Logic, languages, and rules for web data extraction and reasoning over data

Abstract

This paper gives a short overview of specific logical approaches to data extraction, data management, and reasoning about data. In particular, we survey theoretical results and formalisms that have been obtained and used in the context of the Lixto Project at TU Wien, the DIADEM project at the University of Oxford, and the VADA project, which is currently being carried out jointly by the universities of Edinburgh, Manchester, and Oxford. We start with a formal approach to web data extraction rooted in monadic second order logic and monadic Datalog, which gave rise to the Lixto data extraction system. We then present some complexity results for monadic Datalog over trees and for XPath query evaluation. We further argue that for value creation and for ontological reasoning over data, we need existential quantifiers (or Skolem terms) in rule heads, and introduce the Datalog± family. We give an overview of important members of this family and discuss related complexity issues.