLengthMotifs> allClassifiedMotifs,
                                                                  GrammarRules grammarRules,
                                                                  double[] ts,
                                                                  double thresouldCom)

Removes overlapping rules in similar rule set.

Parameters:

allClassifiedMotifs - the set of motifs classified as the same.

grammarRules - the grammar.

ts - the input time series.

thresouldCom - the threshold.

Returns:

a reduced set of the same length rules.

countPointNumberAfterRemoving

public SAXPointsNumber[] countPointNumberAfterRemoving(double[] ts,
                                                       ArrayList<SameLengthMotifs> allClassifiedMotifs)

This method counts how many times each data point is used in REDUCED sequitur rule (i.e. data point 1 appears only in R1 and R2, the number for data point 1 is two). The function will get the occurrence time for all points, and write the result into a text file named as "PointsNumberAfterRemoving.txt".

Parameters:

ts - the input time series.

allClassifiedMotifs - the motifs classified into a same set.

Returns:

points count.

decideRemove

protected boolean decideRemove(SAXMotif motif1,
                               SAXMotif motif2,
                               SAXPointsNumber[] pointsNumberRemoveStrategy)

Decide which one from overlapping subsequences should be removed. The decision rule is that each sub-sequence has a weight, the one with the smaller weight should be removed. The weight is S/(A * L). S is the sum of occurrence time of all data points in that sub-sequence, A is the average weight of the whole time series, and L is the length of that sub-sequence.

Parameters:

motif1 - the motif1.

motif2 - the motif2.

pointsNumberRemoveStrategy - the strategy for cleaning up.

Returns:

the decision.

countPointNumber

protected SAXPointsNumber[] countPointNumber(GrammarRules grammarRules,
                                             double[] ts)

This method counts how many times each data point is used in ANY sequitur rule (i.e. data point 1 appears only in R1 and R2, the number for data point 1 is two). The function will get the occurrence time for all points, and write the result into a text file named as "PointsNumber.txt".

Parameters:

grammarRules - the grammar.

ts - the time series.

Returns:

the points count.

refinePatternsByClustering

protected ArrayList<SameLengthMotifs> refinePatternsByClustering(GrammarRules grammarRules,
                                                                 double[] ts,
                                                                 ArrayList<SameLengthMotifs> allClassifiedMotifs,
                                                                 double fractionTopDist)

Refines patterns by clustering.

Parameters:

grammarRules - the set of grammar rules.

ts - the input time series.

allClassifiedMotifs - all the motifs in groups.

fractionTopDist - the fraction threshold.

Returns:

refined set of patterns.