Reference: Shahar, Y.; Tu, S. W.; & Musen, M. A. Knowledge Acquisition for Temporal-Abstraction Mechanisms. 1992.
Abstract: We describe three general temporal-abstraction mechanisms needed for managing time-stamped data: simple temporal abstraction (a mechanism for abstracting several parameter values into one class); temporal inference (a mechanism for inferring sound logical conclusions over a single interval or two meeting intervals); and temporal interpolation (a mechanism for bridging non-meeting temporal intervals). Developers must acquire the knowledge necessary to instantiate these temporal abstraction mechanisms in any specific domain and task when constructing a knowledge-based system that requires management of temporally oriented records. The knowledge required by these mechanisms includes look-up tables (mapping parameter-value ranges into discrete classes), parameter distributions, parameter inferential properties, abstraction-inference tables (to concatenate abstracted meeting intervals), and abstraction-interpolation tables (to join points into abstracted- intervals, and non-meeting abstracted-intervals into longer, abstracted intervals). Higher-level temporal abstractions need domain-specific rules. Making explicit the knowledge required for temporal abstractions supports the acquisition of problem-solving knowledge needed for planning, assists in problem identification, and enables off-line asynchronous reasoning about the data, independent of work sessions with the application user. It also supports recognition of plan-execution problems and formulation of plan revisions, a necessary prerequisite for modifying a plan while preserving original goals. These mechanisms will be used on the context of our ongoing PROTEGE II project, whose goal is to generate automatically knowledge-based medical decision-support systems, as well as the appropriate knowledge- acquisition tools, custom-tailored to acquire the specific domain and task knowledge needed by the specific problem-solving method