Browse Prior Art Database

Method and System for Sampling and Applying Rules for Online Advertising

IP.com Disclosure Number: IPCOM000201720D
Publication Date: 2010-Nov-19
Document File: 3 page(s) / 44K

Publishing Venue

The IP.com Prior Art Database

Related People

Sumanth Jagannath: INVENTOR [+4]

Abstract

A method and system for sampling and applying rules for an inventory forecasting system used in online advertising is disclosed.

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 54% of the total text.

Method and System for Sampling and Applying Rules for Online Advertising

Abstract

A method and system for sampling and applying rules for an inventory forecasting system used in online advertising is disclosed.

Description

Disclosed is a method and system for sampling and applying rules for an inventory forecasting system used in online advertising.

Advertisers generally require the number of available users that meet a target profile.  Since the data available for a total number of users is large, samples of data that represent the total number of users are considered for determining the number of users matching the target profile.  Thus, a typical user query retrieves a sample of population meeting certain target profile.  User data may have one or more rules defined on the population.  For example, there may be a frequency capping rule which states that a user ‘X’ should not be shown the same advertisement more than ‘y’ times in one hour.  Alternatively, the rules may pertain to geo-confidence and override factors.  Enforcement of the one or more rules is generally an expensive operation.  These rules may be applied on the total user population or the sampled user population.  Accordingly, rules can be applied for all users and thereafter be sampled to provide the requested samples to the advertisers.  This process provides a high accuracy but involves high latency.  Alternatively, the total population data may be sampled initially and then the rules may be applied.  In this case, the latency is low but the accuracy is also low.  Therefore, there is a need for a method and system for providing the sample data with low latency and high accuracy.

Table illustrates the high latency experienced when the one or more rules are enforced on the total population data.  In this case, it is assumed that a user requests for 8 samples of a certain target profile having total population of 2500.  If the one or more rules are enforced on the total population (2500) and 8 samples are returned, the results are obtained with high accuracy, but with a high latency.  Alternatively, 8 samples may be chosen from the total population and thereafter, the one or more rules may be enforced on the 8 samples.  This results in a lower latency, although with a low accuracy.  This method of applying rules on a final sampled count is termed as one-phase sampling.

 

Latency

Error Percentage

All (2500)

93.6

0

Sampled (8)

[One-Phase Sampling]

0.32

12

Table

The method and system disclosed herein utilizes a hybrid process for application of rules and for sampling data.  The hybrid process may be considered as a two-phase sampling process.  Data is initially sampled at a higher rate and thereafter, rules are applied on the samples obtained. ...