Browse Prior Art Database

Method of Estimating Prices of Guaranteed Display Contracts based on Nearest Neighbor

IP.com Disclosure Number: IPCOM000200487D
Publication Date: 2010-Oct-15
Document File: 4 page(s) / 87K

Publishing Venue

The IP.com Prior Art Database

Related People

Jian Yang: INVENTOR [+5]

Abstract

A method is provided to estimate prices of guaranteed display contracts based on the booked prices of nearest neighborhood of historical contracts. The prices are estimated by selecting eligible historical contracts and weighing them based on the nearest neighborhood considering similarity in the dimension of content hierarchy and inventory size.

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.

Method of Estimating Prices of Guaranteed Display Contracts based on Nearest Neighbor

Abstract

A method is provided to estimate prices of guaranteed display contracts based on the booked prices of nearest neighborhood of historical contracts.  The prices are estimated by selecting eligible historical contracts and weighing them based on the nearest neighborhood considering similarity in the dimension of content hierarchy and inventory size.

Description

Disclosed is a method of estimating prices of guaranteed display contracts based on the booked prices of nearest neighborhood of historical contracts.

The method estimates prices of guaranteed display contracts based on nearest neighbor pricing.  The prices are estimated by selecting eligible historical contracts and weighing them based on the nearest neighborhood considering similarity in the dimension of content hierarchy and inventory size.  The method dispenses with the concept of an intermediate impression price and defines the price of a target contract to be a weighted average of contracts from a historical data set that are nearest to the target contract.  A price estimate may then be computed from the nearest neighbor contract prices using the following equation:

Here, "" is the neighborhood of contract defined by the "K" closest contracts, "" in the historical data set.  The price estimate is defined as a weighted average with weights "" and "Z" as the normalization factor.

Closeness among different contracts implies a similarity that may be applied between contracts.  The price of the contract is characterized by its specified targeting attributes.  The primary targeting dimension is a content page where an ad corresponding to the contract may be displayed.  Content in the content page is organized in a hierarchy and each node is identified by a "space id".  Other targeting dimensions that determine available impression inventory and influence the contract price are demographics such as age and gender, and geographic attributes such as country, state, and zip-code of a user.  The hypothesis is that contracts with similar targeting attributes have similar prices and are better predictors for new contract prices than a global average of all related contracts.

In an instance, the method defines two one-dimensional metrics of distance/similarity.  The first metric is based on content hierarchy and content nodes that contracts target.  For two contracts, the method compares the depth from the root of the content hierarchy and defines the distance between the two contracts as the depth difference.  For example, two contracts have zero distance if they target the same node in the content hierarchy, and distance ±1 if one contract targets the parent node of the other.  The content hierarchy and the depth relation between query and historical contracts are illu...