ikarasz commited on
Commit
0d0084a
·
1 Parent(s): 3f3a914

fix list of matched words

Browse files
Files changed (1) hide show
  1. handler.py +9 -8
handler.py CHANGED
@@ -376,13 +376,14 @@ class FocusingQuestionModel:
376
  def load_math_terms():
377
  math_regexes = []
378
  math_terms_dict = {}
 
 
 
 
379
  for term in MATH_WORDS:
380
- if term in MATH_PREFIXES:
381
- math_terms_dict[rf"\b{term}(s|es|d|ed)?\b"] = term
382
- math_regexes.append(rf"\b{term}(s|es|d|ed)?\b")
383
- else:
384
- math_regexes.append(rf"\b{term}\b")
385
  math_terms_dict[rf"\b{term}\b"] = term
 
386
  return math_regexes, math_terms_dict
387
 
388
  def run_math_density(transcript):
@@ -409,9 +410,9 @@ def run_math_density(transcript):
409
  if math_terms_dict[regex] not in student_math_word_cloud:
410
  student_math_word_cloud[math_terms_dict[regex]] = 0
411
  student_math_word_cloud[math_terms_dict[regex]] += len(matches)
412
- match_list.append(math_terms_dict[regex])
413
- # Update matched positions
414
- matched_positions.update((match.start(), match.end()) for match in matches)
415
  num_matches += len(matches)
416
  # print("match group list: ", [match.group(0) for match in matches])
417
  utt.num_math_terms = num_matches
 
376
  def load_math_terms():
377
  math_regexes = []
378
  math_terms_dict = {}
379
+ for term in MATH_PREFIXES:
380
+ math_terms_dict[rf"\b{term}(s|es|d|ed)?\b"] = term
381
+ math_regexes.append(rf"\b{term}(s|es|d|ed)?\b")
382
+
383
  for term in MATH_WORDS:
384
+ if not term in MATH_PREFIXES:
 
 
 
 
385
  math_terms_dict[rf"\b{term}\b"] = term
386
+ math_regexes.append(rf"\b{term}\b")
387
  return math_regexes, math_terms_dict
388
 
389
  def run_math_density(transcript):
 
410
  if math_terms_dict[regex] not in student_math_word_cloud:
411
  student_math_word_cloud[math_terms_dict[regex]] = 0
412
  student_math_word_cloud[math_terms_dict[regex]] += len(matches)
413
+ for match in matches:
414
+ match_list.append(match.group())
415
+ matched_positions.add((match.start(), match.end()))
416
  num_matches += len(matches)
417
  # print("match group list: ", [match.group(0) for match in matches])
418
  utt.num_math_terms = num_matches