Ing. Robin Obůrka

Článek

10.1016/j.dam.2024.03.014

Katedra teoretické informatiky
Katedra softwarového inženýrství

We define a tree pattern border array as a property of linearised trees analogous to border arrays from the string domain. We consider tree patterns with node wildcards and subtree wildcards. Then we use the tree pattern border array in a new forward tree pattern matching algorithm for ordered ranked trees. The algorithm finds all occurrences of a single linearised tree pattern in a linearised input tree. As with the classical Morris–Pratt algorithm for searching in strings, the tree pattern border array is used to determine shift lengths in the searching phase of the tree pattern matching algorithm. We compare this new algorithm with the best performing previously existing algorithms based on backward linearised tree pattern matching algorithms, (non-)linearised tree pattern matching algorithms using finite tree automata or stringpath matchers. We show that our new algorithm outperforms other tree pattern matching algorithms in single tree pattern matching.

Stať ve sborníku

Katedra teoretické informatiky
Katedra softwarového inženýrství

We define a tree pattern border array as a property of linearised trees analogous to border arrays from the string domain. We use it to define a new forward tree pattern matching algorithm for ordered trees, which finds all occurrences of a single given linearised tree pattern in a linearised input tree. As with the classical Morris-Pratt algorithm, the tree pattern border array is used to determine shift lengths in the matching phase of the tree pattern matching algorithm. We compare the new algorithm with the best performing previously existing algorithms based on backward linearised tree pattern matching algorithms, (non-)linearised tree pattern matching algorithms using finite tree automata or stringpath matchers. We show that the presented algorithm outperforms these for single tree pattern matching.