SEFR CUT (Stacked Ensemble Filter and Refine for Word Segmentation)

Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble (EMNLP 2020)
CRF as Stacked Model and DeepCut as Baseline model

Read more:

• Paper: Domain Adaptation of Thai Word Segmentation Models using Stacked Ensemble
• Blog: Domain Adaptation กับตัวตัดคำ มันดีย์จริงๆ

Install

pip install sefr_cut

How To use

Requirements

• python >= 3.6
• python-crfsuite >= 0.9.7
• pyahocorasick == 1.4.0

Example

• Example files are on SEFR Example notebook
• Try it on Colab

Load Engine & Engine Mode

• ws1000, tnhc
  • ws1000: Model trained on Wisesight-1000 and test on Wisesight-160
  • tnhc: Model trained on TNHC (80:20 train&test split with random seed 42)
  • BEST: Trained on BEST-2010 Corpus (NECTEC)

  sefr_cut.load_model(engine='ws1000')
  # OR
  sefr_cut.load_model(engine='tnhc')
  # OR
  sefr_cut.load_model(engine='best')
  

• tl-deepcut-XXXX
  • We also provide transfer learning of deepcut on 'Wisesight' as tl-deepcut-ws1000 and 'TNHC' as tl-deepcut-tnhc

  sefr_cut.load_model(engine='tl-deepcut-ws1000')
  # OR
  sefr_cut.load_model(engine='tl-deepcut-tnhc')
  

• deepcut
  • We also provide the original deepcut

  sefr_cut.load_model(engine='deepcut')
  

Segment Example

• Segment with default k

  sefr_cut.load_model(engine='ws1000')
  print(sefr_cut.tokenize(['สวัสดีประเทศไทย','ลุงตู่สู้ๆ']))
  print(sefr_cut.tokenize(['สวัสดีประเทศไทย']))
  print(sefr_cut.tokenize('สวัสดีประเทศไทย'))
  
  [['สวัสดี', 'ประเทศ', 'ไทย'], ['ลุง', 'ตู่', 'สู้', 'ๆ']]
  [['สวัสดี', 'ประเทศ', 'ไทย']]
  [['สวัสดี', 'ประเทศ', 'ไทย']]
  

• Segment with different k

  sefr_cut.load_model(engine='ws1000')
  print(sefr_cut.tokenize(['สวัสดีประเทศไทย','ลุงตู่สู้ๆ'],k=5)) # refine only 5% of character number
  print(sefr_cut.tokenize(['สวัสดีประเทศไทย','ลุงตู่สู้ๆ'],k=100)) # refine 100% of character number
  
  [['สวัสดี', 'ประเทศไทย'], ['ลุงตู่', 'สู้', 'ๆ']]
  [['สวัสดี', 'ประเทศ', 'ไทย'], ['ลุง', 'ตู่', 'สู้', 'ๆ']]
  

Evaluation

• Character & Word Evaluation is provided by call fuction evaluation()
  • For example

  answer = 'สวัสดี|ประเทศไทย'
  pred = 'สวัสดี|ประเทศ|ไทย'
  char_score,word_score = sefr_cut.evaluation(answer,pred)
  print(f'Word Score: {word_score} Char Score: {char_score}')
  
  Word Score: 0.4 Char Score: 0.8
  
  answer = ['สวัสดี|ประเทศไทย']
  pred = ['สวัสดี|ประเทศ|ไทย']
  char_score,word_score = sefr_cut.evaluation(answer,pred)
  print(f'Word Score: {word_score} Char Score: {char_score}')
  
  Word Score: 0.4 Char Score: 0.8
  
  
  answer = [['สวัสดี|'],['ประเทศไทย']]
  pred = [['สวัสดี|'],['ประเทศ|ไทย']]
  char_score,word_score = sefr_cut.evaluation(answer,pred)
  print(f'Word Score: {word_score} Char Score: {char_score}')
  
  Word Score: 0.4 Char Score: 0.8
  

Performance

How to re-train?

• You can re-train model in folder Notebooks We provided everything for you!!

  Re-train Model

  • You can run the notebook file #2, the corpus inside 'Notebooks/corpus/' is Wisesight-1000, you can try with BEST, TNHC, and LST20 !
  • Rename variable name CRF_model_name
  • Link:HERE

  Filter and Refine Example

  • Set variable name CRF_model_name same as File#2
  • If you want to know why we use filter-and-refine you can try to uncomment 3 lines in score_() function

  #answer = scoring_function(y_true,cp.deepcopy(y_pred),entropy_index_og)
  #f1_hypothesis.append(eval_function(y_true,answer))
  #ax.plot(range(start,K_num,step),f1_hypothesis,c="r",marker='o',label='Best case')
  

  • Link:HERE

  Use your own model?

  • Just move your model inside 'Notebooks/model/' to 'seft_cut/model/' and call model in one line.

  SEFR_CUT.load_model(engine='my_model')
  

Citation

• Wait our paper shown in ACL Anthology

Thank you many code from

• Deepcut (Baseline Model) : We used some of code from Deepcut to perform transfer learning
• @bact (CRF training code) : We used some from https://github.com/bact/nlp-thai in training CRF Model