deeptea

deeptea is a deep learning and natural language processing tool. Its main purpose is to bring the power of deep learning/nlp to everyone, without a steep learning curve. This project was inspired a site I volunteered with. They had many internet trolls, and wanted to practically apply artificial intelligence to defeat them before they strike by learning their patterns. In local tests, the AI worked 10/10 times once it was trained.

Install

Install with npm:

npm install --save deeptea

How about Yarn?

yarn add deeptea

Introduction

This README reflects deeptea v0.0.x and greater currently

TL;DR

• Allows anyone to easily without a giant learning curve create beautiful deep learning / NLP-powered programs
• You can do it in under 10 lines of Javascript
• Built-in models taught to the neural network(s) to help kick-start creating
• Train your own natural language processing models, just supply an input and a category

Examples

Creating NLP-based bot in 10 lines or less

"use strict";
let deep = require('deeptea');
let Deeptea = deep;
let sampleInput = 'Hey there! I was outsied with my sister and dog when I stumbled upon a chess board! I remember I did have a computer virus, but I think I got rid of it.'; // intentional typo - yes, it understands your finger-thumbling
let demo = new Deeptea(sampleInput);
demo.run();
console.log(example_WordType.getNaturalType());

7 Lines. Booyah!

Creating NLP-based bot that'll respond to you, based on types

"use strict";
let deep = require('deeptea');
let Deeptea = deep;
let sampleInput = 'Hey there! I was outside with my sister and dog when I stumbled upon a chess board! I remember I did have a computer virus, but I think I got rid of it.';
let example_WordType = new Deeptea(sampleInput);
// OK, let's train and run the model(s)
example_WordType.run();
// Now we do logical processing
if(example_WordType.getNaturalType() === 'common-phrase-outdoor') {
  var reply = '';
  reply += 'Hey there!\n';
  let scope = example_WordType.getScope();
  if(scope.animals >= 1) {
    reply += 'I love that you have a pet! What\'s their name/names?\n';
  }
  reply += 'Thanks for the cool story!';
  console.log(reply);
}

OK, but how can I understand how it came to the conclusion of common-phrase-outdoor?

Breaking down the madness with 6 lines

Simply run these inputs, and you'll see the exact reasoning how it came to this conclusion:

console.log('Logical scope:');
console.log(example_WordType.getScope()); // ex: { computer: 8, animals: 4, ...}
console.log('Total scope:');
console.log(example_WordType.getTotalScope()); // ex: ['computer','computer'....]
console.log('Natural Type: ');
console.log(example_WordType.getNaturalType()); // ex: 'natural-outdoor-phrase--generic-package'
console.log('Keywords: ');
console.log(example_WordType.getKeywords()); // ex: ['dog', 'cat', 'computer virus', 'ransomware', 'wanacry', 'github', 'Terrain']

That's all for now

But we're still working on it. Stay tuned, help train our default models.

What we're working on next?

• Automatic Data Training
• Web Scraping (uses ADT above to train)
• Saving and making API-accessible endpoints