Thoughts on:

      
      "Fast Subsequence Matching in Time-Series Databases",

      Christos Faloutsos, M. Ranganathan, Yannis Manolopoulos,

      Department of Computer Science and 
		 Institute for Systems Research (ISR)
      University of Maryland at College Park, 1994

A method for indexing sequences for sub-sequence matching is presented.
Bounding boxes in feature space are used to group sequences. Feature
space is the resulting Fourier Transform of the sequences.

The example application to scientific databases looks similar to
our computer performance domain. Of course, this paper focuses on
matching subsequences, not necessarily finding association rules.
It seems directly applicable to finding associations though. It might
work well for comparing across instances.

The method presented could easily be adapted for finding events in
our time series attributes.