Classification of Graph Structures

View Sample PDF

Author(s): Andrzej Dominik (Warsaw University of Technology, Poland)
Copyright: 2009
Pages: 6
Source title: Encyclopedia of Data Warehousing and Mining, Second Edition
Source Author(s)/Editor(s): John Wang (Montclair State University, USA)
DOI: 10.4018/978-1-60566-010-3.ch033

Purchase

View Classification of Graph Structures on the publisher's website for pricing and purchasing information.

Abstract

Classification is a classical and fundamental data mining (machine learning) task in which individual items (objects) are divided into groups (classes) based on their features (attributes). Classification problems have been deeply researched as they have a large variety of applications. They appear in different fields of science and industry and may be solved using different algorithms and techniques: e.g. neural networks, rough sets, fuzzy sets, decision trees, etc. These methods operate on various data representations. The most popular one is information system/decision table (e.g. Dominik, & Walczak, 2006) denoted by a table where rows represent objects, columns represent attributes and every cell holds a value of the given attribute for a particular object. Sometimes it is either very difficult and/or impractical to model a real life object (e.g. road map) or phenomenon (e.g. protein interactions) by a row in decision table (vector of features). In such a cases more complex data representations are required e.g. graphs, networks. A graph is basically a set of nodes (vertices) connected by either directed or undirected edges (links). Graphs are used to model and solve a wide variety of problems including classification. Recently a huge interest in the area of graph mining can be observed (e.g. Cook, & Holder, 2006). This field of science concentrates on investigating and discovering relevant information from data represented by graphs. In this chapter, we present basic concepts, problems and methods connected with graph structures classification. We evaluate performance of the most popular and effective classifiers on two kinds of classification problems from different fields of science: computational chemistry, chemical informatics (chemical compounds classification) and information science (web documents classification).

The IRMA Community

Research IRM

Classification of Graph Structures

Purchase

Abstract

Related Content

IRMA Sponsors