Browse Prior Art Database

A Method and System for Finding Influential Photos for Targeted Advertising

IP.com Disclosure Number: IPCOM000237448D
Publication Date: 2014-Jun-18
Document File: 3 page(s) / 91K

Publishing Venue

The IP.com Prior Art Database

Related People

Liang Wang: INVENTOR [+3]

Abstract

A method and system is disclosed for computing an influence score of a photo depending on how probable the photo would be favored by a large number of people, and results in high production purchase conversion rate. The method and system thereby charge advertisers according to the influential score of the photo.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 52% of the total text.

A Method and System for Finding Influential Photos for Targeted Advertising

Abstract

A method and system is disclosed for computing an influence score of a photo depending on how probable the photo would be favored by a large number of people, and results in high production purchase conversion rate.  The method and system thereby charge advertisers according to the influential score of the photo.

Description

Recently, photo sharing websites have become popular.  People like to communicate about recently acquired products (i.e. cell phones, cars, etc.), traveled places on the photo sharing websites.  People also share opinions about the products by commenting on the photos.  Therefore, considering the user experience information presented both in the photo itself and the comments, the photo sharing website can be considered as a good place to publish advertisements.  However, the power of targeting users with advertisements has been rarely studied previously.  Because, photos are related to various topics and most of them are not product related.  So, the method and system disclosed herein finds influential photos with higher product purchase conversion rate as advertisement targeting sources.

Disclosed is a method and system for computing an influence score of a photo in a photo sharing website, depending on how probable the photo would be favored by large number of people, and results in high production purchase conversion rate.  The method and system thereby charge the advertisers according to the photo’s influential score.  Firstly, the system determines a product on which a particular photo can be related to, and whether users favor it.  Then, the system uses topic learning approaches to learn a set of sentiment words and employ phrase matching to find from comments of the photo, entity words related to certain products.

In accordance with the method and system disclosed herein an image content category of an image/photo is determined as illustrated in Figure below.  The image content category is determined based on three sources i.e. image tags, comments and the image itself.  The final image content category decision is made by a weighted combination of the three sources. The image content categories are defined according to a bazooka category.  

Figure

As illustrated in the Figure, the image tags are predefined and are assigned to the bazooka category manually (i.e. off-line).  There could be generally more than one image tags assigned, and the probability of classifying the image into a particular category is defined according to the ratio of the votes be...