def get_synonym_complexities(self):
if len(self.synonyms) > 1:
synonym_scores = pd.DataFrame()
synonym_scores['synonyms'] = self.synonyms
synonym_scores['sem_sim'] = hf.get_elmo_score(
self.synonyms, self.token_sent, self.index)
synonym_scores[
'complexity'] = complex_word.get_synonym_complexities(
self.synonyms, self.token_sent, self.index)
synonym_scores['grammaticality'] = hf.get_gram_score(
self.synonyms, self.token_sent, self.pos_sent, self.index)
#filtering process, return top word?
synonym_scores = synonym_scores[synonym_scores['sem_sim'] < 0.15]
synonym_scores = synonym_scores.sort_values(by=['complexity'])
return list(
zip(synonym_scores['synonyms'].values,
synonym_scores['complexity'].values))
else:
return None
def get_synonym_dataframe(self):
if len(self.synonyms) > 1:
synonym_scores = pd.DataFrame()
synonym_scores['synonyms'] = self.synonyms
synonym_scores['sem_sim'] = hf.get_elmo_score(
self.synonyms, self.token_sent, self.index)
synonym_scores[
'complexity'] = complex_word.get_synonym_complexities(
self.synonyms, self.token_sent, self.index)
synonym_scores['grammaticality'] = hf.get_gram_score(
self.synonyms, self.token_sent, self.pos_sent, self.index)
synonym_scores = synonym_scores[synonym_scores['sem_sim'] < 0.3]
#Can filter to only replace with words of lower threshold complexity
#synonym_scores = synonym_scores[synonym_scores['complexity']<0.6]
synonym_scores = synonym_scores[synonym_scores['grammaticality'] ==
1]
synonym_scores['combo'] = synonym_scores[
'sem_sim'] + synonym_scores['complexity']
synonym_scores = synonym_scores.sort_values(by=['combo'])
return synonym_scores
else:
return None
def get_ranked_synonyms(self):
if len(self.synonyms) > 1:
synonym_scores = pd.DataFrame()
synonym_scores['synonyms'] = self.synonyms
synonym_scores['sem_sim'] = hf.get_elmo_score(
self.synonyms, self.token_sent, self.index)
synonym_scores[
'complexity'] = complex_word.get_synonym_complexities(
self.synonyms, self.token_sent, self.index)
synonym_scores['grammaticality'] = hf.get_gram_score(
self.synonyms, self.token_sent, self.pos_sent, self.index)
#filtering process
synonym_scores = synonym_scores[synonym_scores['sem_sim'] < 0.15]
synonym_scores = synonym_scores.sort_values(by=['complexity'])
try:
top_synomym = synonym_scores.synonyms.values[0]
except:
return [self.word]
return top_synomym
else:
return [self.word]