Creating a knowledge base through web sync

Customers who have existing knowledge base content on their website can easily keep their chatbot's answers up-to-date using the web sync feature. This feature uses a customer's existing sitemap or allow users to upload a URL to select a target domain, import content, and customize the crawler to select relevant sections of their knowledge base. This website content can be coached just like API and manual content to improve the answers provided to end users.

Note: Aside from web sync, there are other methods in building a knowledge base, including establishing connections with CRM APIs or creating a knowledge base manually.

This article covers:

• How to create a knowledge base through web sync
  • Create a knowledge base using a sitemap
  • Create a knowledge base using a URL upload
  • Create a knowledge base using a link discovery
• Limitations of a website knowledge base

Prerequisites for creating a knowledge base through web sync

• Account owner or admin privileges; or relevant role/privilege
• Basic, Pro, Business, Education, or Enterprise account
• Zoom Virtual Agent license

How to create a knowledge base through web sync

Create a knowledge base using a sitemap

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management and then Knowledge Base.
3. Click Add Knowledge Base.
   A pop-up window will appear.
4. Under Web Sync, select Sitemap. This option allows you to sync your knowledge base using a sitemap.
   Note: The sitemap can support up to 500,000 website URLs
5. Click Connect.
6. Provide the following details:
   • Name your Knowledge Base: Enter a display name for your knowledge base.
   • Extraction Rules: You have the option to extract the following from the knowledge bases:
     • Show Images: To exclude images from being extracted and added to your knowledge base, deselect the box next to Show Images.
     • Show Videos: To exclude videos from being extracted and added to your knowledge base, deselect the box next to Show Videos.
       Note: By default, these options are selected, indicating that images and videos will be included in the extraction process.
     • Extract PDF links as articles: Select the box next to Extract PDF links as articles to create knowledge base entries from discovered .pdf links.
     • Extract .docx links as articles: Select the box next to Extract .docx links as articles to create knowledge base entries from discovered .docx links.
   • URL: Enter the URL from where you want to extract information for your knowledge base.
7. Click Continue.
8. Select the pages to sync. The number of pages to be synced will be shown at the bottom.
9. Select the languages from your knowledge base that you want to add.
10. Click Continue.
11. (Optional) Add the selector elements for your articles to help improve the accuracy. You may need to ask your web manager for help with identifying the selector elements.
    • Page Title Trimmer: Remove keywords that repeat in the page title.
    • Content Selector: Narrow down location of knowledge content. For example, enter .article-content as the selector.
    • Ignore Selector: Removes extra details such as author and dates.
    • Enable JavaScript Support: Select the box next to Enable JavaScript Support to allow the Virtual Agent to interpret and execute JavaScript code when accessing and processing web content. Once selected, configure the additional settings:
      • Page Load Delay: Set the amount of time it takes for a web page to fully load and become interactive after a user initiates a request to access it.
      • Navigation Click Selector: Identify and select specific elements or links on web pages or within documents that the chatbot can use to navigate and gather information. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Click Selector: Engage with and retrieve data from web pages or documents featuring accordion-style components. Accordions represent interface elements commonly encountered on websites or within documents, permitting users to expand or collapse sections. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Behavior: Specify whether one or multiple items can be expanded at the same time.
      • Accordion Title Selector: Expand and collapse one or more sections of text. This can be useful for organizing large amounts of information in a way that is easy to scan and navigate.
      • Dismiss Click Selector: Close an accordion section or other type of expandable content. This is useful for preventing users from having to scroll through a lot of content to find what they are looking for.
12. Click Add.

Create a knowledge base using a URL upload

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management and then Knowledge Base.
3. Click Add Knowledge Base.
   A pop-up window will appear.
4. Under Web Sync, select URL Upload. This option enables you to sync your knowledge base by uploading specific URLs.
5. Click Connect.
6. Provide the following details:
   • Name your Knowledge Base: Enter a display name for your knowledge base.
   • Extraction Rules: You have the option to extract the following from the knowledge bases:
     • Show Images: To exclude images from being extracted and added to your knowledge base, deselect the box next to Show Images.
     • Show Videos: To exclude videos from being extracted and added to your knowledge base, deselect the box next to Show Videos.
       Note: By default, these options are selected, indicating that images and videos will be included in the extraction process.
     • Extract PDF links as articles: Select the box next to Extract PDF links as articles to create knowledge base entries from discovered .pdf links.
     • Extract .docx links as articles: Select the box next to Extract .docx links as articles to create knowledge base entries from discovered .docx links.
7. Click Continue: Upload URLs
8. Select or drag a CSV file containing the URLs from where you want to extract information for your knowledge base.
9. Select the languages from your knowledge base that you want to add.
10. Click Upload and Continue.
11. (Optional) Add the selector elements for your articles to help improve the accuracy. You may need to ask your web manager for help with identifying the selector elements.
    • Page Title Trimmer: Remove keywords that repeat in the page title.
    • Content Selector: Narrow down location of knowledge content. For example, enter .article-content as the selector.
    • Ignore Selector: Removes extra details such as author and dates.
    • Enable JavaScript Support: Select the box next to Enable JavaScript Support to allow the Virtual Agent to interpret and execute JavaScript code when accessing and processing web content. Once selected, configure the additional settings:
      • Page Load Delay: Set the amount of time it takes for a web page to fully load and become interactive after a user initiates a request to access it.
      • Navigation Click Selector: Identify and select specific elements or links on web pages or within documents that the chatbot can use to navigate and gather information. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Click Selector: Engage with and retrieve data from web pages or documents featuring accordion-style components. Accordions represent interface elements commonly encountered on websites or within documents, permitting users to expand or collapse sections. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Behavior: Specify whether one or multiple items can be expanded at the same time.
      • Accordion Title Selector: Expand and collapse one or more sections of text. This can be useful for organizing large amounts of information in a way that is easy to scan and navigate.
      • Dismiss Click Selector: Close an accordion section or other type of expandable content. This is useful for preventing users from having to scroll through a lot of content to find what they are looking for.
12. Click Start Sync.

Create a knowledge base using link discovery

1. Sign in to the Zoom web portal.
2. In the navigation menu, click AI Management and then Knowledge Base.
3. Click Add Knowledge Base.
   A pop-up window will appear.
4. Under Web Sync, select Link Discovery. This option allows customers to bypass the requirement of a sitemap and ensures their data remains up to date with website changes. Users will be able to set a source URL and rules to help guide the system through their website to sync relevant pages.
5. Click Connect.
6. Provide the following details:
   • Name your Knowledge Base: Enter a display name for your knowledge base.
   • Extraction Rules: You have the option to extract the following from the knowledge bases:
     • Show Images: To exclude images from being extracted and added to your knowledge base, deselect the box next to Show Images.
     • Show Videos: To exclude videos from being extracted and added to your knowledge base, deselect the box next to Show Videos.
       Note: By default, these options are selected, indicating that images and videos will be included in the extraction process.
     • Extract PDF links as articles: Select the box next to Extract PDF links as articles to create knowledge base entries from discovered .pdf links.
     • Extract .docx links as articles: Select the box next to Extract .docx links as articles to create knowledge base entries from discovered .docx links.
   • Start URL: Enter the URL from where you want to start extracting information for your knowledge base.
7. Click Continue: URL Paths.
8. Set the URL paths to what pages you want to include in the sync:
   • Under URL Path, expand the drop-down and choose from the available conditions that best match the pages you want to include.
   • If necessary, click Add Condition to add more filtering rules.
9. Select the languages from your knowledge base that you want to add.
10. Click Continue: Content.
11. (Optional) Add the selector elements for your articles to help improve the accuracy. You may need to ask your web manager for help with identifying the selector elements.
    • Page Title Trimmer: Remove keywords that repeat in the page title.
    • Content Selector: Narrow down location of knowledge content. For example, enter .article-content as the selector.
    • Ignore Selector: Removes extra details such as author and dates.
    • Enable JavaScript Support: Select the box next to Enable JavaScript Support to allow the Virtual Agent to interpret and execute JavaScript code when accessing and processing web content. Once selected, configure the additional settings:
      • Page Load Delay: Set the amount of time it takes for a web page to fully load and become interactive after a user initiates a request to access it.
      • Navigation Click Selector: Identify and select specific elements or links on web pages or within documents that the chatbot can use to navigate and gather information. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Click Selector: Engage with and retrieve data from web pages or documents featuring accordion-style components. Accordions represent interface elements commonly encountered on websites or within documents, permitting users to expand or collapse sections. Set the amount of time it takes for a web page to fully load the content.
      • Accordion Behavior: Specify whether one or multiple items can be expanded at the same time.
      • Accordion Title Selector: Expand and collapse one or more sections of text. This can be useful for organizing large amounts of information in a way that is easy to scan and navigate.
      • Dismiss Click Selector: Close an accordion section or other type of expandable content. This is useful for preventing users from having to scroll through a lot of content to find what they are looking for.
12. Click Start Sync.

Limitations of a website knowledge base

• Web knowledge bases are limited to 5000 articles per knowledge base.
• The following CRMs will not work with web sync:
  • Zendesk: This blocks this sort of crawling but is fully supported via API connections
  • Kustomer: This requires Javascript
  • Salesforce: This requires Javascript