Changeset 1620 for branches/ralph/src/main/java/de/ugoe/cs/autoquest/tasktrees/temporalrelation/SequenceForTaskDetectionRuleAlignment.java

Timestamp:

07/23/14 18:18:11 (10 years ago)

Author:

rkrimmel

Message:

More intelligent match finding and creation

File:

• branches/ralph/src/main/java/de/ugoe/cs/autoquest/tasktrees/temporalrelation/SequenceForTaskDetectionRuleAlignment.java (modified) (5 diffs)

branches/ralph/src/main/java/de/ugoe/cs/autoquest/tasktrees/temporalrelation/SequenceForTaskDetectionRuleAlignment.java

@@ -26 +26 @@
 
 import de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.AlignmentAlgorithmFactory;
+import de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.Match;
 import de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.NumberSequence;
 import de.ugoe.cs.autoquest.tasktrees.alignment.matrix.PairwiseAlignmentGenerator;
@@ -155 +156 @@
                 Console.traceln(Level.INFO,"generating substitution matrix");
                 ObjectDistanceSubstitionMatrix submat = new ObjectDistanceSubstitionMatrix(
-                                uniqueTasks, 5);
+                                uniqueTasks, 6,-3);
                 submat.generate();
 
                 Console.traceln(Level.INFO,"generating pairwise alignments");
-                ArrayList<ArrayList<NumberSequence>> matchseqs = new ArrayList<ArrayList<NumberSequence>>();
-                PairwiseAlignmentStorage alignments = PairwiseAlignmentGenerator.generate(numberseqs,submat);
+                ArrayList<Match> matchseqs = new ArrayList<Match>();
+                PairwiseAlignmentStorage alignments = PairwiseAlignmentGenerator.generate(numberseqs,submat,9);
                 
                 Console.traceln(Level.INFO,"retrieving significant sequence pieces");
@@ -167 +168 @@
                         for(int j=0; j< numberseqs.size();j++) {
                                 if(i != j) {
-                                        ArrayList<ArrayList<NumberSequence>> tmp = alignments.get(i, j).getMatches();
-                                        matchseqs.addAll(tmp);
+                                        matchseqs.addAll(alignments.get(i, j).getMatches());
                                 }   
                         }
@@ -178 +178 @@
                 //search this match in every other sequence
                 for(int i=0; i < matchseqs.size();i++) {
-                        int occurencecount = 0;
+                        int sequencecount = 0;
+                        int totalcount = 0;
+                        Match pattern = matchseqs.get(i);
+                        
+                
+                        
+                        //Skip sequences with more 0 events (scrolls) than other events. Both of the pattern sequences are equally long, so the added 0 counts 
+                        //just need to be smaller than the length of one sequence 
+                        if(pattern.getFirstSequence().eventCount(0) + pattern.getSecondSequence().eventCount(0) +1 > pattern.getFirstSequence().size())
+                                continue;
                         Console.traceln(Level.FINEST, "searching for patterns occuring most: " + Math.round((float)i/(float)matchseqs.size()*100) + "%");
-                        for(int j=0; j < matchseqs.size();j++) {
-                                if(i>j) {
-                                        if(matchseqs.get(j).get(0).containsPattern(matchseqs.get(i)) > 0 || matchseqs.get(j).get(1).containsPattern(matchseqs.get(i)) > 0) {
-                                                occurencecount++;
+                        for(int j=0; j < numberseqs.size();j++) {
+                                        int tmpcount = numberseqs.get(j).containsPattern(pattern);
+                                        if(tmpcount > 0) {
+                                                sequencecount++;
+                                                totalcount+=tmpcount;
                                         }
-                                }
-                        }
-                        if(occurencecount > 1) {
-                                System.out.println("Found pattern \n "); 
-                                matchseqs.get(i).get(0).printSequence(); 
-                                matchseqs.get(i).get(1).printSequence(); 
-                                System.out.println(occurencecount + " times");
+                        }
+                        
+                        if(totalcount > 1) {
+                                matchseqs.get(i).getFirstSequence().printSequence(); 
+                                matchseqs.get(i).getFirstSequence().printSequence();
+                                System.out.println("Found pattern in " + sequencecount +"/" + numberseqs.size() + " sequences, total of " + totalcount + " occurences");
                                 System.out.println();
                         }
                 }
-                
-                
-                
                 alignments = null;
                 
@@ -290 +296 @@
                                 }
                         }
+                        templist.setId(numberseqs.size());
                         numberseqs.add(templist);
                         comparator.clearBuffers();