A Hybrid Approach for Data Mart Schema Design from NL-OLAP Requirements

OLAP systems remain difficult to develop for two primary reasons. The first reason is that the REQ analysis step is typically overlooked [1]. The second reason is that, even when attention is accorded to this step, REQ are described from a too technical perspective for decision makers to understand and validate [2]. However, the lack of user involvement in the specification can easily lead to the failure of the project. To ensure the involvement of decision makers in REQ specification, two solutions are offered: non technical notations for OLAP REQ expression (cf., tabular format [3]) and approaches to assist decision makers in identifying their REQ (cf., goal-driven approaches with either a graphical notation [1] or SQL scenarios [4], reuse of MultiDimensional (MD) patterns representing generic OLAP REQ [5]). Despite the assistance they offer, the proposed REQ notations and specification approaches presume that the decision makers are familiar with the MD concepts and their relationships. Within this context, our proposed solution (Fig.1) aims at facilitating the involvement of decision makers in the REQ specification step by offering a NL based notation for expressing the OLAP REQ, and a validation approach to ensure that the specified REQ are realizable. As illustrated in Fig.1, our approach relies on a template for OLAP REQ identification [6]. This template is composed of elements derived from the decision making process to identify the analyzed business process, the relevant actors, the indicators formulas, and the analytical queries used for decision making.