Evaluation of user satisfaction with OLAP recommender systems: an application to RecoOLAP on a agricultural energetic consumption datawarehouse

OLAP and Datawarehouse (DW) systems are technologies intended to support the decision-making process, enabling the analysis of a substantial volume of data. Decision makers explore warehoused data using OLAP operators to discover new trends and/or confirm business hypotheses. In the era of Big Data, the size of warehoused data has increased substantially, and the data have become increasingly difficult to use. One of the goals of recommender systems is to help users navigate large amounts of data. OLAP recommender systems have recently been proposed in the literature because the multidimensional analysis process is often tedious because the user may not know what the forthcoming query should be. However, user satisfaction with these systems has not yet been investigated. Indeed, only time and space performances and classical information retrieval metrics (e.g., accuracy) have been studied on fictive DWs and users. Thus, this work is the first study of the usefulness of OLAP recommender systems from the decision maker’s point of view. Indeed, to the best of our knowledge, although several works have proposed OLAP recommender systems, they did not evaluate them against real-world data and users. With our experiments on a spatial DW concerning agricultural energetic consummation issued from the Energetic French Project, we prove that OLAP recommendation is useful via a real-world case study and confirm the importance of these academic tools.