Time-aware Reasoning in Uncertain Knowledge Bases

Time information is ubiquitous on the Web, and considering temporal constraints among facts extracted from the Web is key for high-precision query answering over time-variant factual data. In this paper, we present a simple and efficient representation model for timedependent uncertainty in combination with first-order inference rules and recursive queries over RDF-like knowledge bases. In the spirit of data lineage, the intensional (i.e., rule-based) structure of query answers is reflected by Boolean formulas that capture the logical dependencies of each derived answer fact back to its extensional roots (i.e., base facts). Our approach incorporates simple weight aggregations for begin, end and during evidences for base facts, but also generalizes the common possibleworlds semantics known from probabilistic databases to histogram-like confidence distributions for derived facts. In particular, we show that adding time to the latter probabilistic setting adds only a light overhead in comparison to a time-unaware probabilistic setting.