Temporal connectives versus explicit timestamps to query temporal databases

Abstract

Temporal databases can be queried either by query languages working directly on a timestamp representation or by languages using an implicit access to time via temporal connectives. We study the differences in expressive power between these two approaches. First, we consider temporal and first-order logic. We show that future temporal logic is strictly less powerful than past-future temporal logic and also that there are queries expressible in first-order logic with explicit timestamps that are not expressible in extended temporal logic. Our proof technique is novel and based on communication complexity. Then, we consider extensions of first-order logic with fixpoints or while-loops. Again the explicit temporal version of these languages, using timestamps, is compared with an implicit one, using instructions for moving in time. We also compare the temporal versions of the fixpoint language with those of the while language. (C) 1999 Academic Press.