Comments on "Stack-based Algorithms for Pattern Matching on DAGs"

TL;DR

This paper critically analyzes the algorithms for pattern matching on DAGs, identifying issues and proposing solutions to improve their correctness and efficiency based on experimental validation.

Contribution

It highlights discrepancies in existing algorithms, analyzes their assumptions, and offers solutions to enhance pattern matching on DAGs with experimental support.

Findings

Original algorithms miss common query solutions.

Assumption impacts algorithm correctness.

Pre-filtering affects pruning efficiency.

Abstract

The paper "Stack-based Algorithms for Pattern Matching on DAGs" generalizes the classical holistic twig join algorithms and proposes PathStackD, TwigStackD and DagStackD to respectively evaluate path, twig and DAG pattern queries on directed acyclic graphs. In this paper, we investigate the major results of that paper, pointing out several discrepancies and proposing solutions to resolving them. We show that the original algorithms do not find particular types of query solutions that are common in practice. We also analyze the effect of an underlying assumption on the correctness of the algorithms and discuss the pre-filtering process that the original work proposes to prune redundant nodes. Our experimental study on both real and synthetic data substantiates our conclusions.