Browse Prior Art Database

Method and System for Predicting Ad Properties using Visual Features Associated with Ads

IP.com Disclosure Number: IPCOM000241867D
Publication Date: 2015-Jun-05
Document File: 1 page(s) / 16K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for predicting Ad properties of an ad using visual features associated with the ad. These ads properties are, but not limited to, CTR (click-through-rate) and Bounce Rate (short surf time). The system and method uses visual features of an ad such appearance of the landing page on smart-phones or creative image directly to train a predictor for predicting its bounce rate and CTR. Thereafter, ad quality can be measured using the bounce rate and CTR.

This text was extracted from a Microsoft Word document.
This is the abbreviated version, containing approximately 59% of the total text.

Method and System for Predicting Ad Properties using Visual Features Associated with Ads

Abstract

A method and system is disclosed for predicting Ad properties of an ad using visual features associated with the ad. These ads properties are, but not limited to, CTR (click-through-rate) and Bounce Rate (short surf time).   The system and method uses visual features of an ad such appearance of the landing page on smart-phones or creative image directly to train a predictor for predicting its bounce rate and CTR.  Thereafter, ad quality can be measured using the bounce rate and CTR.

Description

Disclosed is a method and system for predicting Ad properties of an ad using visual features associated with the ad. These ad properties are, but not limited to, CTR (click-through-rate) and  Bounce Rate (short surf time).   Predicting ad properties are very critical for advertising platforms to assess quality of ads and find the right target audience.

Existing methods for capturing an accurate estimation of bounce rate requires running an ad campaign for a period of time in order to collect enough past events of short clicks.  When a new ad campaign is launched, there is no sufficient short click feedback to train a reliable machine learning model to perform bounce rate prediction, which is commonly known as the cold start problem.  Therefore, users might have bad experience for the reason of seeing many low quality ads served by new campaigns.

The system and method uses visual f...