Currently a plethora of heterogeneous, standalone or web-enabled applications exist providing various functionalities that could be exploited in innumerable contexts for the developmentof personalized agent-basedsolutions for the end user. The integration of these applications in a standard and seamless way to enable content-rich services for the end-user is not generally feasible. The main reason for this is that these app- cations are by nature heterogeneous, developed for different development plat-forms, using different software developmenttechnologies. In this paper we present a reference architecture and support tools designed to address the problem of seamless integration ofheterogeneoussoftwareapplicationsthroughdata mining(DM) onweb service(WS) data ("web service mining") in order to enhance personalization, pervasiveness and - ?ciency on behalf of agent-based end-user applications. 1 Work presented in this paper is part of a European funded project called OASIS , whose main objective is the implementation of an ontology-driven, open reference - chitecture, which will enable and facilitate interoperability, seamless connectivity and sharing of content between different services and ontologiesin application domains for the elderly and beyond. OASIS promotes new ways to integrate all supported appli- tions into a common environment that enables access to information and content from the existing applications through WS-based software interfaces and content delivery to end-user in a pervasive manner through multi-agent applications. The achievement of this objectiveforms the main motivationof our work that results in a WS mining fra- work for the delivery of personalizedservices to the elderly users througha multi-agent system (MAS).