Skip to content

README.md

Latest commit

 

History

History
28 lines (22 loc) · 1.13 KB

README.md

File metadata and controls

28 lines (22 loc) · 1.13 KB
Raw

python-chat-bot

How to run

Install Jupyter or Jupyter notebook. Install nltk, tflearn, numpy, tensorflow Open chat.ipynb with Jupyter and start running.

What I learned:

Deep Neural Network is used interchangably with Multi Layer Perceptron (in tflearn) Data wrangling techniques: tokenizing, POS labeling, chunking. Labeling tokens based on POS labels Categorizing training data NLP techniques: stemming, lemmatizing

What I don't understand:

Why do datasets require a dictionary having all True values?

What's the most important thing?

Thoughts of correctness and statistical techniques are more important than blindly writing code. Jupyter helps a lot.

Techniques

I completed the job with flying colors!

  • Tokenizing: splitting strings into phrases or keywords. E.g. I, completed, the, job, with flying colors
  • Phrase tokenizing: with flying colors (it's an idiom)
  • POS labeling: I (subject), completed (verb), job (noun)...
  • Stemming: completed -> complete (the original word)
  • Lemming: completion -> completion, completed -> complete (one is a noun, one is a verb, also consider other factors)