Auflistung nach Autor:in "Perner, Petra"

Gerade angezeigt 1 - 4 von 4

Dissimilarity based Vector Space Embedding of Graphs using Prototype Reduction Schemes
(Springer, 2009) Riesen, Kaspar; Bunke, Horst; Perner, Petra
04B - Beitrag Konferenzschrift
Generalized graph matching for data mining and information retrieval
(Springer, 2008) Brügger, Alexandra; Bunke, Horst; Riesen, Kaspar; Perner, Petra
04B - Beitrag Konferenzschrift
How to support customer segmentation with useful cluster descriptions
(Springer, 2015) Witschel, Hans Friedrich; Loo, Simon; Riesen, Kaspar; Perner, Petra
Customer or market segmentation is an important instrument for the optimisation of marketing strategies and product portfolios. Clustering is a popular data mining technique used to support such segmentation – it groups customers into segments that share certain demographic or behavioural characteristics. In this research, we explore several automatic approaches which support an important task that starts after the actual clustering, namely capturing and labeling the “essence” of segments. We conducted an empirical study by implementing several of these approaches, applying them to a data set of customer representations and studying the way our study participants interacted with the resulting cluster representations. Major goal of the present paper is to find out which approaches exhibit the greatest ease of understanding on the one hand and which of them lead to the most correct interpretation of cluster essence on the other hand. Our results indicate that using a learned decision tree model as a cluster representation provides both good ease of understanding and correctness of drawn conclusions.
04B - Beitrag Konferenzschrift
Learning Heuristics to Reduce the Overestimation of Bipartite Graph Edit Distance Approximation
(Springer, 2015) Ferrer, Miguel; Serratosa, Francesco; Riesen, Kaspar; Perner, Petra
In data mining systems, which operate on complex data with structural relationships, graphs are often used to represent the basic objects under study. Yet, the high representational power of graphs is also accompanied by an increased complexity of the associated algorithms. Exact graph similarity or distance, for instance, can be computed in exponential time only. Recently, an algorithmic framework that allows graph dissimilarity computation in cubic time with respect to the number of nodes has been presented. This fast computation is at the expense, however, of generally overestimating the true distance. The present paper introduces six different post-processing algorithms that can be integrated in this suboptimal graph distance framework. These novel extensions aim at improving the overall distance quality while keeping the low computation time of the approximation. An experimental evaluation clearly shows that the proposed heuristics substantially reduce the overestimation in the existing approximation framework while the computation time remains remarkably low.
04B - Beitrag Konferenzschrift