Supervised Encoding of Graphs-of-Graphs for Classification and Regression Problems

This paper introduces a novel approach for processing a general class of structured information, viz., a graph of graphs structure, in which each node of the graph can be described by another graph, and each node in this graph, in turn, can be described by yet another graph, up to a finite depth. This graph of graphs description may be used as an underlying model to describe a number of naturally and artificially occurring systems, e.g. nested hypertexted documents. The approach taken is a data driven method in that it learns from a set of examples how to classify the nodes in a graph of graphs. To the best of our knowledge, this is the first time that a machine learning approach is enabled to deal with such structured problem domains. Experimental results on a relatively large scale real world problem indicate that the learning is efficient. This paper presents some preliminary results which show that the classification performance is already close to those provided by the state-of-the-art ones.