Browse Prior Art Database

A Reference Click-Through-Rate (CTR) based Offline Simulation Method for Sponsored Search Optimization

IP.com Disclosure Number: IPCOM000199152D
Publication Date: 2010-Aug-27
Document File: 4 page(s) / 220K

Publishing Venue

The IP.com Prior Art Database

Related People

Qiang Fu: INVENTOR [+3]

Abstract

The method and system for offline simulation based on a reference Click-Through-Rate (CTR) to optimize sponsored search is disclosed. The method performs an effective and efficient offline simulation to optimize various parameters and modules in a sponsored search system using a set of non-linear system metrics.

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 53% of the total text.

A Reference Click-Through-Rate (CTR) based Offline Simulation Method for Sponsored Search Optimization

Abstract

The method and system for offline simulation based on a reference Click-Through-Rate (CTR) to optimize sponsored search is disclosed.  The method performs an effective and efficient offline simulation to optimize various parameters and modules in a sponsored search system using a set of non-linear system metrics.

Description

Disclosed in a method and system for offline simulation based on a reference Click-Through-Rate (CTR) to optimize sponsored search.  The method performs an effective and efficient offline simulation to optimize various parameters and modules in a sponsored search system using a set of non-linear system metrics.  The reference CTR includes the expected CTRs for all ad positions.  The CTR is used to adjust click numbers for all the impressions.

Typically while running a sponsored search advertising system, problems associated with metrics optimization may arise.  Such metrics include click-through-rate (CTR), revenue, and so on.  In a sponsored search scenario, the click behavior is highly correlated with ad placement position.  Reference CTRs for each position are obtained.  The concept of ad position and reference CTRs are shown in Fig. 1 and Fig. 2:

 

Figure 1

Figure 2

In the above example shown in Fig. 2, DUDE = n indicates that ‘n’ ads are shown on the north.  The references CTRs can be obtained by randomly ranking ads with a small traffic.

A smoothed CTR is then calculated for any system factors known to make strong bias in CTR.  Such system factors include position, date, time, traffic source, and so on.  Accordingly, the simulation can be easily adopted in more web service systems, such as, search, contextual advertising, news, and so on.  In the sponsored search scenario, the system factors taken into consideration are DUDE and pos.

Based on the reference CTRs, the method performs offline simulation to reliably predi...