A systematic requirements engineering approach for decision support systems

Abstract

Decision Support Systems have emerged as a dominant technology capable to integrate heterogeneous sources into an analytical fashion to facilitate and provide a better decision-making process. Successful projects of Decision Support Systems implementation have confirmed a highlevel of user satisfaction and return on investment. Despite the potential of these systems, several surveys have indicated that the failure rate of Decision Support System projects in case studies and literature is considerably high. The issue starts from setting the wrong requirements by approaching Decision Support Systems as Operational Systems, without considering that their development is rather different. Particularly, Decision Support Systems are expected to: deal with diverse domain terminology causing poor communication between business users and IT people, integrate heterogeneous sources demanding complex procedures to control the integration and transformation phases, produce high-quality documentation, create analytical results expected to be explored across multiple levels (e.g. dimensions), provide a validated and verified output, provide traceability of sources, and a comprehensible MD (Multidimensional) Design is mandatory. In addition to the aforementioned characteristics, at the early stages of Decision Support System projects, typically, business needs are manually collected (e.g. interviews, etc.) to be then translated into a conceptual MD and ETL Designs; this manual process tends to be error-prone, that demands several rounds of redesign to satisfy all the business requirements stated by the stakeholders. On the basis of the evidence currently available, it seems fair to suggest that leading Information Systems professionals lack of an adequate approach to determine information needs specifically for Decision Support Systems. Inasmuch as, it was needed to define a novel, Requirements Engineering approach, tailored to Decision Support Systems: We have named it "RE4DSS" (Requirements Engineering for Decision Support Systems). Our systematic and block-oriented approach RE4DSS, is coupled with a set of models and techniques, guiding the developer throughout the activities of elicitation, negotiation and documentation to produce goals, scenarios and solution-oriented requirements, followed with the management and validation activities. As RE4DSS aims to improve the manual practices by systematizing the requirements process, while it is accompanied by a system called GEM (Generating ETL and MD Designs) to provide a semi-automatic means translating each of the stakeholders' requirements to finally produce the appropriate and unified ETL and MD Designs. To pursue our main goal, we present a profound literature classification to address the complexities of Requirements Engineering for Decision Support Systems. In closing, a real case study conducted by the Technical University of Catalonia (UPC) in collaboration with the World Health Organization (WHO) demonstrates how RE4DSS approach has been successfully applied during the requirements process to create the ETL and MD Designs of an information surveillance system to control/eliminate the neglected Chagas disease.