-
Notifications
You must be signed in to change notification settings - Fork 0
/
E2E.java
4392 lines (4109 loc) · 228 KB
/
E2E.java
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
package structuredPredictionNLG;
import com.google.common.collect.Lists;
import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileNotFoundException;
import java.io.FileOutputStream;
import java.io.FileReader;
import java.io.IOException;
import java.io.ObjectInputStream;
import java.io.ObjectOutputStream;
import java.io.OutputStreamWriter;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Map;
import java.util.Random;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import edu.stanford.nlp.mt.metrics.BLEUMetric;
import edu.stanford.nlp.mt.tools.NISTTokenizer;
import edu.stanford.nlp.mt.util.IString;
import edu.stanford.nlp.mt.util.IStrings;
import edu.stanford.nlp.mt.util.ScoredFeaturizedTranslation;
import edu.stanford.nlp.mt.util.Sequence;
import gnu.trove.map.hash.TObjectDoubleHashMap;
import imitationLearning.JLOLS;
import jarow.Instance;
import jarow.JAROW;
import jarow.Prediction;
import java.util.List;
import similarity_measures.Levenshtein;
import similarity_measures.Rouge;
import simpleLM.SimpleLM;
class ValueComparator implements Comparator<String> {
Map<String, Integer> base;
public ValueComparator(HashMap<String, Integer> wordDictionary) {
this.base = wordDictionary;
}
// Note: this comparator imposes orderings that are inconsistent with
// equals.
public int compare(String a, String b) {
if (base.get(a) >= base.get(b)) {
return -1;
} else {
return 1;
} // returning 0 would merge keys
}
}
public class E2E extends DatasetParser {
final String singlePredicate = "#PREDICATE";
private JLOLS ILEngine;
public E2E(String[] args) {
super(args);
}
public static void main(String[] args) {
E2E e2e = new E2E(args);
e2e.parseDataset();
e2e.ILEngine = new JLOLS(e2e);
JLOLS.targettedExploration = 10;
JLOLS.batchSize = 1000;
long start_time = System.currentTimeMillis();
e2e.createTrainingData();
long end_time = System.currentTimeMillis();
long running_time = end_time - start_time;
System.out.println("running time is " + running_time);
e2e.performImitationLearning(e2e.ILEngine);
}
@Override
public void parseDataset() {
File trainingDataFile = new File("e2e_traindev/trainset.csv");
File devDataFile = new File("e2e_traindev/devset.csv");
// Initialize the collections
setPredicates(new ArrayList<>());
setAttributes(new HashMap<>());
setAttributeValuePairs(new HashMap<>());
setValueAlignments(new HashMap<>());
getAttributes().put(singlePredicate, new HashSet<>());
getDatasetInstances().put(singlePredicate, new ArrayList<>());
if (isResetStoredCaches() || !loadLists()) {
createLists(trainingDataFile);
getTrainingData().addAll(getDatasetInstances().get(singlePredicate)
//.subList(0, 1000)
);
getDatasetInstances().put(singlePredicate, new ArrayList<>());
createLists(devDataFile);
//Collections.shuffle(getDatasetInstances().get(singlePredicate), new Random(123));
getValidationData().addAll(getDatasetInstances().get(singlePredicate)
//.subList(0, 100)
);
// Create the refs for DEV data, as described in https://github.com/tuetschek/e2e-metrics/tree/master/example-inputs
for (DatasetInstance di : getValidationData()) {
HashSet<String> refs = new HashSet<>();
for (DatasetInstance di2 : getValidationData()) {
if (di2.getMeaningRepresentation().getMRstr().equals(di.getMeaningRepresentation().getMRstr())) {
refs.add(di2.getDirectReference());
}
}
di.setEvaluationReferences(refs);
}
getDatasetInstances().get(singlePredicate).addAll(getTrainingData());
writeLists();
}
// Dataset analysis and re-splitting
/*HashMap<String, String> uniqueValMRsTrain = new HashMap<>();
HashMap<String, String> distinctValMRsTrain = new HashMap<>();
for (DatasetInstance val : getTrainingData()) {
boolean distinct = true;
if (!uniqueValMRsTrain.containsKey(val.getMeaningRepresentation().getAbstractMR())) {
String evalRefs = "";
boolean isFirst = true;
for (String r : val.getEvaluationReferences()) {
if (isFirst) {
evalRefs += r;
} else {
evalRefs += " | " + r;
}
isFirst = false;
}
evalRefs = evalRefs.trim();
uniqueValMRsTrain.put(val.getMeaningRepresentation().getAbstractMR(), "\"" + val.getMeaningRepresentation().getMRstr() + "\"," + evalRefs);
}
for (DatasetInstance tr : getValidationData()) {
if (tr.getMeaningRepresentation().getAbstractMR().equals(val.getMeaningRepresentation().getAbstractMR())) {
//if (tr.getMeaningRepresentation().getAttributeValues().keySet().equals(val.getMeaningRepresentation().getAttributeValues().keySet())) {
distinct = false;
}
}
if (distinct) {
System.out.println(val.getMeaningRepresentation().getAttributeValues().keySet());
System.out.println("\"" + val.getMeaningRepresentation().getMRstr() + "\"," + val.getDirectReference());
if (!distinctValMRsTrain.containsKey(val.getMeaningRepresentation().getAbstractMR())) {
String evalRefs = "";
boolean isFirst = true;
for (String r : val.getEvaluationReferences()) {
if (isFirst) {
evalRefs += r;
} else {
evalRefs += " | " + r;
}
isFirst = false;
}
evalRefs = evalRefs.trim();
distinctValMRsTrain.put(val.getMeaningRepresentation().getAbstractMR(), "\"" + val.getMeaningRepresentation().getMRstr() + "\"," + evalRefs);
}
}
}
HashMap<String, String> uniqueValMRsDev = new HashMap<>();
HashMap<String, String> distinctValMRsDev = new HashMap<>();
for (DatasetInstance val : getValidationData()) {
boolean distinct = true;
if (!uniqueValMRsDev.containsKey(val.getMeaningRepresentation().getAbstractMR())) {
String evalRefs = "";
boolean isFirst = true;
for (String r : val.getEvaluationReferences()) {
if (isFirst) {
evalRefs += r;
} else {
evalRefs += " | " + r;
}
isFirst = false;
}
evalRefs = evalRefs.trim();
uniqueValMRsDev.put(val.getMeaningRepresentation().getAbstractMR(), "\"" + val.getMeaningRepresentation().getMRstr() + "\"," + evalRefs);
}
for (DatasetInstance tr : getTrainingData()) {
if (tr.getMeaningRepresentation().getAbstractMR().equals(val.getMeaningRepresentation().getAbstractMR())) {
//if (tr.getMeaningRepresentation().getAttributeValues().keySet().equals(val.getMeaningRepresentation().getAttributeValues().keySet())) {
distinct = false;
}
}
if (distinct) {
System.out.println(val.getMeaningRepresentation().getAttributeValues().keySet());
System.out.println("\"" + val.getMeaningRepresentation().getMRstr() + "\"," + val.getDirectReference());
if (!distinctValMRsDev.containsKey(val.getMeaningRepresentation().getAbstractMR())) {
String evalRefs = "";
boolean isFirst = true;
for (String r : val.getEvaluationReferences()) {
if (isFirst) {
evalRefs += r;
} else {
evalRefs += " | " + r;
}
isFirst = false;
}
evalRefs = evalRefs.trim();
distinctValMRsDev.put(val.getMeaningRepresentation().getAbstractMR(), "\"" + val.getMeaningRepresentation().getMRstr() + "\"," + evalRefs);
}
}
}
HashMap<Integer, HashMap<String, String>> uniqueValMRsAll = new HashMap<>();
ArrayList<DatasetInstance> all = new ArrayList<>();
all.addAll(getTrainingData());
all.addAll(getValidationData());
for (DatasetInstance val : all) {
if (!uniqueValMRsAll.containsKey(val.getMeaningRepresentation().getAbstractMR())) {
String evalRefs = "";
boolean isFirst = true;
for (String r : val.getEvaluationReferences()) {
if (isFirst) {
evalRefs += r;
} else {
evalRefs += " | " + r;
}
isFirst = false;
}
evalRefs = evalRefs.trim();
if (!uniqueValMRsAll.containsKey(val.getMeaningRepresentation().getAttributeValues().keySet().size())) {
uniqueValMRsAll.put(val.getMeaningRepresentation().getAttributeValues().keySet().size(), new HashMap<>());
}
uniqueValMRsAll.get(val.getMeaningRepresentation().getAttributeValues().keySet().size()).put(val.getMeaningRepresentation().getAbstractMR(), "\"" + val.getMeaningRepresentation().getMRstr() + "\"," + evalRefs);
}
}
HashMap<String, String> uniqueValMRsNewTrain = new HashMap<>();
HashMap<String, String> uniqueValMRsNewDev = new HashMap<>();
for (Integer s : uniqueValMRsAll.keySet()) {
ArrayList<String> keys = new ArrayList<String>(uniqueValMRsAll.get(s).keySet());
Collections.shuffle(keys, new Random(123));
int split = ((Double) java.lang.Math.floor(keys.size() * 0.9)).intValue();
for (String k : keys.subList(0, split)) {
uniqueValMRsNewTrain.put(k, uniqueValMRsAll.get(s).get(k));
}
for (String k : keys.subList(split, keys.size())) {
uniqueValMRsNewDev.put(k, uniqueValMRsAll.get(s).get(k));
}
}
try {
java.io.PrintWriter outT = new java.io.PrintWriter(new java.io.FileWriter("unique_trainset.csv"));
outT.println("mr,ref");
for (String s : uniqueValMRsTrain.values()) {
outT.println(s);
}
outT.flush();
outT.close();
java.io.PrintWriter out = new java.io.PrintWriter(new java.io.FileWriter("unique_devset.csv"));
out.println("mr,ref");
for (String s : uniqueValMRsDev.values()) {
out.println(s);
}
out.flush();
out.close();
java.io.PrintWriter out2 = new java.io.PrintWriter(new java.io.FileWriter("distinct_devset.csv"));
out2.println("mr,ref");
for (String s : distinctValMRsDev.values()) {
out2.println(s);
}
out2.flush();
out2.close();
java.io.PrintWriter out3 = new java.io.PrintWriter(new java.io.FileWriter("uniqueAndDistnct_newTrainSet.csv"));
out3.println("mr,ref");
for (String s : uniqueValMRsNewTrain.values()) {
out3.println(s);
}
out3.flush();
out3.close();
java.io.PrintWriter out4 = new java.io.PrintWriter(new java.io.FileWriter("uniqueAndDistnct_newDevSet.csv"));
out4.println("mr,ref");
for (String s : uniqueValMRsNewDev.values()) {
out4.println(s);
}
out4.flush();
out4.close();
} catch (IOException ex) {
Logger.getLogger(E2E.class.getName()).log(Level.SEVERE, null, ex);
}
System.out.println(uniqueValMRsNewTrain.keySet().size() + " / " + all.size());
System.out.println(uniqueValMRsNewDev.keySet().size() + " / " + all.size());
System.out.println(uniqueValMRsTrain.keySet().size() + " / " + getTrainingData().size());
System.out.println(distinctValMRsTrain.keySet().size() + " / " + getTrainingData().size());
System.out.println(uniqueValMRsDev.keySet().size() + " / " + getValidationData().size());
System.out.println(distinctValMRsDev.keySet().size() + " / " + getValidationData().size());
System.exit(0);*/
System.out.println("Training data size: " + getTrainingData().size());
System.out.println("Validation data size: " + getValidationData().size());
}
public void createLists(File dataFile) {
System.out.println("create lists");
ArrayList<String> dataPart = new ArrayList<>();
//we read in the data from the data files.
try {
BufferedReader br = new BufferedReader(new FileReader(dataFile));
String s;
try {
s = br.readLine();
while (s != null) {
dataPart.add(s);
s = br.readLine();
}
} catch (IOException e) {
e.printStackTrace();
}
try {
br.close();
} catch (IOException e) {
}
} catch (FileNotFoundException e) {
e.printStackTrace();
}
// in this data set we don't have predicate so set a single predicate just for populate the data structure
getPredicates().add(singlePredicate);
// remove the column name line
dataPart.remove(0);
// fix the errors in the data set
for (int i = 0; i < dataPart.size(); i++) {
if (!dataPart.get(i).contains("\",")) {
String line = dataPart.get(i - 1);
line = line + dataPart.get(i);
dataPart.remove(i);
}
}
// for each instance
int num = 0;
for (String line : dataPart) {
//System.out.println(num);
num++;
String MRPart = line.split("\",")[0];
String RefPart = line.split("\",")[1].toLowerCase();
if (RefPart.equals("café")) {
continue;
}
if (MRPart.startsWith("\"")) {
MRPart = MRPart.substring(1);
}
if (RefPart.startsWith("\"")) {
RefPart = RefPart.substring(1);
}
if (RefPart.endsWith("\"")) {
RefPart = RefPart.substring(0, RefPart.length() - 1);
}
String[] MRs = MRPart.split(",");
// original value to delexicalized value
HashMap<String, String> delexicalizedMap = new HashMap<>();
// each instance's attribute value pairs
HashMap<String, HashSet<String>> attributeValues = new HashMap<>();
// for each attribute value pairs
for (String mr : MRs) {
String value = mr.substring(mr.indexOf("[") + 1, mr.indexOf("]")).trim().toLowerCase();
String attribute = mr.substring(0, mr.indexOf("[")).trim().toLowerCase();
if (attribute.equals("name")) {
// delexicalize name values
String delexValue = Action.TOKEN_X + attribute + "_0";
delexicalizedMap.put(delexValue, value);
value = delexValue;
}
if (attribute.equals("near")) {
//delexicalize near values
String delexValue = Action.TOKEN_X + attribute + "_0";
delexicalizedMap.put(delexValue, value);
value = delexValue;
}
if (value.equals("yes") || value.equals("no")) {
value = attribute + "_" + value;
}
getAttributes().put(singlePredicate, new HashSet<>());
if (attribute != null) {
getAttributes().get(singlePredicate).add(attribute);
if (!getAttributeValuePairs().containsKey(attribute)) {
getAttributeValuePairs().put(attribute, new HashSet<String>());
}
if (!attributeValues.containsKey(attribute)) {
attributeValues.put(attribute, new HashSet<>());
}
if (value != null) {
getAttributeValuePairs().get(attribute).add(value);
attributeValues.get(attribute).add(value);
}
}
}
//RefPart = " "+RefPart+" ";
for (String deValue : delexicalizedMap.keySet()) {
String value = delexicalizedMap.get(deValue);
if (RefPart.contains(value)) {
RefPart = RefPart.replace(value, deValue).trim();
}
}
//if(!RefPart.contains("@x@name_0")&&!RefPart.contains("@x@near_0")){
//System.out.println(RefPart);
//}
/*
if(RefPart.contains("name-")){
RefPart = RefPart.replace("name-","name -");
}
if(RefPart.contains("@x@names")){
RefPart = RefPart.replace("@x@names", "@x@name s");
}
if(RefPart.contains("@x@nearian")){
RefPart = RefPart.replace("@x@nearian", "@x@near");
}
if(RefPart.contains("@x@namen")){
RefPart = RefPart.replace("@x@namen", "@x@name");
}
if(RefPart.contains("@x@nearn")){
RefPart = RefPart.replace("@x@nearn", "@x@near");
}
if(RefPart.contains("@x@nears")){
RefPart = RefPart.replace("@x@nears", "@x@near s");
}
if(RefPart.contains("@x@name_0s")){
RefPart = RefPart.replace("@x@name_0s", "@x@name_0 s");
}
if(RefPart.contains("@x@near_0s")){
RefPart = RefPart.replace("@x@near_0s", "@x@near_0 s");
}*/
// create MR for each instance
//MeaningRepresentation MR = new MeaningRepresentation(singlePredicate,attributeValues,MRPart,delexicalizedMap);
// start create the value alignments
ArrayList<String> observedAttrValueSequence = new ArrayList<>();
ArrayList<String> observedWordSequence = new ArrayList<>();
// replace the punctuation in the reference
//RefPart = RefPart.replaceAll("[.,?:;!'-]", " "+Action.TOKEN_PUNCT+" ");
//String[] words = RefPart.replaceAll("[.,?:;!'-]", " "+Action.TOKEN_PUNCT+" ").split(" ");
String[] words = RefPart.replace(", ,", " , ").replace(". .", " . ").replaceAll("[.,?:;!'-]", " $0 ").split("\\s+");
for (String w : words) {
if (w.contains("0f")) {
w = w.replace("0f", "of");
}
Pattern p1 = Pattern.compile("([0-9]+)([a-z]+)");
Matcher m1 = p1.matcher(w);
Pattern p2 = Pattern.compile("([a-z]+)([0-9]+)");
Matcher m2 = p2.matcher(w);
Pattern p3 = Pattern.compile("(£)([a-z]+)");
Matcher m3 = p3.matcher(w);
Pattern p4 = Pattern.compile("([a-z]+)(£[0-9]+)");
Matcher m4 = p4.matcher(w);
Pattern p5 = Pattern.compile("([0-9]+)([a-z]+)([0-9]+)");
Matcher m5 = p5.matcher(w);
Pattern p6 = Pattern.compile("([0-9]+)(@x@[a-z]+_0)");
Matcher m6 = p6.matcher(w);
if (m1.find()) {
observedWordSequence.add(m1.group(1).trim());
observedWordSequence.add(m1.group(2).trim());
} else if (m2.find()) {
observedWordSequence.add(m2.group(1).trim());
observedWordSequence.add(m2.group(2).trim());
} else if (m3.find()) {
observedWordSequence.add(m3.group(1).trim());
observedWordSequence.add(m3.group(2).trim());
} else if (m4.find()) {
observedWordSequence.add(m4.group(1).trim());
observedWordSequence.add(m4.group(2).trim());
} else if (m5.find()) {
observedWordSequence.add(m5.group(1).trim());
observedWordSequence.add(m5.group(2).trim());
observedWordSequence.add(m5.group(3).trim());
} else if (m6.find()) {
observedWordSequence.add(m6.group(1).trim());
observedWordSequence.add(m6.group(2).trim());
} else if (w.contains("@x@name_0") && !w.matches("@x@name_0")) {
String realValue = delexicalizedMap.get("@x@name_0");
realValue = w.replace("@x@name_0", realValue);
delexicalizedMap.put("@x@name_0", realValue);
w = "@x@name_0";
observedWordSequence.add(w.trim());
} else if (w.contains("@x@near_0") && !w.equals("@x@near_0")) {
String realValue = delexicalizedMap.get("@x@near_0");
realValue = w.replace("@x@near_0", realValue);
delexicalizedMap.put("@x@near_0", realValue);
w = "@x@near_0";
observedWordSequence.add(w.trim());
} else {
observedWordSequence.add(w.trim());
}
}
MeaningRepresentation MR = new MeaningRepresentation(singlePredicate, attributeValues, MRPart, delexicalizedMap);
// We store the maximum observed word sequence length, to use as a limit during generation
if (observedWordSequence.size() > getMaxWordSequenceLength()) {
setMaxWordSequenceLength(observedWordSequence.size());
}
// We initialize the alignments between words and attribute/value pairs
ArrayList<String> wordToAttrValueAlignment = new ArrayList<>();
for (String w : observedWordSequence) {
if (w.trim().matches("[.,?:;!'\"]")) {
wordToAttrValueAlignment.add(Action.TOKEN_PUNCT);
} else {
wordToAttrValueAlignment.add("[]");
}
}
ArrayList<Action> directReferenceSequence = new ArrayList<>();
for (int r = 0; r < observedWordSequence.size(); r++) {
directReferenceSequence.add(new Action(observedWordSequence.get(r), wordToAttrValueAlignment.get(r)));
}
DatasetInstance DI = new DatasetInstance(MR, directReferenceSequence, postProcessRef(MR, directReferenceSequence));
getDatasetInstances().get(singlePredicate).stream().filter((existingDI) -> (existingDI.getMeaningRepresentation().getAbstractMR()
.equals(DI.getMeaningRepresentation().getAbstractMR()))).map((existingDI) -> {
existingDI.getEvaluationReferences().addAll(DI.getEvaluationReferences());
return existingDI;
}).forEachOrdered((existingDI) -> {
// We add the direct reference of this DatasetInstance as an available evaluation reference to all previously constructed DatasetInstance that are identical to this one
DI.getEvaluationReferences().addAll(existingDI.getEvaluationReferences());
});
getDatasetInstances().get(singlePredicate).add(DI);
// value alignments
HashMap<String, HashMap<String, Double>> observedValueAlignments = new HashMap<>();
MR.getAttributeValues().keySet().stream().forEach((attr) -> {
MR.getAttributeValues().get(attr).stream().filter((value) -> (!value.startsWith(Action.TOKEN_X)))
.forEachOrdered((value) -> {
String valueToCompare = value;
//if(valueToCompare.contains("familyfriendly")){
//valueToCompare = valueToCompare.replace("familyfriendly", "family friendly");
//}
observedValueAlignments.put(valueToCompare, new HashMap<String, Double>());
// n grams
for (int n = 1; n < observedWordSequence.size(); n++) {
//Calculate the similarities between them and valueToCompare
for (int r = 0; r <= observedWordSequence.size() - n; r++) {
boolean compareAgainstNGram = true;
for (int j = 0; j < n; j++) {
if (observedWordSequence.get(r + j).startsWith(Action.TOKEN_X)
|| wordToAttrValueAlignment.get(r + j).equals(Action.TOKEN_PUNCT)
|| observedWordSequence.get(r + j).isEmpty()) {
compareAgainstNGram = false;
}
}
if (compareAgainstNGram) {
String align = "";
String compare = "";
String backwardCompare = "";
for (int j = 0; j < n; j++) {
// The coordinates of the alignment
align += (r + j) + " ";
compare += observedWordSequence.get(r + j);
backwardCompare = observedWordSequence.get(r + j) + backwardCompare;
}
align = align.trim();
// Calculate the character-level distance between the value and the nGram (in its original and reversed order)
Double distance = Levenshtein.getSimilarity(valueToCompare.toLowerCase(), compare.toLowerCase(), true);
Double backwardDistance = Levenshtein.getSimilarity(valueToCompare.toLowerCase(), backwardCompare.toLowerCase(), true);
// We keep the best distance score; note that the Levenshtein distance is normalized so that greater is better
if (backwardDistance > distance) {
distance = backwardDistance;
}
// We ignore all nGrams that are less similar than a threshold
if (valueToCompare.equals("5 out of 5")
|| valueToCompare.equals("1 out of 5")
|| valueToCompare.equals("3 out of 5")) {
if (distance > 0.1) {
observedValueAlignments.get(valueToCompare).put(align, distance);
}
} else if (valueToCompare.equals("familyfriendly_no")
|| valueToCompare.equals("familyfriendly_yes")) {
if (distance > 0.1) {
observedValueAlignments.get(valueToCompare).put(align, distance);
}
} else if (valueToCompare.equals("more than £30")
|| valueToCompare.equals("£20-25")
|| valueToCompare.equals("less than £20")) {
if (distance > 0.1) {
observedValueAlignments.get(valueToCompare).put(align, distance);
}
} else {
if (distance > 0.65) {
observedValueAlignments.get(valueToCompare).put(align, distance);
}
}
}
}
}
});
});
// We filter out any values that haven't been aligned
HashSet<String> toRemove = new HashSet<>();
for (String value : observedValueAlignments.keySet()) {
if (observedValueAlignments.get(value).isEmpty()) {
toRemove.add(value);
}
}
for (String value : toRemove) {
observedValueAlignments.remove(value);
}
while (!observedValueAlignments.keySet().isEmpty()) {
// Find the best aligned nGram
Double max = Double.NEGATIVE_INFINITY;
String[] bestAlignment = new String[2];
for (String value : observedValueAlignments.keySet()) {
for (String alignment : observedValueAlignments.get(value).keySet()) {
if (observedValueAlignments.get(value).get(alignment) > max) {
max = observedValueAlignments.get(value).get(alignment);
bestAlignment[0] = value;
bestAlignment[1] = alignment;
}
}
}
// Find the subphrase that corresponds to the best aligned nGram, according to the coordinates
ArrayList<String> alignedStr = new ArrayList<>();
String[] coords = bestAlignment[1].split(" ");
if (coords.length == 1) {
alignedStr.add(observedWordSequence.get(Integer.parseInt(coords[0].trim())));
} else {
for (int a = Integer.parseInt(coords[0].trim()); a <= Integer.parseInt(coords[coords.length - 1].trim()); a++) {
alignedStr.add(observedWordSequence.get(a));
}
}
// Store the best aligned nGram
if (!getValueAlignments().containsKey(bestAlignment[0])) {
getValueAlignments().put(bestAlignment[0], new HashMap<ArrayList<String>, Double>());
}
getValueAlignments().get(bestAlignment[0]).put(alignedStr, max);
// And remove it from the observed ones for this instance
observedValueAlignments.remove(bestAlignment[0]);
// And also remove any other aligned nGrams that are overlapping with the best aligned nGram
observedValueAlignments.keySet().forEach((value) -> {
HashSet<String> alignmentsToBeRemoved = new HashSet<>();
observedValueAlignments.get(value).keySet().forEach((alignment) -> {
String[] othCoords = alignment.split(" ");
if (Integer.parseInt(coords[0].trim()) <= Integer.parseInt(othCoords[0].trim()) && (Integer.parseInt(coords[coords.length - 1].
trim()) >= Integer.parseInt(othCoords[0].trim()))
|| (Integer.parseInt(othCoords[0].trim()) <= Integer.parseInt(coords[0].trim()) && Integer.parseInt(othCoords[othCoords.length - 1].
trim()) >= Integer.parseInt(coords[0].trim()))) {
alignmentsToBeRemoved.add(alignment);
}
});
alignmentsToBeRemoved.forEach((alignment) -> {
observedValueAlignments.get(value).remove(alignment);
});
});
// We filter out any values that are no longer aligned (due to overlapping conflicts)
toRemove = new HashSet<>();
for (String value : observedValueAlignments.keySet()) {
if (observedValueAlignments.get(value).isEmpty()) {
toRemove.add(value);
}
}
for (String value : toRemove) {
observedValueAlignments.remove(value);
}
}
getObservedAttrValueSequences().add(observedAttrValueSequence);
}
}
public void writeLists() {
String file1 = "cache/getPredicates()";
String file2 = "cache/attributes";
String file3 = "cache/attributeValuePairs";
String file4 = "cache/getValueAlignments()";
String file5 = "cache/getDatasetInstances";
String file6 = "cache/maxLengths";
String file7 = "cache/getValidationDatasetInstances";
String file8 = "cache/getTrainingDatasetInstances";
FileOutputStream fout1 = null;
ObjectOutputStream oos1 = null;
FileOutputStream fout2 = null;
ObjectOutputStream oos2 = null;
FileOutputStream fout3 = null;
ObjectOutputStream oos3 = null;
FileOutputStream fout4 = null;
ObjectOutputStream oos4 = null;
FileOutputStream fout5 = null;
ObjectOutputStream oos5 = null;
FileOutputStream fout6 = null;
ObjectOutputStream oos6 = null;
FileOutputStream fout7 = null;
ObjectOutputStream oos7 = null;
FileOutputStream fout8 = null;
ObjectOutputStream oos8 = null;
try {
System.out.println("Write lists...");
fout1 = new FileOutputStream(file1);
oos1 = new ObjectOutputStream(fout1);
oos1.writeObject(getPredicates());
///////////////////
fout2 = new FileOutputStream(file2);
oos2 = new ObjectOutputStream(fout2);
oos2.writeObject(getAttributes());
///////////////////
fout3 = new FileOutputStream(file3);
oos3 = new ObjectOutputStream(fout3);
oos3.writeObject(getAttributeValuePairs());
///////////////////
fout4 = new FileOutputStream(file4);
oos4 = new ObjectOutputStream(fout4);
oos4.writeObject(getValueAlignments());
///////////////////
fout5 = new FileOutputStream(file5);
oos5 = new ObjectOutputStream(fout5);
oos5.writeObject(getDatasetInstances());
///////////////////
fout6 = new FileOutputStream(file6);
oos6 = new ObjectOutputStream(fout6);
//ArrayList<Integer> lengths = new ArrayList<Integer>();
//lengths.add(getMaxContentSequenceLength());
//lengths.add(getMaxWordSequenceLength());
oos6.writeObject(getMaxWordSequenceLength());
fout7 = new FileOutputStream(file7);
oos7 = new ObjectOutputStream(fout7);
oos7.writeObject(getValidationData());
fout8 = new FileOutputStream(file8);
oos8 = new ObjectOutputStream(fout8);
oos8.writeObject(getTrainingData());
} catch (IOException ex) {
} finally {
try {
fout1.close();
fout2.close();
fout3.close();
fout4.close();
fout5.close();
fout6.close();
} catch (IOException ex) {
}
try {
oos1.close();
oos2.close();
oos3.close();
oos4.close();
oos5.close();
oos6.close();
} catch (IOException ex) {
}
}
}
@SuppressWarnings("unchecked")
public boolean loadLists() {
String file1 = "cache/getPredicates()";
String file2 = "cache/attributes";
String file3 = "cache/attributeValuePairs";
String file4 = "cache/getValueAlignments()";
String file5 = "cache/getDatasetInstances";
String file6 = "cache/maxLengths";
String file7 = "cache/getValidationDatasetInstances";
String file8 = "cache/getTrainingDatasetInstances";
FileInputStream fin1 = null;
ObjectInputStream ois1 = null;
FileInputStream fin2 = null;
ObjectInputStream ois2 = null;
FileInputStream fin3 = null;
ObjectInputStream ois3 = null;
FileInputStream fin4 = null;
ObjectInputStream ois4 = null;
FileInputStream fin5 = null;
ObjectInputStream ois5 = null;
FileInputStream fin6 = null;
ObjectInputStream ois6 = null;
FileInputStream fin7 = null;
ObjectInputStream ois7 = null;
FileInputStream fin8 = null;
ObjectInputStream ois8 = null;
if ((new File(file1)).exists()
&& (new File(file2)).exists()
&& (new File(file3)).exists()
&& (new File(file4)).exists()
&& (new File(file5)).exists()
&& (new File(file6)).exists()
&& (new File(file7)).exists()
&& (new File(file8)).exists()) {
try {
System.out.println("Load lists...");
fin1 = new FileInputStream(file1);
ois1 = new ObjectInputStream(fin1);
Object o1 = ois1.readObject();
if (getPredicates() == null) {
if (o1 instanceof ArrayList) {
setPredicates(new ArrayList<String>((Collection<? extends String>) o1));
}
} else if (o1 instanceof ArrayList) {
getPredicates().addAll((Collection<? extends String>) o1);
}
///////////////////
fin2 = new FileInputStream(file2);
ois2 = new ObjectInputStream(fin2);
Object o2 = ois2.readObject();
if (getAttributes() == null) {
if (o2 instanceof HashMap) {
setAttributes(new HashMap<String, HashSet<String>>((Map<? extends String, ? extends HashSet<String>>) o2));
}
} else if (o2 instanceof HashMap) {
getAttributes().putAll((Map<? extends String, ? extends HashSet<String>>) o2);
}
///////////////////
fin3 = new FileInputStream(file3);
ois3 = new ObjectInputStream(fin3);
Object o3 = ois3.readObject();
if (getAttributeValuePairs() == null) {
if (o3 instanceof HashMap) {
setAttributeValuePairs(new HashMap<String, HashSet<String>>((Map<? extends String, ? extends HashSet<String>>) o3));
}
} else if (o3 instanceof HashMap) {
getAttributeValuePairs().putAll((Map<? extends String, ? extends HashSet<String>>) o3);
}
///////////////////
fin4 = new FileInputStream(file4);
ois4 = new ObjectInputStream(fin4);
Object o4 = ois4.readObject();
if (getValueAlignments() == null) {
if (o4 instanceof HashMap) {
setValueAlignments(new HashMap<String, HashMap<ArrayList<String>, Double>>((Map<? extends String, ? extends HashMap<ArrayList<String>, Double>>) o4));
}
} else if (o4 instanceof HashMap) {
getValueAlignments().putAll((Map<? extends String, ? extends HashMap<ArrayList<String>, Double>>) o4);
}
///////////////////
fin5 = new FileInputStream(file5);
ois5 = new ObjectInputStream(fin5);
Object o5 = ois5.readObject();
if (getDatasetInstances() == null) {
if (o5 instanceof HashMap) {
setDatasetInstances(new HashMap<String, ArrayList<DatasetInstance>>((Map<? extends String, ? extends ArrayList<DatasetInstance>>) o5));
}
} else if (o5 instanceof HashMap) {
getDatasetInstances().putAll((Map<? extends String, ? extends ArrayList<DatasetInstance>>) o5);
}
///////////////////
fin6 = new FileInputStream(file6);
ois6 = new ObjectInputStream(fin6);
Object o6 = ois6.readObject();
//ArrayList<Integer> lengths = new ArrayList<Integer>((Collection<? extends Integer>) o6);
//setMaxContentSequenceLength(lengths.get(0));
setMaxWordSequenceLength((Integer) o6);
///////////////////
fin7 = new FileInputStream(file7);
ois7 = new ObjectInputStream(fin7);
Object o7 = ois7.readObject();
if (getValidationData() == null) {
if (o7 instanceof ArrayList) {
setValidationData(new ArrayList<DatasetInstance>((Collection<? extends DatasetInstance>) o7));
}
} else if (o7 instanceof ArrayList) {
getValidationData().addAll(new ArrayList<DatasetInstance>((Collection<? extends DatasetInstance>) o7));
}
fin8 = new FileInputStream(file8);
ois8 = new ObjectInputStream(fin8);
Object o8 = ois8.readObject();
if (getTrainingData() == null) {
if (o8 instanceof ArrayList) {
setTrainingData(new ArrayList<DatasetInstance>((Collection<? extends DatasetInstance>) o8));
}
} else if (o8 instanceof ArrayList) {
getTrainingData().addAll(new ArrayList<DatasetInstance>((Collection<? extends DatasetInstance>) o8));
}
System.out.println("done!");
} catch (ClassNotFoundException | IOException ex) {
} finally {
try {
fin1.close();
fin2.close();
fin3.close();
fin4.close();
fin5.close();
fin6.close();
fin7.close();
fin8.close();
} catch (IOException ex) {
}
try {
ois1.close();
ois2.close();
ois3.close();
ois4.close();
ois5.close();
ois6.close();
ois7.close();
ois8.close();
} catch (IOException ex) {
}
}
return true;
} else {
return false;
}
}
/**
*
* @return
*/
public boolean loadAvailableActions() {
if (!isCache()) {
return false;
}
String file1 = "cache/availableContentActions_SF_" + getDataset();
String file2 = "cache/availableWordActions_SF_" + getDataset();
FileInputStream fin1 = null;
ObjectInputStream ois1 = null;
FileInputStream fin2 = null;
ObjectInputStream ois2 = null;
if ((new File(file1)).exists()
&& (new File(file2)).exists()) {
try {
System.out.print("Load available actions...");
fin1 = new FileInputStream(file1);
ois1 = new ObjectInputStream(fin1);
Object o1 = ois1.readObject();
if (getAvailableContentActions() == null) {
if (o1 instanceof HashMap) {
setAvailableContentActions((HashMap<String, HashSet<String>>) o1);
}
} else if (o1 instanceof HashMap) {
getAvailableContentActions().putAll((HashMap<String, HashSet<String>>) o1);
}
fin2 = new FileInputStream(file2);
ois2 = new ObjectInputStream(fin2);
Object o2 = ois2.readObject();
if (getAvailableWordActions() == null) {
if (o2 instanceof HashMap) {
setAvailableWordActions((HashMap<String, HashMap<String, HashSet<Action>>>) o2);
}
} else if (o2 instanceof HashMap) {
getAvailableWordActions().putAll((HashMap<String, HashMap<String, HashSet<Action>>>) o2);
}
System.out.println("done!");
} catch (ClassNotFoundException | IOException ex) {
} finally {
try {
fin1.close();
fin2.close();
} catch (IOException ex) {
}
try {
ois1.close();
ois2.close();
} catch (IOException ex) {
}
}
} else {
return false;
}
return true;
}
/**
*
*/
public void writeAvailableActions() {
String file1 = "cache/availableContentActions_SF_" + getDataset();
String file2 = "cache/availableWordActions_SF_" + getDataset();
FileOutputStream fout1 = null;
ObjectOutputStream oos1 = null;