A missing link in root-to-frontier tree pattern matching

Tree pattern matching (tpm) algorithms play an important role in practical applications such as compilers and XML document validation. Many tpm algorithms based on tree automata have appeared in the literature. For reasons of efficiency, these automata are preferably deterministic. Deterministic root-to-frontier tree automata (drftas) are less powerful than nondeterministic ones, and no root-to-frontier tpm algorithm using drftas has appeared so far. Hoffmann & O’Donnell [HO82] presented a root-to-frontier tpm algorithm based on an Aho-Corasick automaton recognizing tree stringpaths, but no relationship between this algorithm and algorithms using tree automata has been described. In this paper, we show that a specific drfta can be used for stringpath matching in a root-to-frontier tpm algorithm. This algorithm has not appeared in the literature before, and provides a missing link between tpm algorithms using stringpath automata and those using tree automata.