Finding frequent subgraphs from graph structured data with geometric information and its application to lossless compression

Authors:
Yuko Itokawa;Tomoyuki Uchida;Takayoshi Shoudai;Tetsuhiro Miyahara;Yasuaki Nakamura
Affiliations:
Faculty of Human and Social Environment, Hiroshima International University, Japan;Faculty of Information Sciences, Hiroshima City University, Hiroshima, Japan;Department of Informatics, Kyushu University, Kasuga, Japan;Faculty of Information Sciences, Hiroshima City University, Hiroshima, Japan;Faculty of Information Sciences, Hiroshima City University, Hiroshima, Japan
Venue:
PAKDD'03 Proceedings of the 7th Pacific-Asia conference on Advances in knowledge discovery and data mining
Year:
2003

Citing 3
Cited 0

Graph-Based Data Mining

IEEE Intelligent Systems
Fast Algorithms for Mining Association Rules in Large Databases

VLDB '94 Proceedings of the 20th International Conference on Very Large Data Bases
Substructure discovery using minimum description length and background knowledge

Journal of Artificial Intelligence Research

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we present an effective algorithm for extracting characteristic substructures from graph structured data with geometric information, such as CAD, map data and drawing data. Moreover, as an application of our algorithm, we give a method of lossless compression for such data. First, in order to deal with graph structured data with geometric information, we give a layout graph which has the total order on all vertices. As a knowledge representation, we define a layout term graph with structured variables. Secondly, we present an algorithm for finding frequent connected subgraphs in given data. This algorithm is based on levelwise strategies like Apriori algorithm by focusing on the total order on vertices. Next, we design a method of lossless compression of graph structured data with geometric information by introducing the notion of a substitution in logic programming. In general, analyzing large graph structured data is a time consuming process. If we can reduce the number of vertices without loss of information, we can speed up such a heavy process. Finally, in order to show an effectiveness of our method, we report several experimental results.