Evaluating Business Intelligence Platforms: a case study Carlo DELL’AQUILA, Francesco DI TRIA, Ezio LEFONS, and Filippo TANGORRA Dipartimento di Informatica Università di Bari via Orabona 4, 70125 Bari ITALY Abstract: The paper examines some common platforms supporting Business Intelligence activities in order to state evaluation criteria for the system choice. The evaluation considers a software measurement method based on the analysis of the functional complexity of the platforms. The study has been performed on an academic warehouse that uses historical data available in legacy databases. Experimental results are reported which show the advantages and the drawbacks of each considered system. Key-Words: - Data warehouse, Data mart, OLAP system, Functional size measurement. 1. Introduction Modern Business Intelligence (BI) technologies are tightly integrated, easily, and widely deployed and usable for they are based on prepackaged application solutions [1]. For these reasons, Business Intelligence has become so much easy to justify relevant investments and the cost for developing and maintaining a data warehouse has significantly decreased. Traditional users of data warehouses are banks, financial services, or chains of supermarkets; instead, Institutional Organizations (e.g. Academies) in the past were not interested to collect and store large amount of data to use for strategic decision making. Now the trend is reversed: nowadays, we can consider the management of a University as critical as the management of a big business company, because the factors affecting an optimal management of a University are the same involved in the business processes [2]. It is clear that the development of an academic data warehouse can provide a lot of benefits, as these databases represent the source of knowledge for the researchers and for the Academic Decision Makers. However, without an effective Business Intelligence System that allows users to extract vital information, the data often go underutilized; in this case, if there is a very large collection of data to manage and there is an effective and competitive IT-competence, a Business Intelligence solution can help academic staff to ask questions that are impractical in a traditional way [3]. In order to implement a Business Intelligence solution in different business contexts and to maximize the benefits that end users can obtain, its technologies must be organized. The techonology must be deployed within an infrastructure with the capabilities to implement the Business Intelligence process that has been described in this paper and to support the range of applications best suited to every user of every type [4]; this infrastructure is called Business Intelligence Platform. These tools are software designed to support access to all forms of business information, not only the data stored in the data warehouse [1]. In fact, an effective business intelligence tool must be able to access quality information from a variety of sources stored in different forms, even in unstructured forms; in these cases, vertical collection-building and metasearching methods are necessary [5]. It is evident that BI platforms are very different among themselves as concerns performances and features; in spite of everything that, only in the April 2007, Gartner [6] has formalized standard criteria for executing an evaluation and a comparison of the technical and functional characteristics that must be owned by these software tools. The platforms are the client side components of the Business Intelligence Architecture; the server side is called OLAP Server and it is always a subsystem embedded in modern DBMSs, that usually are able to integrate MOLAP and ROLAP technologies [7, 8]. It is possible to find a complete checklist for evaluating OLAP Servers in [9, 10], since this topic is outside the scope of this paper. In particular, the aims of this paper are: (a) performing an effective comparison 7th WSEAS Int. Conf. on ARTIFICIAL INTELLIGENCE, KNOWLEDGE ENGINEERING and DATA BASES (AIKED'08), University of Cambridge, UK, Feb 20-22, 2008 ISSN: 1790-5109 Page 558 ISBN: 978-960-6766-41-1