Section: New Results

Keywords : KDD, preprocessing, data transformation, metadata, knowledge management, viewpoint, ontology, annotation, reusability, distances, dissimilarities.

Introduction

This year we obtained original results as previous years in our three research topics: data transformation and knowledge management, data mining and web usage mining methods. Let us note first results in document mining (cf. the content and structure point of view of an IS) and in data stream mining (cf. the usage point of view of an IS).

In section 6.2 , we describe new results on data transformation and knowledge representation. For the latter, we have studied the use of metadata (cf. the KM point of view), in particular in two ongoing PhD thesis related semantic web and KDD, conducted by H. Behja and A. Baldé. Metadata have been used or annotating global KDD processes in terms of viewpoints to support the management and the reuse of past KDD experiences (cf. section  6.2.6 ), 2) for supporting the interpretation of extracted clusters. Moreover this year we have proposed and studied new distances and dissimilarities in various applicative contexts: XML Sanskrit documents (section  6.2.2 ), tourist itineraries (section  6.2.3 ) and Web navigations and content (cf section  6.2.1 ).

On data mining methods (cf. section  6.3 ), we published new results on self organizing maps (cf. section  6.3.1 ), on functional data analysis (cf. section  6.3.2 ), on a new partitioning dynamic clustering method (cf. section  6.3.3 ) and on an agglomerative 2-3 Hierarchical Clustering in the context of Chelcea'PhD thesis (cf. section  6.3.4 ). This year we started a new research topic related to KDD in the context of data streams (cf. section  6.3.5 ).

Finally on information systems data mining, we started this year to work on visualization problems and we proposed different representations of the organization of a web site based on usage data (cf. section  6.4.1 ). We also obtained our first results on XML document mining and XML search: we studied content and/or structure mining for clustering or classifying XML documents (cf. sections  6.5.1 , 6.5.2 ) as well as the improvement of the relevance in XML search (cf. section  6.5.3 ). More classically we pursued our researches on intersites web usage mining in the context of the ECML/PKDD 2005 discovery Challenge (cf. section  6.4.2 ) and in extracting dense periods of sequential patterns (cf. section  6.4.3 ).