Lightweight collaborative filtering method for binary-encoded data

Abstract

A lightweight method for collaborative filtering is described that processes binary encoded data. Examples of transactions that can be described in this manner are items purchased by customers or web pages visited by individuals. As with all collaborative filtering, the objective is to match a person’s records to customers with similar records. For example, based on prior purchases of a customer, one might recommend new items for purchase by examining stored records of other customers who made similar purchases. Because the data are binary (true-or-false) encoded, and not ranked preferences on a numerical scale, efficient and lightweight schemes are described for compactly storing data, computing similarities between new and stored records, and making recommendations tailored to an individual.