Creating Conversational Experiences using Google Cloud DialogFlow

Google Cloud DialogFlow is an end to end developer platform for building natural and rich conversational experiences, across devices and platforms. It is a powerful tool to understand and process the natural language input. It helps to build smart conversational agents. Natural Language Understanding works by recognizing a user’s intent and responding accordingly based on training dataset. 

DialogFlow is built on some of the world class AI assets that are developed for products like Search based on the top contents and user’s previously searched results.

Few significant capabilities of DialogFlow in Search are illustrated below:

• Google NLP has the power of syntax analysis, which allows extracting tokens and sentences that identify parts of speech and create dependency trees for each sentence.

• Entity recognition enables agents to identify and label by types, line organization, events, products, etc.

• Sentiment analysis gives an understanding of the overall sentiment expressed in a block of text.

• Content classification allows classifying documents in over 700 predefined categories.

• Multi language support includes the ability to easily analyse the text in multiple languages and convert it according to the user's needs

Working Principle of DialogFlow

Training Data ->Proprietary language models by DialoFlow ->Unique Model->User Query Response.

Use Cases of DialogFlow 

• Customer Service: While building conversational interfaces the agent must perform tasks such as following up on past orders, scheduling appointments, assisting end-users with requests.

• IoT Devices: Controlling smart home appliances.

• Commercial insights: Users can know about their order status accordingly without even opening the application.

Chatbot Workflow in Google Cloud DialogFlow

The workflow mechanism of DialogFlow is

Design Phase-> Development Phase -> Deploy Phase

• In the Design Phase (what to build) the developer decides the tone and the personality of the chatbot based on a particular brand.

• In the development phase, the developer uses the DialogFlow to create agents with a combination of adding intents and responses in the console and writing code to connect to backend services.

• The deployment mostly depends on the different components that the chatbot needs to consider and which applications it will touch.

Defining Intents of Chatbot

Intents are the actions that users want to execute. It determines where the conversation will move forward and what an agent should do. The intents are of two types. These are Default Intents and Fallback Intents.

The best practices when defining intents are :

• Training phrases must be representative

• The intent actions should not overlap

• The intents must be prioritized that represent the most common request.