Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Lehrstuhl fr Informatik 9 Zeit: Dienstag, 15.09.09, 16:30 Uhr

Summary: Lehrstuhl für Informatik 9
Zeit: Dienstag, 15.09.09, 16:30 Uhr
Ort: Seminarraum 5056, Ahornstr. 55
Referent: Frau Prof. Dr. Myra Spiliopoulou
Otto-von-Guericke-Universität Magdeburg
Titel: Multi-Table Stream Mining
The multi-relational data mining paradigm deals with the challenge of model learning
over a set of correlated database tables. This paradigm is intended for static datasets
though, while many modern applications require the analysis of streams. It is obvious
that customer transactions in an e-shop constitute a stream; there are mature meth-
ods for mining such a stream in isolation. However, the customers themselves consti-
tute another stream that brings new objects whenever an individual becomes a cus-
tomer. The products build a further, slower stream from which objects are forgotten
when they are no longer offered for sale. Learning a model over such a "multi-table
stream" implies solving problems that do not occur in the static scenario. In this talk,
we will see solutions to these problems.
The methods we discuss deal with (a) the management of the correlated streams and
their transformation into a single stream for mining and (b) the impact of growing ob-


Source: Ábrahám, Erika - Fachgruppe Informatik, Rheinisch Westfälische Technische Hochschule Aachen (RWTH)


Collections: Computer Technologies and Information Sciences