文摘
Ordered graphs, each of whose vertices has a unique order on edges incident to the vertex, can represent graph structured data such as Web pages, $\mbox{\TeX}$ ?sources, CAD and MAP. In this paper, in order to design computational machine learning for such data, we propose an ordered graph pattern with ordered graph structures and structured variables. We define an ordered graph language for an ordered graph pattern g as the set of all ordered graphs obtained from g by replacing structured variables in g with arbitrary ordered graphs. We present a polynomial time pattern matching algorithm for determining whether or not a given ordered graph is contained in the ordered graph language for a given ordered graph pattern. We also implement the proposed algorithm on a computer and evaluate the algorithm by reporting and discussing experimental results.