An Empirical Study of Recommendations in OLAP Reporting Tool

This paper presents the results of the experimental study that was performed in laboratory settings in the context of the OLAP reporting tool developed and put to operation at the University. The study was targeted to explore which of the modes for generating recommendations in the OLAP reporting tool has a deeper impact on users (i.e. produces more accurate recommendations). Each of the modes of the recommendation component – report structure, user activity, and semantic – employs a separate content-based method that takes advantage of OLAP schema metadata and aggregate functions. Gained data are assessed (i) quantitatively by means of the precision/recall and other metrics from the log-table analysis and certain statistical tools, and (ii) qualitatively by means of the user survey and feedback given in a free form.