RAMSYS - A methodology for supporting rapid remote collaborative data mining projects Steve Moyle (1), Alípio Jorge (2)(3) (1) Oxford University Computing Laboratory, UK, (2) LIACC University of Porto, Rua Campo Alegre 823, 4150 Porto, Portugal. (3) Faculty of Economics, University of Porto, Portugal amjorge@liacc.up.pt, http://www.niaad.liacc.up.pt/~amjorge Abstract. In this paper we propose the basic principles and philosophy for a collaborative methodology for performing data mining work. Such a methodology allows the data mining effort to be expended at very different locations communicating via a web-based tool. The aim of the methodology is to enable information and knowledge sharing, as well as the freedom to experiment with any problem solving technique. The data mining work methodology follows and extends the CRISP-DM methodology. 1 Introduction The leading edge of European Data Mining expertise is currently spread over a number of Data Mining Units consisting of research laboratories and companies across Europe. Each one of these Data Mining Units has developed areas of specific expertise to solve particular data mining problems. Frequent on site collaboration, often required in solving data mining problems, is made difficult by geographical distance and costs. In this report, we propose both a Data Mining methodology and a system outline for remote collaborative data mining projects: RAMSYS 1 . According to the proposed methodology, each Data Mining (DM) project will be developed by a number of Data Mining Units – or nodes -- in a network of expertise. Each node represents a work site operated by an expert, an expert team, or a technical support team. All the participating nodes in the project will work as a team. Some network members form the Management Committee which has special obligations, including managing the interface with the client, defining project evaluation criteria, and performing solution selection. 1 RApid collaborative data M ining SYS tem.