Section: New Results

Introduction

This year we obtained original results in our previous four research topics: a) data transformation and knowledge management in KDD, b) data mining methods, c) Web usage and Internet mining and d) document mining and information retrieval.

First on data transformation, document validation and knowledge management in KDD (cf. section  6.2 ), we started new research on summarizing data streams and on a clustering approach for reducing the size of data and on a knowledge base for ontology learning. We developed a rule based language called SeXML for XML document verification based on our CLF framework (cf. section  6.2.3 . We pursued also our research on feature selection (cf. section  6.2.2 ) and on critical edition of sanskrit texts (cf. section  6.2.6 ). We also studied from several years the use of metadata and ontologies (cf. the KM point of view in KDD) 1) for annotating global KDD processes in terms of viewpoints to support the management and the reuse of past KDD experiences (cf. the on-going PhD thesis of H. Behja, section  6.2.4 ), 2) for supporting the interpretation of extracted clusters with the definition of an ontology and an interpretation model this year (cf. the Baldé'PhD thesis defended this year [Oops!] ).

Secondly concerning data mining methods (cf. section  6.3 ), we published original results on a new partitioning dynamic clustering method (cf. section  6.3.1 ) and started research on mining solid itemsets (cf. section  6.3.7 ) and on extracting temporal gradual rules from sequential data. We pursued also our research on self organizing maps (cf. section  6.3.2 ), on functional data analysis (cf. section  6.3.3 ), on visualisation (cf. section  6.3.4 ) and on sequential pattern extraction in Data streams (cf. section  6.3.5 . Let us note the defence of Chelcea'thesis on the agglomerative 2-3 Hierarchical Clustering [Oops!] .

Thirdly on information systems data mining and more precisely on usage mining, we obtained original results in the two following topics:

• construction and analysing of evolving data summaries (cf. section  6.4.1 ),

• mining interesting periods from Web Access Logs (cf. section  6.4.2 ).

Finally we pursued our research on a method based on Ergonomics and WUM for analysing a Web site (cf. section  6.4.3 ). We published also our work (made in 2006) on a usage mining based approach for supporting ontology evolution as a book chapter  [Oops!] .

Finally on document mining and information retrieval, we started research on Entity extraction and Entity ranking, in order to validate XML-based Information Systems (IS) at a finer granularity than the one offered by the structure. We pursued our research on ontology construction from Web HTML pages (cf. section  6.5.2 ) and on semantic and conceptual context-aware Information Retrieval. Our work on entity extraction was published this year   [Oops!] as well as the one on scientific and technical watch  [Oops!] . For generic XML document mining, our previous work (described in our 2006 annual report) has also been published this year as a book chapter of a book on "Data Mining Patterns: New Methods and Applications"  [Oops!] . We were also very active in the context of INEX initiative involved in three tracks (cf. see section  9.1.3 ).