Modelling Ideas

[ian-whitestone]

• [ ] Distress score prediction
• [ ] Topic modelling
  • unsupervised clustering, then manually tag based on keyword frequencies
  • or use i-carol tagged categories
• [ ] Sentiment analysis using out of box method
  • Change in sentiment over the course of the chat
  • identify convos with large change in sentiment to serve as examples/training material or for recognition of counsellors who did a great job
• [ ] Model change in sentiment/distress over time
  • Need a prior probability (based on survey info entered), plus updates over time
• [ ] Message response suggestions
• [ ] Queue wait time prediction

[ian-whitestone]

Message response suggestions

Articles

• [ ] See Google's smart reply: https://arxiv.org/pdf/1606.04870.pdf
• [ ] https://arxiv.org/pdf/1506.05869.pdf
• Chatbots
  • [ ] NLP + Chatsbots
  • [ ] Chatbots
• [ ] AI generated reviews to help local businesses

Other

• DialogFlow
• LSTMs
  • LSTMs Intro
  • LSTMs Tensorflow
  • LSTMs Medium Post

Ideas

• Beginning of conversation

  • common opener phrases

• After period of inactivity

  • are you still there?
  • after x minutes of activity, give default response: 'Okay, I am going to go now but if you would still like to talk please come back to chat or call us anytime!'

[ian-whitestone]

Distress Score Prediction

Articles

• [x] https://blog.insightdatascience.com/how-to-solve-90-of-nlp-problems-a-step-by-step-guide-fda605278e4e

  • [x] Accompanying notebook

• [ ] https://medium.com/@lowweihong/5-simple-steps-to-create-a-simple-sentiment-analysis-on-movie-reviews-using-natural-language-69c7cba4b79b

• [ ] https://medium.com/@lowweihong/apply-simple-natural-language-model-to-predict-movie-reviews-using-python-bd06b1362be2

• [x] https://medium.freecodecamp.org/text-classification-and-prediction-using-bag-of-words-8aeb1396cded

• [ ] https://towardsdatascience.com/sentiment-analysis-with-python-part-1-5ce197074184

• WORD2VEC

  • [x] https://towardsdatascience.com/mapping-word-embeddings-with-word2vec-99a799dc9695
  • [x] https://medium.freecodecamp.org/how-to-get-started-with-word2vec-and-then-how-to-make-it-work-d0a2fca9dad3
  • [ ] https://skymind.ai/wiki/word2vec
  • [ ] https://towardsdatascience.com/word-embedding-with-word2vec-and-fasttext-a209c1d3e12c
  • [ ] http://mccormickml.com/2016/04/19/word2vec-tutorial-the-skip-gram-model/

• [ ] Norvig on spellchecker implementation

• [x] NLP Python Series

Text pre-processing

• strip out machine generated text (i.e. 'Conversation Terminated')
• convert to lowercase
• spellcheck??
  • can you quantify how many misspelled words there are?
    • Check if word is in vocab/corpus
  • pyhunspell
  • SymSpell (not a python package)
• tokenization
• remove stop words
• stemming and/or lemmatization
  • spacy has a lemmatizer
  • stemming can be done with nltk

Feature Ideas

• Use parts of speech for feature generation
• Create separate word encodings for kid vs counsellor
• tf-idf
• word2vec

Models

• bag of words + tf-idf + logistic regression
• bag of words + tf-idf + GBM
• deep learning
  • CNNs

Interpretability

• Plotting with PCA after each method
• LIME

Packages

• nltk
• spacy

[ian-whitestone]

General NLP Resources - Books

• Foundations of Statistical Natural Language Processing
• Speech and Language Processing