Description

Early approaches to KDD typically relied on one-size-fits-all solutions. More recently, however, the role of available prior knowledge as well as specific profile of the user have been increasingly taken into account. Such contextual information may help select the suitable data, prune the space of hypothesis and represent the output in a most comprehensible way. Ontological grounding is a pre-requisite for efficient automated usage of such information with respect to a particular mining session. Notably, availability of domain ontologies also enables to automatically expose the mining results on the semantic web, to provide some KDD tools in the form of (semantic) web services, or to handle heterogeneous and complex objects when mining web data for the purpose of (semantic) web personalisation.

In some domains large bodies of consensual knowledge already exist. This is the case in medicine: although e.g. UMLS or Foundational Model of Anatomy are not ideal ontologies (i.e. formal theories) in the strictly logical sense, they express large-scale and long-term pragmatic structuring of domain knowledge. In many other domains, however, it might be necessary to start from a collection of data (esp. text) and to design the first version of ontology inductively, via ontology learning. The result of ontology learning (possibly fine-tuned by the human designer) eventually becomes input for further empirical discovery.

The workshops follows up with the successful KDO-04 workshop in Pisa (at ECML/PKDD 2004), as well as with (semantic) web mining workshops at ECML/PKDD 2001-2003. It aims to bring together researchers and practiotioners interested in the intersection of two popular areas: knowledge discovery and applied ontology.

Areas of Interest

Submissions are invited on all aspects of the interaction between knowledge discovery and applied ontology:

Use of ontologies, taxonomies and other forms of prior knowledge

Data, text and web mining techniques for

Closed loop between "ontologies for KDD" and "KDD for ontologies"

Other techniques addressing problems arising in the intersection of KDD and

Architectures, tools and experimental evaluation for the above

Applications of the above, in particular in

Workshop Schedule and Publication

The workshop will consist of an invited lecture, thematic sessions of (long and short) presentations, and a wrap-up discussion. There is also a thematical association with tutorial on Ontology Learning and Population from Text. There are also links to the various ECML/PKDD events on Web Mining / Web Personalization.

In addition to the ECML/PKDD workshop proceedings, it is intended to publish a selection of accepted papers in a Springer volume of the 'Hot Topics' series.