In this paper, we discuss the most important aspects of METALA, a software tool for meta-learning that we have developed to perform inductive learning in a distributed and component based fashion. The distribution comes from the use of a well posed distributed application development standard as is J2EE, and the component basis comes from the methodology we developed to integrate new learning algorithms and other software utilities into the system.

Keywords: Software architecture, web usage mining, inductive learning, knowledge models, J2EE, XML.