Prof. Takeo Higuchi
Director of Idea-Marathon System Institute, Lecturer of Osaka Institute of Technology, Electro-Communications University and Mie University

Title: Analysis of Effect and Mechanism of “Group Idea-Marathon System” with the Addendum of Creativity Proposal for Vietnam.





















  Prof. Nada Lavrac
Jozef Stefan Institute,
Department of Knowledge Technologies,
Jamova 39, 1000 Ljubljana, Slovenia.

Title:  Semantic Data Mining for Creative Knowledge Discovery

A major challenge for next generation data mining systems is creative knowledge discovery from highly diverse and distributed data/knowledge sources [1]. This talk presents a recently developed approach to information fusion and creative knowledge discovery from semantically annotated knowledge sources, where ontology information is used as background knowledge for subgroup discovery [2]. Case studies from medicine and functional genomics are used to present the lessons learned in semantic subgroup discovery [2,3]. Current directions in creative knowledge discovery through bisociative data analysis, as investigated in the European FP7 project BISON, are also outlined [4,5].


[1] Nada Lavrac, Joost Kok, Jeroen de Bruin, Vid Podpecan (eds.) Proceedings of the ECML/PKDD-08 Workshop on Third Generation Data Mining: Towards Service-oriented Knowledge Discovery, Antwerpen, September 2008 (,

[2] Igor Trajkovski, Filip Zelezny, Nada Lavrac, Jakub Tolar. Learning relational destriptions of differentially expressed gene groups. IEEE Transactions on systems, man and cybernetics, Part C, 38(1): 16-25, 2008.

[3] Igor Trajkovski, Nada Lavrac, Jakub Tolar. SEGS: Search for enriched gene sets in microarray data. Journal of biomedical informatics 41(4): 588-601, 2008.

[4] BISON: Bisociation networks for creative information discovery (

[5] Michael R. Berthold, Fabian Dill, Tobias Kötter, Kilian Thiel. Supporting creativity: Towards associative discovery of new insights. In T. Washio et al. (eds.): Proceedings of PAKDD 2008, Springer LNAI 5012, 14–25, 2008.

[4] BISON: Bisociation networks for creative information discovery (















  Prof. Chidchanok Lursinsap
Chidchanok Lursinsap Advanced Virtual and Intelligent Computing (AVIC) Center Department of Mathematics Faculty of Science Chulalongkorn University Bangkok, Thailand

Title: Fast Behavioral Learning Model for Functional Approximation and Pattern Recognition.

Abstract: The problems of functional approximation and pattern recognition can be viewed as a problem of constructing a manifold on a high dimensional space. The disadvantage of adapting the Euclidean distance sum-squared error is that the square function conceals the actual topological relation between the target and output values. It creates fluctuating phenomenon between the constructed and target manifolds. A new cost function measuring the parallel aspect between the constructed manifolds and target manifold is introduced. This function compares the natural gradients, called behavior, of both manifolds formed at various locations according to the given sequence of the training set. In addition, it is unnecessary to force the constructed manifold to fit the target manifold during the training process. The learning speed of this approach ranges from 1 to 30 times faster than the learning speed based on the classical sum-squared error when tested on several benchmarked functional approximation problems. To fit both manifolds, a shifting distance is computed and added to the constructed manifold. Some applications of this technique to simulate natural phenomena will be demonstrated.













Asso. Prof. Tru Hoang Cao

Faculty of Computer Science and Engineering,

Ho Chi Minh City University of Technology.

Title: Named Entity-Centred Information Processing

Abstract: Information on the Web is pervaded with real individuals, i.e., named entities, in the world. The content of a text is mainly defined by keywords and named entities occurring in it. Moreover, named entities have ontological features, namely, their aliases, types, and identifiers, which are hidden from their textual appearance but may be of user concern.  First, this talk presents a hybrid statistical and rule-based incremental approach to recognize ambiguous named entities. Second, it introduces a multi-vector space model to combine keywords and named entity features for document searching and clustering. Third, it presents a robust named entity-based method for translating natural language queries to conceptual graphs.