Best Answers over Incomplete Data : Complexity and First-Order Rewritings

Amélie Gheerbrant (1, 2, 3) , Cristina Sirangelo (1, 2, 3)

1 UPCité - Université Paris Cité
2 IRIF (UMR_8243) - Institut de Recherche en Informatique Fondamentale
3 CNRS - Centre National de la Recherche Scientifique

Answering queries over incomplete data is ubiquitous in data management and in many AI applications that use query rewriting to take advantage of relational database technology. In these scenarios one lacks full information on the data but queries still need to be answered with certainty. The certainty aspect often makes query answering unfeasi-ble except for restricted classes, such as unions of conjunctive queries. In addition often there are no, or very few, certain answers, thus expensive computation is in vain. Therefore we study a relaxation of certain answers called best answers. They are defined as those answers for which there is no better one (that is, no answer true in more possible worlds). When certain answers exist the two notions coincide. We compare different ways of casting query answering as a decision problem and characterise its complexity for first-order queries, showing significant differences in the behaviour of best and certain answers. We then restrict attention to best answers for unions of conjunctive queries and produce a practical algorithm for finding them based on query rewriting techniques.

Domaines: Base de données [cs.DB]; Logique en informatique [cs.LO]; Intelligence artificielle [cs.AI]; Complexité [cs.CC]

Amélie Gheerbrant, Cristina Sirangelo. Best Answers over Incomplete Data : Complexity and First-Order Rewritings. the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI 2019), Aug 2019, Macao, China. ⟨hal-02971723⟩