Two Phase Description Logic Reasoning for Efficient Information Retrieval

With the proliferation of knowledge intensive applications, there is a vivid research in the domain of knowledge representation. Description Logics are designed to be a convenient means for such representation task. One of their main advantages over other formalisms is a clearly defined semantics which opens the possibility to provide reasoning services with mathematical rigorousness. My PhD work is concerned with Description Logic reasoning. I am particularly interested in ABox reasoning when the available data is really large. This domain is much less explored than TBox reasoning. Nevertheless, reasoning over large ABoxes is useful for problems like web-based reasoning. I am one of the developers of the DLog data reasoner which implements a two phase reasoning: the first phase uses complex reasoning to turn the TBox into a set of simple rules, while the second phase is geared towards very fast query answering over large ABoxes. DLog currently supports the SHIQ DL language. We are trying to extend the reasoner to more expressive languages, hopefully until SROIQ, the logic behind OWL 2.
Published in 2010