Evaluating systematic transactional data enrichment and reuse

Abstract

A library account-based recommender system was developed using machine learning processing over transactional data of 383,828 check-outs sourced from a large multi-unit research library. The machine learning process utilized the FP-growth algorithm [13] over the subject metadata associated with physical items that were checked-out together in the library. The purpose of this paper is to evaluate the results of systematic transactional data reuse in machine learning. The analysis herein contains a large-scale network visualization of 180,441 subject association rules and corresponding node metrics.