What are Intents?
Updated
Overview
Sprinklr's Conversational AI understands user intentions to help them achieve their goals. This article explains how intents are detected and classified, their importance, and how Sprinklr leverages data analysis for accurate intent recognition.
Understanding Intents
Conversational AI identifies user intentions through tokenization, dividing text into meaningful pieces. Multiple intents can be detected and the primary intent is defined during intent creation. Intents represent categories of similar phrases, capturing user goals like making a purchase.
Entities
Entities enrich intents by identifying specific details in user messages, such as colors or products. They extract information like phone numbers or email addresses. Understanding entities enhances AI comprehension.To learn more about entities,See What are Entities?
Data Requirements for Robust AI Models
Creating robust AI models requires diverse data. Having too few or overly similar expressions for an intent limits the model's performance. Sprinklr leverages a large dataset to train models accurately.
Unsupervised Clustering
Unsupervised clustering automates grouping data into clusters without human annotation. Sprinklr utilizes this technique to analyze vast social inbound data, creating clusters that serve as building blocks for intents, entities, and FAQs.
Types of Intents
Issue intents address customer concerns
Dialogue intents capture non-issue-related conversation segments.
Example: Customer: Hello, My laptop does not turn on. Agent: I’m sorry to hear that. Do you know the serial number of the device? Customer: Yes, it’s Y29576FB Agent: Perfect, thank you. I see you have a Power Mini Pro is that correct? Customer: Yes that’s correct Agent: Thanks for that. I’m going to ask you some questions so that we can properly pinpoint the issue. Agent: Does it power on if you have the charger plugged in? Customer: Let me check, I’ll be right back. In this example, the customer’s first sentence defines the Issue Type Intent “Hello, My laptop does not turn on.” The issue type intent here could be ‘Laptop Boot issue’. The customer’s third & fourth sentences, however, are Dialogue Intents. “Yes that’s correct”: The dialogue intent here is ‘Affirmation’. Similarly “Let me check, I’ll be right back”: The dialogue intent here might be ‘Customer carrying out diagnosis’ |
Permissions
To grant users access to Intents in Sprinklr, they must have the following permission: Conversational AI > Conversational AI - Message Processing > All |
