https://github.com/Axilent/talkback

A declarative framework for chatbots and voice agent skills.
https://github.com/Axilent/talkback

Last synced: about 1 month ago
JSON representation

A declarative framework for chatbots and voice agent skills.

• Host: GitHub
• URL: https://github.com/Axilent/talkback
• Owner: Axilent
• License: mit
• Created: 2016-06-23T14:54:26.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2016-07-11T14:41:10.000Z (about 8 years ago)
• Last Synced: 2024-04-04T10:32:23.296Z (6 months ago)
• Language: Python
• Size: 52.7 KB
• Stars: 5
• Watchers: 4
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• starred-awesome - talkback - A declarative framework for chatbots and voice agent skills. (Python)

README

        
# Talkback

> A declarative framework for chatbots and voice agent skills.

The Talkback framework allows you to create chatbots and voice agent skills. Using a declarative syntax, you can define apps that interact with conversational interfaces. The framework will provide integration with popular platforms, and provide built-in natural language processing.

## Target Platforms

### Voice Agents

* Amazon Alexa

* Microsoft Cortina

* Apple Siri

* Google Home

### Messaging Platforms

* Facebook Messenger

* Slack

* Skype

* iMessage

## Core Concepts

### Intents

A *Talkback* app consists of one or more **Intents**. An Intent represents a user goal - something a user is attempting to accomplish. The developer can define any number of phrases that indicate the Intent, in an Intent Map that accompanies the Intent. *Talkback* will train it's built-in natural language processor (NLP) with the phrases in the Intent Map.

### Sessions

When an Intent is invoked, it is passed a **Session**. A Session represents the specific interaction with the conversational platform. The business logic of the Intent communicates with the user via the Session.

A Session can have associated Metadata, which can inform the Intent to particulars about the interaction.

A Session has several built-in methods of interaction that an Intent can use to communicate with the user:

* An **Interview** is a set of one or more questions sent to the user, for which their answers are collected. It can be thought of as analgeous to a form on a web or mobile app. The user will experience the Interview as a series of questions asked by the conversational interface.

* An Intent may use a Session to simply **Speak** to the user. This can be in the form of text or audio (depending on the underlying platform), or may in some cases allow the attachment of a media file like an image.

* An Interview question or Speaking may be accompanied by **Options**: pre-defined choices for the user that directly trigger action.

## Example Code

	from talkback.core import Intent

	

	class PayBill(Intent):

	    """User wants to pay their phone bill."""

	    def invoke(self,session):

	        user_id = session.meta['user_id']

	        total = get_bill_total(user_id)

	        pay_options = PayOptions(user_id)

	        session.speak("Your balance is %f. Do you want to pay your bill?" % total,options=pay_options)