GETTING STARTED
Kore.ai XO Platform
Virtual Assistants Overview
Natural Language Processing (NLP)
Concepts and Terminology
Quick Start Guide
Accessing the Platform
Working with the Builder
Building a Virtual Assistant
Using Workspaces
Release Notes
Current Version
Previous Versions
Deprecations

CONCEPTS
Design
Storyboard
Dialog Tasks
Overview
Dialog Builder
Node Types
Intent Node
Dialog Node
Entity Node
Form Node
Confirmation Node
Message Nodes
Logic Node
Bot Action Node
Service Node
Webhook Node
Script Node
Group Node
Agent Transfer
User Prompts
Voice Call Properties
Dialog Task Management
Connections & Transitions
Component Transition
Context Object
Event Handlers
Knowledge Graph
Introduction
Knowledge Extraction
Build Knowledge Graph
Add Knowledge Graph to Bot
Create the Graph
Build Knowledge Graph
Add FAQs
Run a Task
Build FAQs from an Existing Source
Traits, Synonyms, and Stop Words
Manage Variable Namespaces
Update
Move Question and Answers Between Nodes
Edit and Delete Terms
Edit Questions and Responses
Knowledge Graph Training
Knowledge Graph Analysis
Knowledge Graph Import and Export
Importing Knowledge Graph
Exporting Knowledge Graph
Creating a Knowledge Graph
From a CSV File
From a JSON file
Auto-Generate Knowledge Graph
Alert Tasks
Small Talk
Digital Skills
Digital Forms
Views
Introduction
Panels
Widgets
Feedback Survey
Train
Introduction
ML Engine
Introduction
Model Validation
FM Engine
KG Engine
Traits Engine
Ranking and Resolver
NLP Configurations
NLP Guidelines
Intelligence
Introduction
Contextual Memory
Contextual Intents
Interruption Management
Multi-intent Detection
Amending Entities
Default Conversations
Sentinment Management
Tone Analysis
Test & Debug
Talk to Bot
Utterence Testing
Batch Testing
Conversation Testing
Deploy
Channels
Publish
Analyze
Introduction
Conversations Dashboard
Performance Dashboard
Custom Dashboards
Introduction
Meta Tags
Dashboards and Widgets
Conversations History
Conversation Flows
Feedback Analytics
NLP Metrics
Containment Metrics
Usage Metrics
Smart Bots
Universal Bots
Introduction
Universal Bot Definition
Universal Bot Creation
Training a Universal Bot
Universal Bot Customizations
Enabling Languages
Store
Manage Assistant
Plan & Usage
Overview
Usage Plans
Support Plans
Invoices
Authorization
Multilingual Virtual Assistants
Masking PII Details
Variables
IVR Settings
General Settings
Assistant Management
Data Table
Table Views
App Definitions
Sharing Data Tables or Views

HOW TOs
Build a Flight Status Assistant
Design Conversation Skills
Create a Sample Banking Assistant
Create a Transfer Funds Task
Create a Update Balance Task
Create a Knowledge Graph
Set Up a Smart Alert
Design Digital Skills
Configure Digital Forms
Configure Digital Views
Add Data to Data Tables
Update Data in Data Tables
Add Data from Digital Forms
Train the Assistant
Use Traits
Use Patterns for Intents & Entities
Manage Context Switching
Deploy the Assistant
Configure an Agent Transfer
Use Assistant Functions
Use Content Variables
Use Global Variables
Web SDK Tutorial
Widget SDK Tutorial
Analyze the Assistant
Create a Custom Dashboard
Use Custom Meta Tags in Filters

APIs & SDKs
API Reference
API Introduction
API List
API Collection
koreUtil Libraries
SDK Reference
SDK Introduction
SDK Security
SDK Registration
Web Socket Connect and RTM
Using the BotKit SDK
BotKit SDK Tutorial - Blue Prism

ADMINISTRATION
Introduction
Assistant Admin Console
Administration Dashboard
User Management
Add Users
Manage Groups
Manage Roles
Assistant Management
Enrollment
Invite Users
Send Bulk Invites
Import User Data
Synchronize Users from AD
Security & Compliance
Using Single-Sign On
Security Settings
Cloud Connector
Analytics
Billing
  1. Home
  2. Docs
  3. Virtual Assistants
  4. Intelligence
  5. Dialog Management11 min read

Dialog Management11 min read

Human conversations are characterized by twists and turns, and no two directions are ever the same. Natural conversations often tend to go beyond linear intent resolution paths defined in the bots. Ideally, a user allows the bot to take a user intent to its logical conclusion before initiating another, but that is not always the case.

Consider the following conversation for example:

It includes an entity value for the bot to proceed with the current intent, while also presenting a new user requirement or intent.

Bots must account for this by allowing users to pause a task, start and complete another task, and seamlessly return to the original task without losing important contextual data and conversation continuity.

Kore.ai provides you with granular control over hold and resume functionality at the bot, task, and node levels, and allows you to control context switching rules and behaviors. You can handle such interruptions in intent flows by providing a whole range of options to define the task-switching experience.

The Manage Interruptions option can be set up at the bot, task, and node levels to ensure the configurations are layered to suit your various business requirements. You can also add conditional exceptions between tasks with the ability to pass contextual data between them.

The following sections describe:

  • The various Manage Interruption configurations,
  • The hierarchy of the setups at bot, task, and node levels.

Manage Interruptions

The Manage Interruptions options allow you to select if and how a user can switch to another task and also the exit strategies. The configuration for Manage Interruptions is done either at the bot Level or at the task or dialog level. We will discuss the settings at the Bot Level in this section. These settings can be overridden at the task level as discussed in the Interruptions Hierarchy section.

You can access the interruption settings from the Build > Intelligence > Manage Interruptions page.

The platform provides options for Interruptions Behavior under the following three categories:

  • Allow interruptions: Provides options for you to choose for holding a task and options to resume it later.
  • Do not allow interruptions: Stops new tasks from interrupting the current task.
  • Allow the end user to select the behavior: Seeks end-user confirmation to switch to a new task and provides several resume options for you to pick from.

NOTE: When the user inputs after a session has ended, i.e. after at least 15 minutes of inactivity, and if the input is within the previous session context, it is treated as an interruption scenario. If instead, the user comes back with a new utterance, then the old context is discarded and a new conversation is started.

Allow Interruptions

Once you allow for task interruption, you need to specify the bot behavior in such scenarios as what should happen to the current task, where should the conversation go once the interruption task has been completed, etc.

Following are the options under Allow interruptions.

Option Description
Hold the current task and resume back once the new task is completed This option lets the bot switch to a new intent as soon as it is detected irrespective of the current intent flow. When you select this option, the Resume Options section appears providing further choices to define how to resume the on-hold task once the new intent task is completed.  Refer to the Resume Options section below for details.
Discard the current task and switch to new When you select this option, the bot discards the current task and switches to a new task soon after it encounters another intent. The bot sends a message to notify the user before switching to the new task and the current task would be discarded. You can customize the message from the Manage Response setting.
Switch to a new task without any notification to the user and discard the current task When you select this option, the bot discards the current task and switches to a new task soon after it encounters another intent. The bot does not notify the user about the switch and does not resume the current task later.
Continue with the current task and add a new task to the follow-up task list When you select this option, the bot continues on the current task even if it encounters a new intent. It, however, adds the new intent to the Follow-up intents array.

Do not allow interruptions

When you select this option, the bot level interruptions is turned off. However, you may override this behavior for selected tasks, linked dialogs (exceptions), or node levels. Refer to the Interruptions Hierarchy section below for more information.

Allow the end user to select the behavior

When you select this option, the bot presents the end-user with a confirmation message asking if the user wants to switch to a new task. The bot switches to the new task only if the end-user chooses to do so. You can customize the confirmation message sent to the user by clicking the Manage Response link next to the option.

You need to define a Resume option that the bot should take if the user decides to switch the task. Refer to the Resume Options section below for details.

Resume Options

Once you allow the task to be put on hold and proceed with the interruption task, you must configure the Resume option i.e. how the on-hold task should be handled once the interrupting task is completed.

 

Option Description
Get confirmation with the user before resuming an on-hold task After executing the new task, the bot gives Yes or No options to the user to return to the last on-hold task. You can write a custom message for the user by clicking the Manage Response link next to the option.
Example:

User: Hi
Bot: Hello!
User: Can you book me a flight for tomorrow?
Bot: From which city are you flying?
User: Los Angeles
Bot: Where to?
User: By the way, what’s the weather forecast for tomorrow?
Bot: Please enter the name of the city for the forecast.
User: Los Angeles
Bot: Weather Forecast for Los Angeles
Date: March 15
Temperature: 25 C
Condition: Mostly Sunny
Bot: Should I continue with the task 'Book Flight'?
User: Yes
Bot: Enter the name of the destination airport
Notify the user with a message that the on-hold task is being resumed Notifies the user with a message and resumes the last task that’s kept on hold without user confirmation.
You can customize the notification message by clicking the Manage Response link next to the option.
Example:

User: Hi
Bot: Hello!
User: Can you book me a flight for tomorrow?
Bot: From which city are you flying?
User: Los Angeles
Bot: Where to?
User: By the way, what’s the weather forecast for tomorrow?
Bot: Please enter the name of the city for the forecast.
User: Los Angeles
Bot: Weather Forecast for Los Angeles
Date: March 15
Temperature: 25 C
Condition: Mostly Sunny
Bot: Resuming interrupted task Book Flight
Bot: Enter the name the of the destination airport
Resume the on hold task without any specific message to the user Directly resumes the last task that’s kept on hold soon after the current task is executed.
Example:

User: Hi
Bot: Hello!
User: Can you book me a flight for tomorrow?
Bot: From which city are you flying?
User: Los Angeles
Bot: Where to?
User: By the way, what’s the weather forecast for tomorrow?
Bot: Please enter the name of the city for the forecast.
User: Los Angeles
Bot: Weather Forecast for Los Angeles
Date: March 15
Temperature: 25 C
Condition: Mostly Sunny
Bot: Enter the name of the destination airport
Always resume the on hold task without any specific message to the user if the task is ended in a single response Sometimes, the switched-over tasks end with a single bot response. For example, in the weather forecast example above, if the user had directly entered By the way, what’s the weather forecast for LA tomorrow? the bot would have merely responded with the forecast, with no further steps.
In such cases, if you select this option, the bot switches back to the task on hold without any confirmation or notification, regardless of the option you’ve chosen.

Note: The switching happens directly, without any notification to the end-user.

On Hold Quantity
You can select the number of tasks that you want to keep on hold from the On Hold Quantity field. The bot keeps a maximum of this number of tasks on hold and ignores the tasks once this value is reached.

For example, if you specify the On Hold Quantity as 1 and if a third intent comes into play, the bot will not allow the new task to be initiated regardless of interruptions options.

Note: Tasks always resume in reverse chronological order. So, if you kept task 2 on hold after task 1, task 2 resumes before task 1.

Interruptions Hierarchy

Apart from defining the generic interruptions options at the bot level, you can customize these options at the dialog task or dialog node level. When you set up the Hold and Resume options in more than one levels, the order of precedence of the settings works as follows:

Node Level Settings

You can customize the interruptions settings at the node level. If you customize the settings for a node, this configuration takes the highest precedence.

To customize the interruptions settings for a node:

  1. Open the dialog task and then the node to change settings.
  2. Click the instance tab and select Advanced Controls.
  3. Under the Interruptions section, select Customize for this node and make the necessary configurations. These configurations are the same as discussed above for the bot-level Interruptions Settings.

Dialog Level Settings

The dialog level Interruptions customizations have higher priority over bot level settings but rank lower than the node level customizations or any exceptions.

To customize the Interruptions settings for a dialog:

  1. Open the Dialog task you want to customize.
  2. On the top-right of the dialog builder, click the more options icon and then click Manage Interruptions.
  3. Under the Interruptions Behavior section, select Customize for this task and make the necessary configurations. These settings are the same as discussed above for the bot Level Interruptions.
Note: If you do not customize Interruptions settings at the node or task level, the bot level settings come into play.

Manage Behavior for FAQs

You can choose whether FAQs should honor the interruptions settings or if FAQs should be always responded to.

The flag Interruption Behavior for FAQs will allow you to:

  • Always identify and respond to FAQs when interruptions are enabled. This is the default setting.
  • Always identify FAQs when interruptions are enabled and proceed as per the Hold and Resume settings.

Manage Behavior for Ambiguous Intents

You can handle ambiguous intents during a conversation flow by setting the flag Behavior for Ambiguous Interruptions. It allows you to:

  • Present the ambiguous intents to the end-user and proceed as per the Hold and Resume Settings. In case the user selects a task and the interruptions setting is:
    • Continue the current task and add a new task to the follow-up task list, then the current task will be kept on hold and the new task will be initiated.
    • Allow the end-user to select the behavior, then the new task will be initiated without asking the user since the selection has been already made.
    • All other hold options will be applied as usual.
  • Do not present ambiguous intents to the end-users and continue on the current task. This is the default setting.

Manage Behavior for Small Talk

You can handle small talk during a conversation flow by setting the flag Interruption Behavior for Small Talk. It allows you to:

  • Respond to Small Talk and resume the on-hold task – this is the default setting.
  • Execute the Small Talk using Hold & Resume settings.

Manage Behavior for User Authorization

When a user gives any input at a prompt message for authorization that is not relevant, the following flow is followed:

  • If Small Talk intent is identified, then the Small Talk’s response is displayed, and the authorization is re-prompted.
  • The platform checks for the presence of any intents (tasks or FAQs) in the user utterance, irrespective of the configurations set for Manage Interruptions:
    • If the intent is identified, confirmation to discard the current task and trigger the new task is sought.
    • If multiple intents are identified, an ambiguity dialog along with the option to ignore the utterance and continue with the current task is presented.
  • If no intent is identified, then the user is re-prompted with the authorization link.
Menu