Towards an XML-based framework for web usage mining

Abstract

Current systems for Web Usage Mining (WUM) offer graphical and interactive ways of using the implemented methods, but do not support individual multilevel or complex analyses. Based on those systems it is neither possible to perform flexible nor regular automated evaluations based on predefined evaluation schemes for offering WUM as a service. We present an extensible framework for WUM, which describes a unified WUM process and hence supports the application of any existing WUM method. The core of our framework is a XML language called Web Usage Mining Markup Language (WUMML) that provides a vocabulary for representing and querying WUM data. WUMML uses XML standards like XPath or XQuery and supports the specification of new commands for previously not integrated WUM methods. By using WUM software that implements the WUM framework, a company is able to perform WUM analysis efficiently and is - in combination with the technical expertise - capable to professionally offer WUM as service.