Last updated
Was this helpful?
Last updated
Was this helpful?
To add new web-scraped data, follow the steps below:
Open your AI Chat
Login using !login command
Click 'Get Started'
Click Commands 'Add this URL' or 'Add new URL'
Go to Studio -> NLP Manager -> Responses to ensure the data is carried over and formatted correctly.
There are two ways to update web-scraped data:
Open your AI Chat
Login using !login command
Hit 'Get Started'
Click Add this URL or Add New URL to rescrape the page with the updated information
Go to Studio -> NLP Manager -> Responses to ensure the data is carried over and formatted correctly
Locate the response you'd like to edit
Once you click on the response name, select the three dots in the right-hand corner of the window
Uncheck the box next to Content Subscription.
Click the pencil icon
Make any necessary edits
Hit Save and Publish or just Save to keep your changes in draft mode
Why Would You Manually Update Scraped Data?
Add Topic Headers (ex. Topic: Group Ticket Perks)
Update information that hasn't been updated on your website yet
Improve the chunking of data
Removing unnecessary scraped data (webpage alerts, footer text, etc.)
Adding redacted information such as emails/phone numbers when applicable
Locate the response you'd like to delete
Click the delete button (trash can icon)
Utilize the power of our web scraping tool without having to surface information publicly on your website. By creating hidden pages on your website, you create a controlled environment that serves as a great data resource for your AI Chat to train from and curate content.
We suggest making a list of the URLs you want to scrape and train in advance. Once you locate the popup on the first URL, keep your list handy and add more URLs by clicking the 'Add new URL' button. This way, you won't need to switch between pages and have complete control over the URLs you've already scraped.
Don't attempt to scrape every web page you have! Focus your scraping efforts on web pages where your customers typically learn key information such as A-Z Guides, Things to Do, Directories, etc.
Avoid scraping:
Web pages that are frequently updated (schedules, rosters, stats, etc.). In these cases, we recommend using a prewritten response.
Any third-party web pages you do not manage.
Web pages lacking rich content such as image-heavy pages, landing pages, etc.
We always recommend that you not only review results from our scraping tool within the response library but also see how responses are generated and exposed within your chat.
If the information you scraped into the dashboard is not being understood by the LLM, this may be due to poor formatting within the response. To fix this:
Find the corresponding URL labeled response in the library (usually named after the web page it was scraped from)
Click Edit
Ensure that:
There is a header description related to the data's topic in each section
All sections are separated by a 5 “-----”
No section is very short or extremely long
Note: Manual edits can be made within web-scraped data responses in the dashboard. However, if a page is rescraped, those manual edits will be overwritten.
While in the , go to Studio -> NLP Manager -> Responses
While in the , go to Studio -> NLP Manager -> Responses
Real-World Examples: ,
