Temporal data mining based on temporal abstractions

Abstract

Analyzing data for support of diagnostic tasks in dynamic domains, such as medicine, plant pathology, or information and communication technology security, requires an explicit representation and consideration of the temporal semantics of the data. However, discovering temporal knowledge is a challenging task. Temporal abstraction is a common task, based on temporal reasoning, which provides an intelligent interpretation and summary of large amounts of raw data. We suggest the application of temporal data mining mainly to time intervals of temporally abstracted data, instead of to only time-stamped raw data, and discuss its potential benefits.