Anne-Marie Vercoustre has recently joined the Axis research group at INRIA. AxIS is a pluri-disciplinary team (Artificial Intelligence, Data Mining, Data Analysis, Software Engineering). It aims at conceiving methods and tools directed by usage, for assistance with the design and the analysis of knowledge and/or information systems (IS).

I will present some of Axis research activities, focusing on Web usage mining and XML Documents mining. In both cases we are using clustering methods to infer information that is not explicitly represented in the document collection (or site). Log-based Web usage mining tries to caracterise users' behavior when browsing a site. The goal is to identify classes of behavior in order to provide assistance to those users, or to help redesigning the site if necessary.

Our work in XML Document mining was to measure the impact of using different parts of the documents (based on the XML structure) when clustering the collection. More generally we are interested in developing technology for clustering XML collection based on their content and/or structure.