Although many QoS management architectures have been recently introduced with a lot of advanced features, they have never been widely used in the existing applications due to the lack of interoperation between providers and users, or between network operators. One of the main issues is the heterogeneity of QoS information coming from different sources: clients, communication networks, servers, data .etc. In the context of Quality-Driven Delivery (QDD) referring to the ability of a system to deliver data objects while considering the end-users expectations, all components of a distributed multimedia system have to contribute to satisfy users requirements. The mapping activity is therefore essential for dealing with the variety of QoS information of these components. In this paper, we propose an approach aimed at creating QoS mapping rules using statistical data analysis and data mining techniques combined with monitoring tools. The automatic generation of QoS mapping rules allows adapt...