Finding Intuitive Abstractions of High-Frequency Data

In this paper we describe ways to transform a curve constituted by a series of data points delivered by monitoring devices into a series of bends and lines between them. The resulting qualitative representation of the curve can easily be utilized in a Knowledge Based System. Since the representation obtained resembles the way humans describe curves, e.g. "this bend is not sharp enough to be normal", the transformation of the data described in this paper promises to facilitate knowledge acquisition in the eld of interpreting high-frequency data in the medical domain. Our analysis performs best on curves that are periodical but too irregular to be analyzed by Fast Fourier Transformation or other standard methods.