IntelliBot Demo

AI Based chatbot using Python, AIML, Angular and Bootstrap

Requirements

Python >= 3.5x
AIML >= 0.9
Flask
Lxml
Beautifulsoup4
Jieba
Tensorflow 1.2 
MySQL 5.5

Installation

Clone and navigate to chatbot directory.
Install the required packages.
```
$ pip install -r requirements.txt
```
You're done, chat with your IntelliBot :)

Run in the browser

$ cd Medius_Chatbot/core
$ python web/server.py

> ......
> * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

Run in Command Prompt (CMD)

$ cd chatbot_3.6
$ python web\server.py  

> Kernel bootstrap completed
> User > 
> ......
> * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

###Processing flow Step 1: Pretreatment Limit word count Filter sensitive words (disgusting, political, pornographic, illegal...)

Step 2: Knowledge Base Matching (AIML) Basic functions: say hello, gossip... Exception handling: the problem is too long, a blank question, no response can be found... Emotional answer: expression, praise, ridicule... If the match does not match, go to step three.

Step 3: Internet Searching Article----A daily article Jokes----Anecdote Weather----weather If the search does not find answer, go to step four.

Step 4: Neural Network The next-generation dialogue engine based on the Seq2Seq model not only trains the best answer in the existing answer, but can create a human-like answer.

Original Cornell Dataset: [cleaned it using Python script] http://www.cs.cornell.edu/~cristian/Cornell_Movie-Dialogs_Corpus.html

Reddit Dataset: https://www.reddit.com/r/datasets/comments/3bxlg7/i_have_every_publicly_available_reddit_comment/

Pre-trained Dataset: https://drive.google.com/file/d/1mVWFScBHFeA7oVxQzWb8QbKfTi3TToUr/view

Standford NLP https://stanfordnlp.github.io/CoreNLP/

Screenshot

Author

Mohammad Nuruzzaman, PhD
University of New South Wales, Australia.
Currently working as a Data Scientist at ELMO Software Ltd.

Additional Resources

python -m nltk.downloader punkt
python -m nltk.downloader stopwords
import nltk
nltk.download('wordnet')
tensorboard --logdir Data\Result\
java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000
SET PASSWORD FOR 'root'@'localhost' = PASSWORD('soho1234');

CChengz/MyPhD-Chatbot