Evaluating First-Order Formulas in Structured Graphs (Invited Talk)

Abstract

A central problem in database theory concerns the complexity of the query evaluation problem, also called the model-checking problem in finite model theory: the problem of evaluating a given formula in a given structure. Here, I will focus on formulas of first-order logic, and the data complexity (or parameterized complexity) of their evaluation. Leveraging tools from structural graph theory, I will assume that the input structure is a graph which comes from a fixed class of well-structured graphs, such as the class of planar graphs, classes of bounded treewidth or clique-width, or much more general "tame" graph classes, such as the nowhere dense graph classes introduced by Ossona de Mendez and Nešetřil, or classes of bounded twin-width studied by Bonnet, Thomassé, and coauthors.
I will survey the recent progress in this area, which connects tools from structural graph theory, from model theory - such as stability and dependence - and from statistical learning theory and computational geometry - such as VC-dimension and ε-nets.