Data Pool
    • 29 May 2024
    • PDF

    Data Pool

    • PDF

    Article summary

    Customize the Copilot’s Language Models (LLMs) for End Users' specific needs.



    Requirements

    • SysAdmin or AI Admin permissions

    • SysAid Copilot license

    The Data Pool is the information source for the organization's AI Chatbot. 

    SysAdmins use Data Pool data to customize the Copilot’s Language Models (LLMs) for End Users' specific needs.

    The Data Pool consists of Datasets; containers that SysAdmins add and configure to improve the Copilot’s knowledge and personalize responses to user queries.

    The Data Pool contains six types of Datasets:

    1. Service Record Data

    2. Q&A Sets

    3. Knowledge Base Articles

    4. Documents 

    5. URL

    6. SharePoint Connector

    Data Pool Page Structure

    SysAdmins can manage Datasets and their sources on the Data Pool Page.

    To access and manage the Data Pool, go to Self-Service Portal > AI Chatbot > Settings > Data Pool.

    The Data Pool Page consists of two main sections:

    • Dataset Side Panel

    A side panel that groups the account’s list of Datasets with indication of their type, as well as a button to create a new Dataset

    • Table of Sources

    The Table displays all items added to the selected Dataset. Each item can be viewed or deleted from the table through the ‘three dots’ action menu.

    The Table displays the Datasets with two sortable Columns: Status and Last Updated. 

    The Table Header contains the following chips:

    Chip

    Definition

    Sanitization

    Whether or not the data is sanitized

    Visibility

    Who can view the Dataset

    Default

    Appears automatically when the Dataset is out-of-the-box

    Notes

    • All out-of-the-box Datasets are automatically marked “Default” and “Sanitized”, but can be edited

    • URL Source Items update automatically 1x per week

    • Datasets can be enabled or disabled by selecting the ‘three dot’ action menu in the table and clicking “Enable”/Disable”

    Create a Dataset

    To create a Dataset for your Data Pool, click on:

    1. the +New Dataset button

    2. The relevant Dataset-type from the dropdown.

    This will open the Dataset creation form, as demonstrated below (according to Dataset-type)

    Out-of-the-box Datasets

    The Data Pool’s six out-of-the-box Datasets are:

    Out-of-the-box Dataset

    Dataset Type

    Incident History

    Service Record Data

    SSP Knowledge Base Articles

    Knowledge Base Articles (published to the Self-Service-Portal)

    Verified Answers

    Q&A Sets (submitted as “Verified” in the “Monitor & Fine-tune” area)

    Documents

    Documents uploaded by the SysAdmin

    Websites

    URL (public URLs) 

    SharePoint Connector

    Imported SharePoint Sites

    To create a new Dataset, click on the “plus” button at the top of the side panel.

    Service Record Datasets

    Configure the Service Record Dataset

     

    The Service Record Datasets derive historical Service Record data from a container of Service Records – enabling the Copilot to access past Service Records which may contain the solution to the current user’s queries. 

    SysAdmins can create Service Record Datasets and define the following properties: 

    Property

    Setting

    Default Value

    Title

    Name of Dataset

    (Blank)

    Time Period

    Time range of Service Records in Dataset (either by Weeks, Months, or Years)

    2 Years

    Visibility Permissions

    Service Records to include in Dataset – based on who has permissions to view them

    All Users

    Sanitized

    Whether to sanitize the PII (Personal Identifiable Information)

    Sanitized

    Configure which Service Record Fields are processed

    AI Admins can configure which Service Record Fields the AI Chatbot processes from each Service Record Dataset.

    AI Chatbots can process the content of following Service Record Fields:

    • Messages

    • External Notes

    • Chat Console

    • Solution

    • Resolution

    • Activities

    Default Fields

    All Service Record Fields are checked off by default, except for Resolution and Activities

    To reconfigure a Dataset, check/uncheck the relevant Service Record Fields from the list, and press Save.

    The Dataset is then cleared and reset based on the new configuration.

    Knowledge Base Articles

    Knowledge Base Article Datasets derive data from the AI Admin’s organizational Knowledge Base – enabling the AI Chatbot to refer to relevant Article content that may enhance the quality of the AI Chatbot responses and/or refer users to useful article links. 

    AI Admins can create Knowledge Base Article Datasets and define the following properties: 

    Property

    Setting

    Default Value

    Title

    Name of Dataset

    (Blank)

    Permissions

    Knowledge Base Articles to include in Dataset – based on who has permissions to view them

    All Users

    Import End User articles

    Whether end-user articles should be imported into the Dataset

    Selected

    Import Admin Articles

    Whether Admin articles should be imported into the Dataset

    Unselected

    Sanitization

    Whether to sanitize the PII (Personal Identifiable Information)

    Sanitized

    You can also perform the following actions for individual Articles: 

    • View Article content

    • Remove an Article

    Notes

    At least 1 of the Import methods must be selected

    Knowledge Base modifications and updates sync automatically with the Data Pool

    Q&A Set

    Q&A Set Datasets derive data from Q&A Sets created from the selected Dataset – enabling the Copilot to refer to conversations that may contain the solution to the current user’s queries. 

    SysAdmins can create Q&A Set Datasets and define the following properties:

    Property

    Setting

    Default Value

    Title

    Name of Dataset

    (Blank)

    Permissions

    Q&A Sets to include in Dataset – based on who has permissions to view them

    All Users

    Sanitized

    Whether to sanitize the PII (Personal Identifiable Information)

    Sanitized

    Documents

    Document Datasets contain Documents (.pdf, .doc, .csv, .pptx, and .xlsx files) whose content includes relevant information to topics referenced in user queries. 

    You can configure the following Settings and apply them to a Dataset’s Document sources:

    Property

    Setting

    Default Value

    Title

    Name of Dataset

    (Blank)

    Visibility Permissions

    Documents to include in Dataset – based on who has permissions to view them

    All Users

    Sanitized

    Whether to sanitize the PII (Personal Identifiable Information)

    Sanitized

    URLs

    URL Datasets contain public URLs (no authentication required) whose content and data include information and data that contribute to the AI Chatbot’s responses. 

    You can choose to enable Deep Crawling for a URL Dataset and determine Levels; which of the links nested in the provided website will be processed into the Dataset. 

    You can configure the following Settings and apply them to a Dataset’s URL sources:

    Property

    Setting

    Default Value

    Title

    Name of Dataset

    (Blank)

    Visibility Permissions

    URLs to include in Dataset – based on who has permissions to view them

    All Users

    Sanitized

    Whether to sanitize the PII (Personal Identifiable Information)

    Sanitized

    You can perform the following actions for individual URLs:

    • Add a URL

    • Delete a URL

    SharePoint Connector

    sharepoint dataset

    SharePoint Connector Datasets contain all published SharePoint Pages, SharePoint Sites (and their nested files) that AI Admins have chosen to import as a knowledge source for the AI Chatbot’s responses.

    Whitelisting Nested SharePoint Sites

    AI Admins can list selective nested sites within the SharePoint URL (separated by commas) to include (whitelist) from the SharePoint Dataset.

    If no values are added to this field, all sites within the provided URL are imported and processed into the SharePoint Dataset.

    SharePoint Datasets are auto-synced with the SharePoint site every 10 minutes – so changes made by Admins in SharePoint stay consistent as ‘one source of truth’.

    Creating a SharePoint Dataset for the Data Pool requires the AI Admin to authenticate their SharePoint account.

    This requires providing the following SharePoint Account Details:

    • Tenant ID

    • Client ID

    • Client Secret

    • Site URL

    Once all Required Fields are added, the “Confirm” button can be clicked to complete authentication and begin importing the SharePoint site into the Dataset.

    Edit a Dataset

    To edit an existing Dataset:

    1. Open the Dataset

    2. Click the three dots in the Dataset window (top right)

    This will display the Dataset creation form (shown above) — where you can make the necessary changes.

    Coming soon

    The Data Pool will soon offer a feature to limit the Dataset according to Chatbot-type (Employee or Agent).

    The Agent Chatbot is a new feature that will soon be released