We provide two tutorials for this edition of IEEE ICDIM. Our presenters have worked hard to distill the essence of their topics into a half-day (3 to 4 hours) tutorial. We hope you will enjoy both of the two tutorials.
After a general presentation of Bayesian networks (Knowledge representation formalism, inference, model design), various methods of automatic learning of Bayesian networks from data will be described. We will see thus how to approach all the data mining problematic: missing values processing and data base imputation, unsupervised learning to discover the direct probabilistic relations that hold between all the variables of the data base, supervised learning to induce the probabilistic characterization of a target variable (supervised analysis, scoring), semi-supervised learning to induce a model focused on a specific variable, unsupervised classification (clustering) of data to discover groups of lines sharing the same characteristics, unsupervised classification of the variables to discover latent variables (new concepts). Finally, we will illustrate these methods by using examples of concrete applications.
Received the Ph.D. degree in Computer Science from the Université of Rennes I, Rennes, France, in 1997. After one year dedicated to the industrialization of the results of his Ph.D. research (Fuzzy Inference System learning by Reinforcement methods – automatic pig house environment control), he received the Inov’Space Award and the medal of the town of Rennes. He then joined the ESIEA as a Professor/Researcher where he began his research on Bayesian network learning from data. He then co-founded Bayesia in 2001, a company specialized in Bayesian networks technology. He is the Managing Director and is especially in charge of BayesiaLab, the leading Bayesian networks software in Knowledge Modelling and Data mining. He applies BayesiaLab in particular in marketing, industry, health, bank, and, defence (risk management, safety & reliability, data analysis, scoring, satisfaction analysis, usage & attitude analysis, corporate & individual branding analysis, diagnostic & prognostic …)
In this tutorial, we discuss database-related XML technologies. After describing XML basics, we introduce the concept of XML-nativeness, by which we can categorize XML databases into XML native databases and XML-enabled databases. We focus on storage and query processing of each category. We also present database-related standards such as XQuery and SQL/XML. Lastly we discuss some topics on XML data mining. We keep historical aspects of the development of these technologies in mind throughout this tutorial.
Received the B.S. and Ph.D degrees in Computer Science from the University of Tokyo. After working for Fujitsu Laboratories and being a professor of Tokyo Metropolitan University, he is now a professor of Shizuoka University. His research interests include database and Web mining. He has published actively in international, refereed journals and conferences, such as ACM TODS, IEEE TKDE, VLDB, IEEE ICDE. He authored some books, which include books entitled Object-Oriented Database System (Springer-Verlag) and Next-Generation of Databases and Data Mining (CQ Publishing). He received the Sakai Memorial Distinguished Award from Information Processing Society of Japan (IPSJ) and the Director General Award from Science and Technology Agency of Japan. He was an invited professor at the Polytechnic School of the University of Nantes, France in 2003 and 2005. He is a trustee board member of the Database Society of Japan and was an editorial board member of VLDB Journal. He was also the chairman of the SIG on Database Systems of IPSJ and an editor-in-chief of IPSJ Trans. on Databases.
|
