Query Containment for Highly Expressive Datalog Fragments [article]

Pierre Bourhis and Markus Krötzsch and Sebastian Rudolph
2014 arXiv   pre-print
The containment problem of Datalog queries is well known to be undecidable. There are, however, several Datalog fragments for which containment is known to be decidable, most notably monadic Datalog and several "regular" query languages on graphs. Monadically Defined Queries (MQs) have been introduced recently as a joint generalization of these query languages. In this paper, we study a wide range of Datalog fragments with decidable query containment and determine exact complexity results for
more » ... is problem. We generalize MQs to (Frontier-)Guarded Queries (GQs), and show that the containment problem is 3ExpTime-complete in either case, even if we allow arbitrary Datalog in the sub-query. If we focus on graph query languages, i.e., fragments of linear Datalog, then this complexity is reduced to 2ExpSpace. We also consider nested queries, which gain further expressivity by using predicates that are defined by inner queries. We show that nesting leads to an exponentially increasing hierarchy for the complexity of query containment, both in the linear and in the general case. Our results settle open problems for (nested) MQs, and they paint a comprehensive picture of the state of the art in Datalog query containment.
arXiv:1406.7801v1 fatcat:cn5rocfztfg5rhey4kgsexu5o4