Fraud detection - Waiviogit/waivio GitHub Wiki

Technology for detecting fraud attempts in reviews. The tab displays the results of the Fraud detection assistant.
Page for displaying campaigns where “dubious” photos were identified in review.

A tab is added under the Rewards / Campaigns section and is accompanied by a disclaimer stating that this is an experimental service only and should not be relied upon.

If no cases are found, then the message is displayed: "Fraud attempts have not been detected."

The selection of cards is limited to 30 days.

At the moment, the check is carried out as follows:

  1. the models of the devices on which the pictures for the review were taken are compared, if they differ, the user is suspected of fraud.
  2. if the image has gps coordinates, then it is checked that they are made within a radius of 1 km from the restaurant (provided that the object has a map)
  3. the date of creation of the image is checked - it must be no earlier than 14 days before the reservation of the company.

  4. 1st two digits - code, 2nd last random.
    Code:
    01 - metadata is missing on all photos;
    02 - metadata is missing in one of the photos;
    03 - photo resolutions differ (no triggering on vertical / horizontal);
    04 - the dates are different in the pictures;
    05 - GPS coordinates are different in the pictures;
    06 - pictures taken by devices with different IDs;
    07 - the date on the photo differs from the post date by 14 or more days;
    08 - GPS coordinates more than 1 km from the object.

    Page elements
    TITLE "Fraud detection assistant"
    DESCRIPTION "Disclaimer: It is an experimental service with a limited scope and is provided "as is" with no guarantee of applicability for the detection of probable fraud attempts. All submissions must always be manually verified and confirmed by the campaign sponsor."
    SORT Sort by
    • Reservation
    • Action (date)
    CARD secondary cards with status "Completed" are displayed
    CODES codes with the reason for suspicion of fraud are displayed, under the comment field

    image

⚠️ **GitHub.com Fallback** ⚠️