Hierarchical models of consumer attributes for targeting content in a privacy-preserving manner

Abstract

A system and method provide for the interpretation and augmenting of structured documents electronically delivered to an individual consumer's computer using consumer profiles developed from and maintained with information reflecting the consumer's online and offline transactions, by selecting the variable content alternatives encoded in the structured documents that most closely match the consumer's profile. The consumer profiles are logically controlled by the consumer's computer, thus providing for enhanced security over information that is personal and confidential to the consumer, yet still allowing third parties such as web sites and others electronically delivering structured documents to the consumer to have such documents customized based on the consumer's profile. The consumer profile includes hierarchical attribute vectors which encode attributes of a consumer at progressively higher levels of abstraction, and allowing for querying of any combination of abstracted data or abstracted attributes of a consumer. The consumer profiles are updated with a process that reflects the relevance of each transaction to the consumer's profile, and accommodates a decay in the influence of transactions over time. A selection process for selecting content allows for multiple items of content to be selected for sequenced display to the consumer, through a limited presentation space.