Understanding Why an Alert was Generated
The Rationale tab within a Beacon alert surfaces relevant information into why an alert was generated. This may give counselors insight into how our model makes a decision about generating an alert. Sometimes when alerts are generated, the supplementing screenshot may not always appear to display any content related to suicide or self-harm.
When pages are analyzed by GoGuardian Beacon and are determined to contain self-harm related content, screenshots of the page the user is on are taken. However, many websites have additional text content outside the bounds of a screenshot either above or below where the screenshot was taken.
To aid in understanding why certain browsing content generated a Beacon alert, the Rationale tab displays screenshots and the analyzed text within the page when alerts are generated.
Screenshots of browsing activity from 15 minutes before and after alerts are generated are displayed with a corresponding timestamp. A screenshot from the source of the alert will be displayed in the center with yellow highlighting. Use the arrows on either side of the screenshots list to scroll, and click on an individual screenshot to enlarge it.
Zoomed Into Screenshot
Words on the Page
Words on the Page allows users to review all text found within the page, and highlights specific terms that are found to be related to suicide or self-harm. Users can download a PDF of the Words on the Page by clicking the Download Text button.