posted on 2001-06-01, 00:00authored byB Cenk Gazen, Jaime G. Carbonell, Philip J. Hayes, Chun Jin, Eugene Fink
New Hypothesis formation may be an analyst-initiated activity, an automated process
whereby a novel trend is discovered and tracked, or a hybrid one. In the hybrid case, the
system offers its discovery of novel, potentially interesting patterns for analyst review,
leading to new hypothesis being formed and tracked, or to discarding the novelty as
coincidental or uninteresting. The ARGUS project assumes the third paradigm, where a
combination of analysis of massive data – both historical and real-time streams – leads to
automated creation of potential hypothesis for analysts to consider, discard, embellish,
combine, and/or instruct ARGUS to track as a new persistent interest profile.