A New Tree Distance Metric for Structural Comparison of Sequences

In this paper we consider structural comparison of sequences, that is, to compare sequences not by their content but by their structure. We focus on the case where this structure can be defined by a tree and propose a new tree distance metric that capture structural similarity. This metric satisfies nonnegativity, identity, symmetry and the triangle inequality. We give algorithms to compute this metric and validate it by using it as a distance function for a clustering process of slightly modified copies of trees, outperforming an existing measure.