Skip to content

Rule based chat-bot using CNN based on multi class text classification which responds to all queries on Deep Learning class. As a fallback option added generative chat-bot trained on cornell movie dialog corpus using sequence to sequence RNN model.

Notifications You must be signed in to change notification settings

Dhrumil1808/chatbot

 
 

Repository files navigation

chatbot

TensorFlow = 1.3 Python: 2.7

Generate Data :

	python DeterministicGenerator.py test.gram 
  

Pre-processing (requires data generation) :

	python data_preprocessing.py
  

This file contains the latest checkpoint and the labels in the josn format. It also contains the information regarding word count and this folder needs to be placed in the sqs directory.

Predict CNN

    python predict.py 

For Sequence to Sequence Generator

  1. download trained model from: https://drive.google.com/open?id=1Ths1P1OPSTV6TGClo5vMmJPTJijFRh-L
  2. Unzip this in sqs/ckpt/cornell_corpus folder

Run on python 2.6/2.7 from here

Slackbot :

pip install slackclient

For NER:

pip install git+https://github.com/mit-nlp/MITIE.git

Also download :

https://www.dropbox.com/s/d4ncdbg88j4zzvs/new_ner_model.dat?dl=0 https://www.dropbox.com/s/3yhg2fm9qnzxu5y/total_word_feature_extractor.dat?dl=0

Make sure they are all in the same directory.



python starterbot.py

About

Rule based chat-bot using CNN based on multi class text classification which responds to all queries on Deep Learning class. As a fallback option added generative chat-bot trained on cornell movie dialog corpus using sequence to sequence RNN model.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%