Information integration in research information systems

Information integration is an on-going challenge in data management and various approaches have been proposed in database research. New technologies and application areas create different requirements for integration systems. Research information management (RIM) is yet another challenge for data integration. RIM has many properties that are typical for data integration scenarios: many data sources, various modeling languages and data models, heterogeneity in syntax and semantics. Furthermore, many stakeholders are involved in RIM, usually with diverting goals. The combination of these properties makes RIM a particular difficult integration problem. In this paper, we discuss the applicability of data integration approaches to research information management. In particular, we want to highlight the lessons which have been learned in data integration in the recent years. Early approaches in data integration focused on the data models and the problems with schema integration. Recent work rather concentrates on the mappings between models and integration processes. Our main argument in this paper is that mappings should be also considered as key objects in research information systems.