Transcript and Visual Compliance Scan for Video Approval Requests

When end users create posts that include video content and submit them for review in Yext, they will be transcribed and scanned. Compliance admins will be alerted if a keyword or visual object is detected in the video itself in the Supervision Queue.

To learn more, click on your desired section and/or review the video.

Enabling Intelligent Pre-Approval Review

Reviewing Video Approval Requests

Video Supervision Scan Process

Speech and Text Lexicon Check

Disclosure

Enabling Intelligent Pre-Approval Review

Admins will need to first configure this setting to allow for Intelligent Pre-Approval Review for Video Approval Requests. Follow the steps below to enable this setting.

1. From the Administration view, select Settings > Organization Settings. Then click Review & Archive tab.

2. Scroll down until you locate the field Intelligent Pre-Approval Review. Switch the setting to Enabled.

Reviewing Video Approval Requests

1. Select the Video Approval Request from your Supervision Queue (Supervision > Dashboard).

2. In the Pending Post section, you’ll notice the Post Copy along with a Visual, Thumbnail, and Transcript tab appear at the top of the submitted video.

• The Visual tab notifies you of problematic images, like alcohol or nudity, as well as on-screen text that violates your lexicon, like words in a news headline.  If there is a lexicon keyword detected in the video, it will be highlighted in red, under Visual Recognition Alerts and indicate the frame the alert appears in. 

• The Transcript tab shows the transcript along with the lexicon keyword that is triggered. Click Download Transcript to download the transcript for the submitted video.

3. Use the buttons at the top or bottom of the page to manage the video approval request according to your organization's compliance policies.

Video Supervision Scan Process

The visual content of the video is analyzed using a machine-learning algorithm to detect inappropriate content. If any inappropriate content is detected, then specific timestamps for the violations are noted and the specific content category violation is also identified. This information regarding the inappropriate content is then sent to administrators for review. Our algorithm scans for the following inappropriate content categories:

1. Explicit Nudity:
   1. Nudity
   2. Graphic Male Nudity
   3. Graphic Female Nudity
   4. Sexual Activity
   5. Illustrated Explicit Nudity
   6. Adult Toys
2. Violence:
   1. Graphic Violence Or Gore
   2. Physical Violence
   3. Weapon Violence
   4. Weapons
   5. Self Injury
3. Visually Disturbing:
   1. Emaciated Bodies
   2. Corpses
   3. Hanging
4. Rude Gestures
   1. Middle Finger
5. Alcohol:
   1. Drinking
   2. Alcoholic Beverages
6. Hate Symbols:
   1. Nazi Party
   2. White Supremacy
   3. Extremist

Speech and Text Lexicon Check

Any speech in the video is digitally transcribed to text. Then a lexicon check is performed on the transcribed content, which detects compliance violations, profanity, or personally identifiable information. Administrators can also add custom keywords to the lexicon check to catch any idiosyncratic violations pertinent to their organization.

A lexicon check is also performed on video body texts (such as attached subtitles), video titles, and caption files.

On-screen text that may appear in the video, such as news headlines, is also subjected to a lexicon check. 

Yext generates alerts or sends violations detected in the lexicon check for pre-approval. The lexicon violation and text-detection warnings are also displayed in the Advisor/Agent Publisher when they try to publish a video post.

Disclosure

The intention for these supervision features is not to replace human-led supervision. Every organization’s compliance team is still responsible for approving or denying video posts.