Similarity Analysis of Time Interval Data Sets Regarding Time Shifts and Rescaling
Marc Haßler, Sabina Jeschke, Tobias Meisen · RWTH Publications (RWTH Aachen) · 2017
Comparing things like objects, tasks, texts or audio is a common task in computer science.To do so, first a definition for similarity is required.In many fields of application, common and generic distance measures like the Minkowski distance or more specific measures like Dynamic Time Warping to compare temporal sequences are already defined and used.Based on our state of knowledge, there is no applicable measurement for calculating the similarity between time interval data sets in a manlike understanding.In this paper.we present a novel method to compare time interval data sets while using an adapted distance measurement.With our approach we look at the data sets as the disjoint parts of a bigraph, such that we can use methods from graph theory.In particular, our solution provides the opportunity to take dynamic changes (like rescaling or time-shifting) into account and thus allows the comparison of real data in humanoid fashion.Hence, it allows to compare real data with e.g.scale models.