Faster algorithms for tree similarity based on compressed enumeration of bounded-sized ordered subtrees

Abstract

In this paper, we study efficient computation of tree similarity for ordered trees based on compressed subtree enumeration. The compressed subtree enumeration is a new paradigm of enumeration algorithms that enumerates all subtrees of an input tree T in the form of their compressed bit signatures. For the task of enumerating all compressed bit signatures of k-subtrees in an ordered tree T, we first present an enumeration algorithm in O(k)-delay, and then, present another enumeration algorithm in constant-delay using O(n) time preprocessing that directly outputs bit signatures. These algorithms are designed based on bit-parallel speed-up technique for signature maintenance. By experiments on real and artificial datasets, both algorithms showed approximately 22% to 36% speed-up over the algorithms without bit-parallel signature maintenance. © 2013 Springer-Verlag.