Cold-Start recommender system problem within a multidimensional data warehouse

Abstract

Data warehouses store large volumes of consolidated and historized multidimensional data for analysis and exploration by decision-makers. Exploring data is an incremental OLAP (On-Line Analytical Processing) query process for searching relevant information in a dataset. In order to ease user exploration, recommender systems are used. However when facing a new system, such recommendations do not operate anymore. This is known as the cold-start problem. In this paper, we provide recommendations to the user while facing this cold-start problem in a new system. This is done by patternizing OLAP queries. Our process is composed of four steps: patternizing queries, predicting candidate operations, computing candidate recommendations and ranking these recommendations.