Tighter integration in dl-programs

We introduce a mechanism called lifting to share predicates between the two components of a dl-program, integrating them in a tighter way. Using lifting, one can reason about the predicates being shared both via the description logic knowledge base and via Datalogstyle rules, and the deductions one makes are automatically reflected globally on both components. This is a capability not directly present in dl-programs, since changes to the knowledge base only affect the queries where they occur. We show that lifting has nice theoretical properties, making it suitable for modular design of dl-programs. Furthermore, dlprogram processors can easily incorporate lifting as a new operator, and we have extended dlvhex to work with dl-programs with lifting.