A Framework for Knowledge-Based Temporal Interpolation

Temporal interpolation is the task of bridging gaps between time-oriented data or abstracted concepts in a context-sensitive manner. It is one of the subtasks important for solving the temporal-abstraction taskÑ abstraction of interval-based, higher-level concepts from time-stamped data. We present a knowledge-based approach to the temporal-interpolation task and discuss in detail the precise knowledge required by that approach, its theoretical foundations, and the implications of the approach. The temporal-interpolation computational mechanism we discuss relies, among other knowledge types, on a temporalpersistence model. The temporal-persistence model employs local temporal-persistence functions that are temporally bidirectional (i.e., extend a belief measure in a predicate both into the future and into the past) and global, maximal-gap temporal-persistence functions that bridge gaps between interval-based predicates. We investigate the quantitative and qualitative properties implied by both types of persistence functions. The approach has been implemented and evaluated in several domains. Our goal is to formulate temporal-abstraction knowledge so as to facilitate its acquisition, maintenance, reuse for the same task in different domains, and sharing among different applications in the same domain. 1. Temporal-Abstraction and Temporal Interpolation Time-stamped data often need to be abstracted in a context-sensitive manner into intervalÐbased concepts, meaningful for a specific domain of application and a particular task. We term this interpretation task the temporalabstraction (TA) task. An example is providing physicians or automated decision-support systems with concise, context-sensitive summaries of time-oriented patient data. A meaningful summary characterizes significant features over periods of time; a clinical example is abstraction of "2 weeks of grade-II bone-marrow toxicity in the context of therapy for complications of a bone-marrow transplantation eventÓ from time-stamped clinical data (Figure 1). A key subtask of the TA task is the temporal-interpolation task: bridging gaps between pointor interval-based temporally disjoint time-oriented predicates, to create longer intervals when data was not acquired directly (see Figure 1). Temporal interpolation requires, among other knowledge types, a measure of temporal persistence of temporal predicates (the very notion of an episode implies some form of bounded persistence). In this paper, we present a knowledge-based approach to context-specific temporal interpolation, discuss the computational mechanism which uses that ontology, and analyze its theoretical foundations and their implications for acquisition and maintenance of temporal-interpolation knowledge.