Building A Conversation Product with Google’s DialogFlow

A Quick Introduction to Building A Conversation Product with Google’s DialogFlow #DevFest17

Transcript

1. Building A Conversation Product with Google’s DialogFlow By Goodness Kay #DevFest17
2. Summary 1. An Introduction 2. Examples of Conversational Applications 3. Why Conversational Products/Applications? 4. The Core of Conversational Applications 5. Starting With Google’s DialogFlow( Formerly Api.ai) 6. Demo - Chinedu
3. Introduction • We are used to the old way of doing things and have not noticed there is something better. • Real-Time Communication: Humans like asking questions. • Communication Rules the World.
5. Why Conversational Products/Applications? • Real-time access to information • Structured Information (Business Intelligence) • Humans are communication beings and you will always find them on social media networks • Better understanding of how the platform works because of structured directions.
6. The Core of Conversational Applications (Artificial Intelligence) • "machines will be capable, within 20 years, of doing any work a man can do." (AI innovator Herbert Simon.1965) • AI can be put in place in Game playing, Natural Language Processing,Speech recognition ,Expert Systems,Robotics and so on. From here, the focus will be on Natural Language Processing
7. Starting With Google’s DialogFlow( Formerly Api.ai) Humans are creative in the way they communicate and a particular statement can be said in different ways. Chat Bots are computer programs and not humans. The question is how do one manage a particular expression communicated in different ways? The answer is there is no way you can handle that on your own and moreover, it is most effective to make use of Natural Language Processing Platforms like DialogFlow(Api.ai) to handle such situations. A cool thing about this platform is that you can deploy faster with them and make your applications more interactive. I will work you through how DialogFlow works and then demo a bot I built called Chinedu
8. Creating A Conversational Application With DialogFlow • Set up a DialogFlow Account at https://dialogflow.com/ .
9. • Create an Agent by clicking on the ‘create agent’ button • Once the Agent is created, we can now talk about Intents, Entities,Integrations and Small talks
10. Intents From the word intention, it has to do with what the user has in mind. A user(A lady) may want to say ‘hello’ instead of ‘bros, how far’. So, the intents option in DialogFlow helps to acheive that by giving answers to a group of statements that look alike. • To create an intent in DialogFlow, click on the ‘Create Intent` Button though by default you have 2 intents E.g Default fallback intent (List of default responses to statement made and not understood by the application) & Default Welcome Intent(List of default responses to initiated or welcome statements by users) • You should state the name of the intent, the action, what a user can most likely say and the bot’s response to that scenerio. • Then you are good to go but know that you can create as many as possible intents depending on the kind of application being built.
12. Entities Entities can help in extracting parameter values from natural language inputs. Any important data you want to get from a user's request, will have a corresponding entity. For example, the numbers in a user’s statement. You create entities from the ‘Entities Tab’ on your dashboard.
13. Integrations With DialogFlow, after creating intents and entities and also configuring small talk, you can work with the chat bot by integrating with social media platforms or integrating on a website. To do this, you just need to click on the integrations tab on your dashboard and set up for the platform you want to run on.
14. Small Talks Small Talk is another cool feature of DialogFlow. Small Talks include some chitty chatty type of conversation such that even if you don’t have intents set up, you can still converse with the bot. It can be activated by clicking on the Small Talk link on your DialogFlow dashboard.
15. • Some added features of DialogFlow is Training. The machine learning part of the conversational application. • There is the analysis part for analysing the performance of the bot(sessions).
16. Now, Let’s Play With Chinedu Original Igbo Boy Bot. Sells Okirika At Alaba Market, Lagos http://bit.ly/broschinedu
17. Thank You Goodness Kayode - SWE @Accounteer - Pusher of Codes - Linguist Turned Engineer - Scotch.io Advocate - Expert Mentor at Codementor @goodnesskayode @goodnesskay @goodnesskay