Heuristic Selection of Aggregated Temporal Data for Knowledge Discovery

Autor: Howard J. Hamilton, Dee Jay Randall
Rok vydání: 1999
Předmět:
Zdroj: Multiple Approaches to Intelligent Systems ISBN: 9783540660767
IEA/AIE
DOI: 10.1007/978-3-540-48765-4_76
Popis: We introduce techniques for heuristically ranking aggregations of data. We assume that the possible aggregations for each attribute are specified by a domain generalization graph. For temporal attributes containing dates and times, a calendar domain generalization graph is used. A generalization space is defined as the cross product of the domain generalization graphs for the attributes. Coverage filtering, direct-arc normalized correlation, and relative peak ranking are introduced for heuristically ranking the nodes in the generalization space, each of which corresponds to the original data aggregated to a specific level of granularity.
Databáze: OpenAIRE