Class: DegreeOfRelevance

Inherits:
Object
  • Object
show all
Defined in:
lib/automated_metareview/degree_of_relevance.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Instance Attribute Details

#review ⇒ Object

Returns the value of attribute review.



7
8
9
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 7

def review
  @review
end

#vertex_match ⇒ Object

creating accessors for the instance variables



6
7
8
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 6

def vertex_match
  @vertex_match
end

Instance Method Details

#compare_edges_diff_types(rev, subm, num_rev_edg, num_sub_edg) ⇒ Object

DIFFERENT TYPE COMPARISON!!

* Compares the edges from across the two graphs to identify matches and quantify various metrics
* compare SUBJECT-VERB edges with VERB-OBJECT matches and vice-versa
* SUBJECT-VERB, VERB-SUBJECT, OBJECT-VERB, VERB-OBJECT comparisons are done!



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 306

def compare_edges_diff_types(rev, subm, num_rev_edg, num_sub_edg)
  # puts("*****Inside compareEdgesDiffTypes :: numRevEdg :: #{num_rev_edg} numSubEdg:: #{num_sub_edg}")   
  best_SV_VS_match = Array.new(num_rev_edg){Array.new}
  cum_edge_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  wnet = WordnetBasedSimilarity.new  
  for i in (0..num_rev_edg - 1)
    if(!rev[i].nil? and rev[i].in_vertex.node_id != -1 and rev[i].out_vertex.node_id != -1)
      #skipping edges with frequent words for vertices
      if(wnet.is_frequent_word(rev[i].in_vertex.name) and wnet.is_frequent_word(rev[i].out_vertex.name))
        next
      end
      #identifying best match for edges
      for j in (0..num_sub_edg - 1) 
        if(!subm[j].nil? and subm[j].in_vertex.node_id != -1 and subm[j].out_vertex.node_id != -1)
          #checking if the subm token is a frequent word
          if(wnet.is_frequent_word(subm[j].in_vertex.name) and wnet.is_frequent_word(subm[j].out_vertex.name))
            next
          end 
          #for S-V with S-V or V-O with V-O
          if(rev[i].in_vertex.type == subm[j].in_vertex.type and rev[i].out_vertex.type == subm[j].out_vertex.type)
            #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
            sum = 0.0
            cou = 0
            if(!@vertex_match[rev[i].in_vertex.node_id][subm[j].out_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j].out_vertex.node_id]
              cou +=1
            end
            if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].in_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].in_vertex.node_id]
              cou +=1
            end  
            if(cou > 0)
              best_SV_VS_match[i][j] = sum.to_f/cou.to_f
            else
              best_SV_VS_match[i][j] = 0.0
            end
            #-- Vertex and SRL
            best_SV_VS_match[i][j] = best_SV_VS_match[i][j]/ compare_labels(rev[i], subm[j])
            flag = 1
            if(best_SV_VS_match[i][j] > max)
              max = best_SV_VS_match[i][j]
            end
          #for S-V with V-O or V-O with S-V
          elsif(rev[i].in_vertex.type == subm[j].out_vertex.type and rev[i].out_vertex.type == subm[j].in_vertex.type)
            #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
            sum = 0.0
            cou = 0
            if(!@vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id]
              cou +=1
            end
            if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id]
              cou +=1
            end  
            if(cou > 0)
              best_SV_VS_match[i][j] = sum.to_f/cou.to_f
            else
              best_SV_VS_match[i][j] =0.0
            end
            flag = 1
            if(best_SV_VS_match[i][j] > max)
              max = best_SV_VS_match[i][j]
            end
          end
        end #end of the if condition
      end #end of the for loop for submission edges
        
      if(flag != 0) #if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
        # puts("**** Best match for:: #{rev[i].in_vertex.name} - #{rev[i].out_vertex.name} -- #{max}")
        cum_edge_match = cum_edge_match + max
        count+=1
        max = 0.0 #re-initialize
        flag = 0
      end
    end #end of if condition
  end #end of for loop for review edges
    
  avg_match = 0.0
  if(count > 0)
    avg_match = cum_edge_match.to_f/ count.to_f
  end
  return avg_match
end

#compare_edges_non_syntax_diff(rev, subm, num_rev_edg, num_sub_edg) ⇒ Object

  • SAME TYPE COMPARISON!!

    * Compares the edges from across the two graphs to identify matches and quantify various metrics
* compare SUBJECT-VERB edges with SUBJECT-VERB matches
* where SUBJECT-SUBJECT and VERB-VERB or VERB-VERB and OBJECT-OBJECT comparisons are done



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 150

def compare_edges_non_syntax_diff(rev, subm, num_rev_edg, num_sub_edg)
  # puts("*****Inside compareEdgesnNonSyntaxDiff numRevEdg:: #{num_rev_edg} numSubEdg:: #{num_sub_edg}")   
  best_SV_SV_match = Array.new(num_rev_edg){Array.new}
  cum_edge_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  
  wnet = WordnetBasedSimilarity.new
  for i in (0..num_rev_edg - 1)
    if(!rev[i].nil? and rev[i].in_vertex.node_id != -1 and rev[i].out_vertex.node_id != -1)
    #skipping edges with frequent words for vertices
    if(wnet.is_frequent_word(rev[i].in_vertex.name) and wnet.is_frequent_word(rev[i].out_vertex.name))
      next
    end
    
    #looking for best matches
    for j in (0..num_sub_edg - 1)
      #comparing in-vertex with out-vertex to make sure they are of the same type
      if(!subm[j].nil? && subm[j].in_vertex.node_id != -1 && subm[j].out_vertex.node_id != -1)
        
        #checking if the subm token is a frequent word
        if(wnet.is_frequent_word(subm[j].in_vertex.name) and wnet.is_frequent_word(subm[j].out_vertex.name))
          next
        end     
              
        #carrying out the normal comparison
        if(rev[i].in_vertex.type == subm[j].in_vertex.type && rev[i].out_vertex.type == subm[j].out_vertex.type)
          if(!rev[i].label.nil?)
            if(!subm[j].label.nil?)
              #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
              sum = 0.0
              cou = 0
              if(!@vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id].nil?)
                sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id]
                cou +=1
              end
              if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id].nil?)
                sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id]
                cou +=1
              end  
              #--Only vertex matches
              if(cou > 0)
                best_SV_SV_match[i][j] = sum.to_f/cou.to_f
              else
                best_SV_SV_match[i][j] = 0.0
              end
              #--Vertex and SRL - Dividing it by the label's match value
              best_SV_SV_match[i][j] = best_SV_SV_match[i][j]/ compare_labels(rev[i], subm[j])
              flag = 1
              if(best_SV_SV_match[i][j] > max)
                max = best_SV_SV_match[i][j]
              end
            end
          end
        end
      end
    end #end of for loop for the submission edges
    
    #cumulating the review edges' matches in order to get its average value
    if(flag != 0) #if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
      # puts("**** Best match for:: #{rev[i].in_vertex.name} - #{rev[i].out_vertex.name} -- #{max}")
      cum_edge_match = cum_edge_match + max
      count+=1
      max = 0.0#re-initialize
      flag = 0
      end
    end
  end #end of 'for' loop for the review's edges
  
  #getting the average for all the review edges' matches with the submission's edges
  avg_match = 0.0
  if(count > 0)
    avg_match = cum_edge_match/ count
  end
  return avg_match
end

#compare_edges_syntax_diff(rev, subm, num_rev_edg, num_sub_edg) ⇒ Object

  • SAME TYPE COMPARISON!!

    * Compares the edges from across the two graphs to identify matches and quantify various metrics
* compare SUBJECT-VERB edges with VERB-OBJECT matches and vice-versa
* where SUBJECT-OBJECT and VERB_VERB comparisons are done - same type comparisons!!



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 235

def compare_edges_syntax_diff(rev, subm, num_rev_edg, num_sub_edg)
  # puts("*****Inside compareEdgesSyntaxDiff :: numRevEdg :: #{num_rev_edg} numSubEdg:: #{num_sub_edg}")    
  best_SV_VS_match = Array.new(num_rev_edg){Array.new}
  cum_edge_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  wnet = WordnetBasedSimilarity.new  
  for i in (0..num_rev_edg - 1)
    if(!rev[i].nil? and rev[i].in_vertex.node_id != -1 and rev[i].out_vertex.node_id != -1)
      #skipping frequent word
      if(wnet.is_frequent_word(rev[i].in_vertex.name) and wnet.is_frequent_word(rev[i].out_vertex.name))
        next
      end
      for j in (0..num_sub_edg - 1)
        if(!subm[j].nil? and subm[j].in_vertex.node_id != -1 and subm[j].out_vertex.node_id != -1)
          #checking if the subm token is a frequent word
          if(wnet.is_frequent_word(subm[j].in_vertex.name) and wnet.is_frequent_word(subm[j].out_vertex.name))
            next
          end 
          if(rev[i].in_vertex.type == subm[j].out_vertex.type and rev[i].out_vertex.type == subm[j].in_vertex.type)
            #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
            sum = 0.0
            cou = 0
            if(!@vertex_match[rev[i].in_vertex.node_id][subm[j].out_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j].out_vertex.node_id]
              cou +=1
            end
            if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].in_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].in_vertex.node_id]
              cou +=1
            end              
            
            if(cou > 0)
              best_SV_VS_match[i][j] = sum.to_f/cou.to_f
            else
              best_SV_VS_match[i][j] = 0.0
            end
            
            flag = 1
            if(best_SV_VS_match[i][j] > max)
              max = best_SV_VS_match[i][j]
            end
          end
        end #end of the if condition
      end #end of the for loop for the submission edges     
          
      if(flag != 0)#if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
        # puts("**** Best match for:: #{rev[i].in_vertex.name} - #{rev[i].out_vertex.name}-- #{max}")
        cum_edge_match = cum_edge_match + max
        count+=1
        max = 0.0 #re-initialize
        flag = 0
      end
        
    end #end of the if condition
  end #end of the for loop for the review
   
  avg_match = 0.0
  if(count > 0)
    avg_match = cum_edge_match.to_f/count.to_f
  end
  return avg_match
end

#compare_labels(edge1, edge2) ⇒ Object

SR Labels and vertex matches are given equal importance

* Problem is even if the vertices didn't match, the SRL labels would cause them to have a high similarity.
* Consider "boy - said" and "chocolate - melted" - these edges have NOMATCH for vertices, but both edges have the same label "SBJ" and would get an EXACT match, 
* resulting in an avg of 3! This cant be right!
* We therefore use the labels to only decrease the match value found from vertices, i.e., if the labels were different.
* Match value will be left as is, if the labels were the same.



547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 547

def compare_labels(edge1, edge2)
  result = EQUAL
  if(!edge1.label.nil? and !edge2.label .nil?)
    if(edge1.label.downcase == edge2.label.downcase)
      result = EQUAL #divide by 1
    else
      result = DISTINCT #divide by 2
    end
  elsif((!edge1.label.nil? and !edge2.label.nil?) or (edge1.label.nil? and !edge2.label.nil? )) #if only one of the labels was null
      result = DISTINCT
  elsif(edge1.label.nil? and edge2.label.nil?) #if both labels were null!
      result = EQUAL
  end  
  return result
end

#compare_SVO_diff_syntax(rev, subm, num_rev_edg, num_sub_edg) ⇒ Object

——————————————#——————————————



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 472

def compare_SVO_diff_syntax(rev, subm, num_rev_edg, num_sub_edg)
  # puts("***********Inside compare SVO edges with syntax difference numRevEdg:: #{num_rev_edg} numSubEdg:: #{num_sub_edg}")
  best_SVO_OVS_edges_match = Array.new(num_rev_edg){ Array.new}
  cum_double_edge_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  wnet = WordnetBasedSimilarity.new  
  for i in (0..num_rev_edg - 1) 
    if(!rev[i].nil? and !rev[i+1].nil? and rev[i].in_vertex.node_id != -1 and rev[i].out_vertex.node_id != -1 and 
      rev[i+1].out_vertex.node_id != -1 and rev[i].out_vertex == rev[i+1].in_vertex)
      #skipping edges with frequent words for vertices
      if(wnet.is_frequent_word(rev[i].in_vertex.name) and wnet.is_frequent_word(rev[i].out_vertex.name) and wnet.is_frequent_word(rev[i+1].out_vertex.name))
        next
      end
        
      for j in (0..num_sub_edg - 1)
        if(!subm[j].nil? and !subm[j+1].nil? and subm[j].in_vertex.node_id != -1 and subm[j].out_vertex.node_id != -1 and subm[j+1].out_vertex.node_id != -1 and subm[j].out_vertex == subm[j+1].in_vertex)
          #making sure the types are the same during comparison
          if(rev[i].in_vertex.type == subm[j+1].out_vertex.type and rev[i].out_vertex.type == subm[j].out_vertex.type and 
            rev[i+1].out_vertex.type == subm[j].in_vertex.type)
            #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
            sum = 0.0
            cou = 0
            if(!@vertex_match[rev[i].in_vertex.node_id][subm[j+1].out_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j+1].out_vertex.node_id]
              cou +=1
            end
            if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id]
              cou +=1
            end
            if(!@vertex_match[rev[i+1].out_vertex.node_id][subm[j].in_vertex.node_id].nil?)
              sum = sum + @vertex_match[rev[i+1].out_vertex.node_id][subm[j].in_vertex.node_id]
              cou +=1
            end
            #comparing s-v-o (from review) with o-v-s (from submission)
            if(cou > 0)
              best_SVO_OVS_edges_match[i][j] = sum.to_f/cou.to_f
            else
              best_SVO_OVS_edges_match[i][j] = 0.0
            end
            flag = 1
            if(best_SVO_OVS_edges_match[i][j] > max)
              max = best_SVO_OVS_edges_match[i][j]
            end
          end  
        end #end of 'if' condition
      end #end of 'for' loop for 'j'
      if(flag != 0)#if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
        # puts("**** Best match for:: #{rev[i].in_vertex.name} - #{rev[i].out_vertex.name} - #{rev[i+1].out_vertex.name}-- #{max}")
        cum_double_edge_match = cum_double_edge_match + max
        count+=1
        max = 0.0 #re-initialize
        flag = 0
      end
      
    end #end of if condition
  end #end of for loop for 'i'
    
  avg_match = 0.0
  if(count > 0)
    avg_match = cum_double_edge_match.to_f / count.to_f
  end
  return avg_match
end

#compare_SVO_edges(rev, subm, num_rev_edg, num_sub_edg) ⇒ Object

——————————————#——————————————



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 395

def compare_SVO_edges(rev, subm, num_rev_edg, num_sub_edg)
  # puts("***********Inside compare SVO edges numRevEdg:: #{num_rev_edg} numSubEdg:: #{num_sub_edg}")
  best_SVO_SVO_edges_match = Array.new(num_rev_edg){Array.new}
  cum_double_edge_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  wnet = WordnetBasedSimilarity.new  
  for i in (0..num_rev_edg - 1)
    if(!rev[i].nil? and !rev[i+1].nil? and rev[i].in_vertex.node_id != -1 and rev[i].out_vertex.node_id != -1 and 
      rev[i+1].out_vertex.node_id != -1  and rev[i].out_vertex == rev[i+1].in_vertex)
      #skipping edges with frequent words for vertices
      if(wnet.is_frequent_word(rev[i].in_vertex.name) and wnet.is_frequent_word(rev[i].out_vertex.name) and wnet.is_frequent_word(rev[i+1].out_vertex.name))
        next
      end
        #best match
        for j in (0..num_sub_edg - 1)
          if(!subm[j].nil? and !subm[j+1].nil? and subm[j].in_vertex.node_id != -1 and subm[j].out_vertex.node_id != -1 and 
            subm[j+1].out_vertex.node_id != -1 and subm[j].out_vertex == subm[j+1].in_vertex)
            #checking if the subm token is a frequent word
            if(wnet.is_frequent_word(subm[j].in_vertex.name) and wnet.is_frequent_word(subm[j].out_vertex.name))
              next
            end 
            #making sure the types are the same during comparison
            if(rev[i].in_vertex.type == subm[j].in_vertex.type and rev[i].out_vertex.type == subm[j].out_vertex.type and 
              rev[i+1].out_vertex.type == subm[j+1].out_vertex.type)
              #taking each match separately because one or more of the terms may be a frequent word, for which no @vertex_match exists!
              sum = 0.0
              cou = 0
              if(!@vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id].nil?)
                sum = sum + @vertex_match[rev[i].in_vertex.node_id][subm[j].in_vertex.node_id]
                cou +=1
              end
              if(!@vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id].nil?)
                sum = sum + @vertex_match[rev[i].out_vertex.node_id][subm[j].out_vertex.node_id]
                cou +=1
              end
              if(!@vertex_match[rev[i+1].out_vertex.node_id][subm[j+1].out_vertex.node_id].nil?)
                sum = sum + @vertex_match[rev[i+1].out_vertex.node_id][subm[j+1].out_vertex.node_id]
                cou +=1
              end
              #-- Only Vertex match
              if(cou > 0)
                best_SVO_SVO_edges_match[i][j] = sum.to_f/cou.to_f
              else
                best_SVO_SVO_edges_match[i][j] = 0.0
              end
              #-- Vertex and SRL
              best_SVO_SVO_edges_match[i][j] = best_SVO_SVO_edges_match[i][j].to_f/ compare_labels(rev[i], subm[j]).to_f
              best_SVO_SVO_edges_match[i][j] = best_SVO_SVO_edges_match[i][j].to_f/ compare_labels(rev[i+1], subm[j+1]).to_f
              #-- Only SRL
              if(best_SVO_SVO_edges_match[i][j] > max)
                max = best_SVO_SVO_edges_match[i][j]
              end
              flag = 1
           end
          end #end of 'if' condition
        end #end of 'for' loop for 'j'
        
        if(flag != 0) #if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
          # puts("**** Best match for:: #{rev[i].in_vertex.name} - #{rev[i].out_vertex.name} - #{rev[i+1].out_vertex.name} -- #{max}")
          cum_double_edge_match = cum_double_edge_match + max
          count+=1
          max = 0.0 #re-initialize
          flag = 0
        end
      end #end of 'if' condition
    end #end of 'for' loop for 'i'
    
  avg_match = 0.0
  if(count > 0)
    avg_match = cum_double_edge_match.to_f/ count.to_f
  end
  return avg_match
end

#compare_vertices(pos_tagger, rev, subm, num_rev_vert, num_sub_vert, speller) ⇒ Object

  • every vertex is compared with every other vertex

    * Compares the vertices from across the two graphs to identify matches and quantify various metrics
* v1- vertices of the submission/past review and v2 - vertices from new review



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 85

def compare_vertices(pos_tagger, rev, subm, num_rev_vert, num_sub_vert, speller)
  # puts("****Inside compare_vertices:: rev.length:: #{num_rev_vert} subm.length:: #{num_sub_vert}")
  #for double dimensional arrays, one of the dimensions should be initialized
  @vertex_match = Array.new(num_rev_vert){Array.new}
  wnet = WordnetBasedSimilarity.new
  cum_vertex_match = 0.0
  count = 0
  max = 0.0
  flag = 0
  
  for i in (0..num_rev_vert - 1)
    if(!rev.nil? and !rev[i].nil?)
      rev[i].node_id = i
      # puts("%%%%%%%%%%% Token #{rev[i].name} ::: POS tags:: rev[i].pos_tag:: #{rev[i].pos_tag} :: rev[i].node_id #{rev[i].node_id}")
      #skipping frequent words from vertex comparison
      if(wnet.is_frequent_word(rev[i].name))
        next #ruby equivalent for continue 
      end
      #looking for the best match
      #j tracks every element in the set of all vertices, some of which are null
      for j in (0..num_sub_vert - 1)
        if(!subm[j].nil?)
          if(subm[j].node_id == -1)
            subm[j].node_id = j
          end
          # puts("%%%%%%%%%%% Token #{subm[j].name} ::: POS tags:: subm[j].pos_tag:: #{subm[j].pos_tag} subm[j].node_id #{subm[j].node_id}")
          if(wnet.is_frequent_word(subm[j].name))
            next #ruby equivalent for continue 
          end
          #comparing only if one of the two vertices is a noun
          if(rev[i].pos_tag.include?("NN") and subm[j].pos_tag.include?("NN"))
            @vertex_match[i][j] = wnet.compare_strings(rev[i], subm[j], speller)    
            #only if the "if" condition is satisfied, since there could be null objects in between and you dont want unnecess. increments
            flag = 1
            if(@vertex_match[i][j] > max)
              max = @vertex_match[i][j]
            end
          end
        end
      end #end of for loop for the submission vertices
      
      if(flag != 0)#if the review edge had any submission edges with which it was matched, since not all S-V edges might have corresponding V-O edges to match with
        # puts("**** Best match for:: #{rev[i].name}-- #{max}")
        cum_vertex_match = cum_vertex_match + max
        count+=1
        max = 0.0 #re-initialize
        flag = 0
      end
    end #end of if condition
  end #end of for loop

  avg_match = 0.0
  if(count > 0)
    avg_match = cum_vertex_match/ count
  end
  return avg_match  
end

#get_relevance(reviews, submissions, num_reviews, pos_tagger, core_NLP_tagger, speller) ⇒ Object

Identifies relevance between a review and a submission



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
 
# File 'lib/automated_metareview/degree_of_relevance.rb', line 11

def get_relevance(reviews, submissions, num_reviews, pos_tagger, core_NLP_tagger, speller) #double dimensional arrays that contain the submissions and the reviews respectively
  review_vertices = nil
  review_edges = nil
  subm_vertices = nil
  subm_edges = nil
  num_rev_vert = 0
  num_rev_edg = 0 
  num_sub_vert = 0 
  numSubEdg = 0
  vert_match = 0.0
  edge_without_syn = 0.0
  edge_with_syn = 0.0
  edge_diff_type = 0.0
  double_edge = 0.0
  double_edge_with_syn = 0.0
  
    #since Reviews and Submissions "should" contain the same number of records review - submission pairs
    g = GraphGenerator.new
    #generating review's graph
    g.generate_graph(reviews, pos_tagger, core_NLP_tagger, true, false)
    review_vertices = g.vertices
    review_edges = g.edges
    num_rev_vert = g.num_vertices
    num_rev_edg = g.num_edges
    
    #assigning graph as a review graph to use in content classification
    @review = g.clone
      
    #generating the submission's graph
    g.generate_graph(submissions, pos_tagger, core_NLP_tagger, true, false)
    subm_vertices = g.vertices
    subm_edges = g.edges
    num_sub_vert = g.num_vertices
    num_sub_edg = g.num_edges
      
    vert_match = compare_vertices(pos_tagger, review_vertices, subm_vertices, num_rev_vert, num_sub_vert, speller)
    if(num_rev_edg > 0 and num_sub_edg > 0)
      edge_without_syn = compare_edges_non_syntax_diff(review_edges, subm_edges, num_rev_edg, num_sub_edg)
      edge_with_syn = compare_edges_syntax_diff(review_edges, subm_edges, num_rev_edg, num_sub_edg)
      edge_diff_type = compare_edges_diff_types(review_edges, subm_edges, num_rev_edg, num_sub_edg)
      edge_match = (edge_without_syn.to_f + edge_with_syn.to_f )/2.to_f #+ edge_diff_type.to_f
      double_edge = compare_SVO_edges(review_edges, subm_edges, num_rev_edg, num_sub_edg)
      double_edge_with_syn = compare_SVO_diff_syntax(review_edges, subm_edges, num_rev_edg, num_sub_edg)
      double_edge_match = (double_edge.to_f + double_edge_with_syn.to_f)/2.to_f
    else
      edge_match = 0
      double_edge_match = 0
    end
      
    #differently weighted cases
    #tweak this!!
    alpha = 0.55
    beta = 0.35
    gamma = 0.1 #alpha > beta > gamma
    relevance = (alpha.to_f * vert_match.to_f) + (beta * edge_match.to_f) + (gamma * double_edge_match.to_f) #case1's value will be in the range [0-6] (our semantic values) 
    scaled_relevance = relevance.to_f/6.to_f #scaled from [0-6] in the range [0-1]
    
    #printing values
    # puts("vertexMatch is [0-6]:: #{vert_match}")
    # puts("edgeWithoutSyn Match is [0-6]:: #{edge_without_syn}")
    # puts("edgeWithSyn Match is [0-6]:: #{edge_with_syn}")
    # puts("edgeDiffType Match is [0-6]:: #{edge_diff_type}")
    # puts("doubleEdge Match is [0-6]:: #{double_edge}")
    # puts("doubleEdge with syntax Match is [0-6]:: #{double_edge_with_syn}")
    # puts("relevance [0-6]:: #{relevance}")
    # puts("scaled relevance on [0-1]:: #{scaled_relevance}")
    # puts("*************************************************")
    return scaled_relevance
end