Chatbot Intro Project

Build

• createdb fitbot
• npm install
• npm run seed
• add secret files to root project directory

production

• npm start

development

• npm run start-dev-dialog-flow
• npm run start-dev-lex

Running unit tests

• createdb fitbot-test
• npm test

test dialog flow

• npm run unit

test lex

• npm run unit-lex

Bot Design

planning intents and entities

agent design checklist

Running end-to-end tests

• create a folder 'bin' in root directory
• download selenium-server-standalone-VERSION.jar and place in bin

test dialog-flow

• npm run start-dev-dialog-flow
• npm run end2end

Angular front end on port 4200

• cd fitbot-angular
• ng serve

Rasa

• checkout train-rasa
• cd fitbot-rasa
• make nlu-json
• make core-server
• make action-server

test lex

• npm run start-dev-lex
• npm run end2end

Introduction

This project will serve as an introduction to the CBA practice at (Cedrus)[http://cedrus.digital]. As a member of the CBA team, you will be expected to be heavily involved with both automation (RPA) and machine learning (AI/ML).

This project is designed to provide experience working with multiple platforms for Chatbots (required is use of either Watson Assistant and Amazon Lex and a second of your choice).

The chatbot's purpose can be your choice but should be able to help a user solve some kind of problem and be capable of using an external API to get the user useful information.

A leisurely example might be a chatbot assist you in finding a TV show or Movie to watch. A more practical (preferred) example could be an internal app to help book a conference room at Galvanize. Both would require the date, time, number of users, and to incorporate with an external system.

Requirements

• UI build in framework of your choosing
• Chat input with scrollable chat history
• Dynamic response nodes that may include:
• Customs styles
• External links
• Link Previews
• Generated Documenation using library of your choice
• Login with User Authentication
• Build chatbot using minimally 2 AI platforms (Watson Assistant and Amazon Lex)
• NodeJs Backend
• End to End and Unit testing
• Ability to select chatbot platform by app launch configuration
• Deployment procedure configurable by environment.
• Source Control w/ Git with appropriate branch management.
• Protection of assets (API keys and any proprietary info should not be committed)

Nice to have

• CI Deployment
• Additional Chatbot Platforms
• Containerization

Phase I

The major technical concepts will cover:

• Feature-branch development with git
• Automated testing
• API
• Browser DOM, XPath, and CSS selectors
• Containers and VM's

Key software methodoloy concepts:

• Sprints and milestone tracking
• Kanban boards (feature development)
• Bug tracking / QA
• Sofware delivery and deployment
• Automatable and repeatable processes

Getting started

• Create a source code repo using github, gitlab, or bitbucket
• Create a kanban board
• Create a project wiki (for Release Notes)
• Create your development environment

During this phase you should do some basic setup and configure you development environment, build process and automated testing process. Depending on your chosen framework some tooling may be be provided out of the box.

The first view should be a login page with a username an password field.

This is good time to verify that your end to end tests are working. You should have a test verifying the expected rendering of your login form.

You should also have a unit test verifying connectivity with one of the chat platforms

Note

The details and suggestions for this project will grow and likely change but this is a good place to start

Use this file for the README.md in the project, placing whatever you think is important at the top of this file

API DOCS

Table of Contents

• queryFood
  • Parameters
• buildCaloriesStatus
  • Parameters
• caloriesRemaining
  • Parameters
• initiateDialogFlow
  • Parameters
• messageDialogFlow
  • Parameters
• handleResponseDialogFlow
  • Parameters
• Bot
  • Parameters
  • Properties
• initiateLex
  • Parameters
• messageLex
  • Parameters
• handleResponseLex
  • Parameters
• saveFoodLog
  • Parameters
• initiateWatson
  • Parameters
• messageWatson
  • Parameters
• handleResponseWatson
  • Parameters
• getNutritionInfo
  • Parameters
• handleDialogCodeHook
  • Parameters
• handleFulfillmentCodeHook
  • Parameters
• handleQueryFood
  • Parameters

queryFood

fulfillment handler for dialog flow

Parameters

• agent Object dialog flow webhook client
• parameters Object parameters used to fulfill intent

buildCaloriesStatus

Creates a response depending on whether a user is over or under their caloric goal

Parameters

• dailyGoals Number number of calories allowed per day for user
• calories Number number of user's actual calories

Returns String conditional response string dependant on net calories

caloriesRemaining

Gets user's foodlogs by today's date. Calculates net calories and returns a response message based on the result.

Parameters

• user Object current user
• foodName String? food name is included when this function is called immediately after logging an item.

Returns String resultant response to user

initiateDialogFlow

initiates the dialog flow bot

Parameters

• user Object currently logged in user
  • user.id String user id

Returns String session id

messageDialogFlow

sends user input to Dialog Flow

Parameters

• sessionUserId String id used to reference current session with user
• text String input text user is sending
• callback Function function that handles error, or sends response back to user

handleResponseDialogFlow

handles bot response depending on intent name and if all requeired params are present

Parameters

• user Object currently logged in user
• response Object object received from dialog flow
  • response.fulfillmentText String default response sent from dialog flow
  • response.allRequiredParamsPresent Boolean true if all required params are present for fulfillment
  • response.intent Object contains info about current intent
  • response.parameters Object contains parameters obtained through slot filling
  • response.outputContexts Object contains parameters within contexts persisted throughout the conversation

Returns String response message to user

Bot

bot object client interacts with.

Parameters

• type String identifies the bot service (Lex, Watson, or DialogFlow)

Properties

• initiate Function creates a new service instance, returns a session id
• message Function sends a user input, callback returns a response to user
• handleResponse Function handles intent fulfillment, this is called within the callback of this.message

initiateLex

initiates the lex service

Parameters

• user Object currently logged in user
  • user.id String user id

Returns String session id

messageLex

sends user input to Lex

Parameters

• sessionUserId String id used to reference current session with user
• text String input text user is sending
• callback Function function that handles error, or sends response back to user

handleResponseLex

handles bot response depending on intent fulfillment

Parameters

• user Object currently logged in user
  • user.id String user id
• response Object object received from lex
  • response.intentName String identifies intent to be handled
  • response.dialogState String used to determine if intent is ready for fulfillment
  • response.sessionAttributes Object attributes that can persist throughout a conversation
    • response.sessionAttributes.foodName String if defined, will be part of caloriesRemaining result string
  • response.slots Object contains parameters needed to fulfill intent

Returns String response message to user

saveFoodLog

saves a food log to the database

Parameters

• user Object user associated with the food log
• foodLog Object object with required properties

Returns Object saved foodLog returned from database

initiateWatson

initiates the watson service

Parameters

• callback Function sends session id back to user

messageWatson

sends user input to Watson

Parameters

• sessionUserId String id used to reference current session with user
• text String input text user is sendng
• callback Function function that handles error, or sends response back to user

handleResponseWatson

handles bot response depending on action object

Parameters

• user Object currently logged in user
  • user.id String user id
• response Object object received from watson, contains action parameters needed to handle action

Returns String response message to user

getNutritionInfo

calls external api to get nutrition info

Parameters

• query String string created using buildFoodQuery
• success Function success handler
• failure Function failure handler

Returns Object nutrition info object

handleDialogCodeHook

handles slot filling and validations

Parameters

• request Object lambda request object
• sessionAttributes Object attributes that persist throughout a session
• currentIntent Object
  • currentIntent.slots Ojbect parameters used for fulfilling intent
• inputTranscript String user input string

Returns Object lambda response object

handleFulfillmentCodeHook

handles fulfillment by asking user to confirm log

Parameters

• request Object lambda request object

Returns Object lambda response object

handleQueryFood

main queryFood intent handler

Parameters

• request Object
  • request.invocationSource String determines whether intent ready for fulfillment or not.

Returns Object lambda response object

FULFILLMENT HANDLERS

Table of Contents

• queryFood
  • Parameters
• getNutritionInfo
  • Parameters
• handleDialogCodeHook
  • Parameters
• handleFulfillmentCodeHook
  • Parameters
• handleQueryFood
  • Parameters

queryFood

fulfillment handler for dialog flow

Parameters

• agent Object dialog flow webhook client
• parameters Object parameters used to fulfill intent

getNutritionInfo

calls external api to get nutrition info

Parameters

• query String string created using buildFoodQuery
• success Function success handler
• failure Function failure handler

Returns Object nutrition info object

handleDialogCodeHook

handles slot filling and validations

Parameters

• request Object lambda request object
• sessionAttributes Object attributes that persist throughout a session
• currentIntent Object
  • currentIntent.slots Ojbect parameters used for fulfilling intent
• inputTranscript String user input string

Returns Object lambda response object

handleFulfillmentCodeHook

handles fulfillment by asking user to confirm log

Parameters

• request Object lambda request object

Returns Object lambda response object

handleQueryFood

main queryFood intent handler

Parameters

• request Object
  • request.invocationSource String determines whether intent ready for fulfillment or not.

Returns Object lambda response object

FRONT END DOCS

Table of Contents