MovieMate A chatbot that knows everything about movies!

Movie Mate - Smart (NLP) Chatbot

• It uses the wit.ai services.

wit.ai

• It was acquired in 2015 by Facebook.
• Provides speech & text recognition and processing as a service.
• Uses machine learning and natural language processing.
• wit.ai - 40+ languages, large community, free for commercial applications as well.

Speech/Text -------> wit.ai --------> Entities (intent / other entities);

• wit can be used directly over http.
• It has ready to use SDKs for Ruby, Python and Nodejs.
• Login in - your developed apps will be displayed on the dashboard.
• In order to create new App - "+" , language, private/public.

wit.ai | Intent & Training

• Click on the app -> understanding -> Create the intent and entities.
  • tell me about war of the worlds.
    • intent - movieinfo
    • movie - "war of the worlds" (entites)
    • validate
  • click on the entities - see all the training data.

Built-in NLP and wit.ai

• Built-in NLP <------> FB Messenger <-------> Chatbot (Nodejs APP)
• Built-in NLP automatically parse messages into fundamental entities.
  • Hi, I'll meet you tomorrow at 10 am.
  • entities:{ greetings: [{confidence: 0.9689, value: true}], datetime: [{confidence: 0.9509, value: 'Sun...}]}

wit.ai <---> Built-in NLP <---> FB Messenger (Query) <---> Nodejs chatbot App -> entities.

• developers.facebook App - built-in NLP section - Turn it on.
• wit.ai - settings - server access token - copy.
• Back in developers.facebook App ---> built-in NLP section
  • select the page.
  • Paste Wit Token - save.

Extracting Entities

FB Messenger <---> FBeamer <--> Extract entity values <---> getMoviesData --(The MOVIE DB) -- getDirector() ---> createResponse --> FBeamer <--> FBMessenger

Movie Database Application

• themoviedb
• Sign up and login.
• Profile -> Settings -> API
• Generate the API Key -> by providing basic Info.
• Insert the generated key in the (tmdb) config/development.json file.

Understanding the Movie API

• Developers - themoviedb
• For movies -> Search -> Search movies
• GET / search/movie
  • Try it out tab
    • Variables: api.key -> insert the tmdb key generated previously.
    • Query String - Query - (Movie Name)
    • Send Request.
• GET /movie/{movie_id}/credits

Summary

• Built a simple Q&A chatbot.
• Understood NLP helps build smarter chatbots.
• Introduced to Wit.ai.
• Used Movie DB to fetch movie related information.

Step for Review

Facebook Page Creation and App Id Generation Details

I. Creation of a Page

• Facebook - create a page.
• Choose the category - Business or brand.
• Update the information accordingly - Test the button to update the messager chat.

II. Creation of Application Id

• Facebook for developers - https://developers.facebook.com/;
• MyApps - Add New App
• Create a New App Id - Display Name , create App Id.
• Developers Facebook Page opens - Messanger - Settings
  • Token Generation - Link the respective created page and generate the Access token.
    • In the code base - config/development.json file - page access token.
  • Verify Token Generation
    • In cmd
      • node
      • require('crypto').randomBytes(24).toString('hex')

III. Creation of Webhooks

• Downloads Ngrok - Run it.
• http E.g D:/.../ngrok http 3000
• This generate url -> https://1332f745.ngrok.io;
• Set up Webhooks
  • Callback URL - the generated url from ngrok.
  • Verify Token - generated crypto 24 byte token - config/development.json file.
  • Subscription field - messages.
• Choose the Page (Movie Mate) -> Subscribe;

IV. Wit.ai (Natural Language Processing - Tool)

• wit.ai homepage.
• login - github, etc;
• Create a new App

V. Movie Database Application

• URL -> https://www.themoviedb.org
• Sign up and login.
• Profile -> Settings -> API
• Generate the API Key -> by providing basic Info.
• Insert the generated key in the (tmdb) config/development.json file.

VI. Understanding the Movie API

• URL -> https://developers.themoviedb.org
• For movies -> Search -> Search movies
• GET / search/movie
  • Try it out tab
    • Variables: api.key -> insert the tmdb key generated previously.
    • Query String - Query - (Movie Name)
    • Send Request.
• GET /movie/{movie_id}/credits

VII. Deploying it in Heroku Cloud

• URL -> https://dashboard.heroku.com/login;
• Create a free account -> go to the dashboard.
• Install the Heroku CLI for windows.
• URL -> https://devcenter.heroku.com/articles/heroku-cli
• While Deploying on Heroku
  • package.json file -> remove the main key, mention the exact node version.
  • invoke the start, engine -> node engine.
  • delete app.js file as it not necessary.
  • create .gitignore file.
    • node_modules
    • config/development.json
  • Ensure to have git install.
    • In cmd -> git init
• Login in to the heroku account from the cmd
  • heroku login
  • email id, password
  • git add , git commit, git push heroku master
• Heroku offers cloud computing
  • heroku create
  • an app is created and a https link is also provided.
  • In the facebook developers page -> Webhooks
    • Edit Subsciption -> the https link, verify token.
• Back to the Heroku Dashborad -> settings
  • reveal config var -> define the key values.
• Test the chatbot of Facebook.
  • In case it doesn't work.
    • messanger - settings - [webhook]
    • select the page.
    • unsubscribe and then subscribe it.

VIII. Making an APP public on Facebook / Approval Process

• URL -> Facebook.developers.com
• App Review :
  • Face developers page
  • Messenger - settings - bottom part - review.
  • Click on Add to submission - show the required actions.
  • privacy policy - iubenda.com;
  • provide examples and complete the approval process.
  • On approval, click the slide -> turn on the chat bot and make it publically available.
• Tester - add ppl to test the chatbot.