How to address Your SAP customers pain points with NLP Chatbot Tool
In the previous blog, I have given an overview of HOW TO ADDRESS YOUR CUSTOMER’S PAIN POINTS INSTANTLY WITH SAP INTEGRATED CHATBOT? In this blog, I’ll discuss in detail about the NLP Chatbot tool.
Agents, Intents, Entities, Actions, Parameters, Webhooks and Integrations are the very basic things in Dialogflow. In this blog, I will be giving a brief idea of all the important components and probably develop a bot which can give an automatic reply. To start with, we shall take a very basic scenario and test it in the console of Dialogflow. First, let us understand the components before we dive into the development.
The bot itself is an agent. We can create multiple agents with one Dialogflow account. Every agent will have a unique Project ID, client access token and developer access token. These tokens are used in the code in the later phases of chatbot development.
This is the place where we train our bot. The bot matches the intents based on the user’s question. Here we have training phrases, these are nothing but the variety of possible sentences a user can ask a bot for a specified scenario. The sentences will include parameters inside them. The training entirely relies on the phrases and parameters. The conversation flow starts with the recognition of appropriate intent.
Certain useful information must be recognized from the user’s question before responding back to the user. Thus, Entities are created to recognize useful information in the content. For instance, in the below scenario plant name must be recognized. So, we create a system entity @sys.any which recognizes the text after plant as PlantName.
The activity being performed by an intent will be represented in the action name. Actions are used to refer to the respective intent. This will be better understood when we start coding.
The parameter name is an identity to refer to the entity value. This is nothing but a variable name in common terms. This acts as an input value to obtain the desired response.
Webhook is a web service which helps in bringing data from our server to the Dialogflow. The API is entered as a URL in the fulfillment section. It is very important to enable the webhook in every intent if we want to get data from external resources.
The above data might be very theoretical, so let us try getting an automatic reply from the bot.
Follow these steps
1) Go to https://dialogflow.com/.On the top right corner click on GO TO CONSOLE. Sign in with your Google account.
2) Click on the drop down. You can see create a new agent. Click on it and give a name to your agent.
3) Open your agent and create an intent. On the top right you can find a blue button with the name “CREATE INTENT”. Give a name and enter the training phrases mentioned below.
Heyy!! Did you observe that it has recognized the currency by itself? Also, a system entity is assigned to it. It has a predefined parameter name.
4) Now let’s create a response something like “You mean 500 euros”. The user-entered Unit Currency is the response. Pretty simple and interesting right!! Let’s do it. Enter the below text in the response section.
5) Now, save the intent. REAL TEST!! Let’s test in the Dialogflow console on the top right corner which says “Try it now”. For instance, type “Can u lend me 1000 rupees” in the console. You can try with any other text.
WOAHHH!!!!! Have you seen the response in the above screenshot?? Yeah…..!!!!
We have tried developing a basic scenario in Dialogflow. But this is not enough, we have to bring data from our backend based on the input entered by the user in the Dialogflow console.
Watch out this space for my next blog, where I’ll discuss how the Backend code is developed. I will be using node.js as my backend framework.