Model-based time-distorted Contexts for efficient temporal Reasoning

Intelligent systems continuously analyze their context to autonomously take actions. Building a proper knowledge representation of the context is key to take adequate actions. This requires context models, e.g. formalized as ontologies or meta-models. As these systems evolve in dynamic contexts, reasoning processes typically need to analyze and compare the current context with its history. A common approach consists in a temporal discretization, which regularly samples the context at specific timestamps (snapshots) to keep track of history. Fig. 1 shows a context sampled at three different timestamps. Reasoning processes would then need to mine a huge amount of data, extract a relevant view, and finally analyze it. This would require lot of computational power and be time-consuming, conflicting with the near real-time response time requirements of intelligent systems. To address these issues, we define time-distorted contexts as time-aware context models. Fig. 2 shows a context representation, where the context variables belong to different timestamps. Our approach considers temporal information as first-class property crosscutting any context element, and enables building timedistorted views of a context composed by elements from different times rather than a mere stack of snapshots. We claim that these time-distorted views can efficiently empower continuous reasoning processes and outperform traditional full sampling approaches by far.