项目作者: rounakdatta

项目描述 :
Natural Language Conversation
高级语言: Python
项目地址: git://github.com/rounakdatta/digital-assistant.git
创建时间: 2018-08-05T06:08:22Z
项目社区:https://github.com/rounakdatta/digital-assistant

开源协议:

下载


Usage

Add data to MongoDB

make data

Get the application running

make run

Head over to localhost:5000

Demo

System Design

/dataprovider/dataprovider.py - After core query is parsed, control passes to get_short_answer() where the processing logic lies. There, MongoDB lookup occurs. The priority of search is MongoDB -> wolfram -> wikipedia.

/nlp/rasa.py - Parsing logic - What to do with intents + all the hard-coded replies are defined here. Based on intent, if falls into hard-coded case, randomly one of the replies in the reply array is returned.

initDb.py - General documentation of how to push/pull to/from MongoDB - also includes the data which has already been inserted to MongoDB. This is the file needed to insert data to MongoDB (manually).

/data/training_data.json - Training dataset containing all the intents as well as entities. Use rasa-nlu-trainer instead of manual filling.

Development Status

  • The bot learns to speak
  • Database integration
  • Flask API
  • Front-End to interact
  • The bot remembers
  • The bot gives price, COD details, discounts
  • The bot can bargain (randomly bring down product price upto given limit)
  • The bot can accept an order and redirect you to the payment page
  • The bot is mature
  • The bot doesn’t faint at all

Tip: Make use of Studio 3T on any platform to view MongoDB using an UI.