Systems and methods for self-adaptive episode mining under the threshold using delay estimation and temporal division

Patent metadata

Field	Value
Publication number	US-8965830-B2
Application number	US-201213474083-A
Country	US
Kind code	B2
Filing date	May 17, 2012
Priority date	May 17, 2012
Publication date	Feb 24, 2015
Grant date	Feb 24, 2015

Abstract

Embodiments relate to systems and methods for self-adaptive episode mining under time threshold using delay estimation and temporal division. An episode mining engine can analyze a set of episodes captured from a set of network resources to detect all sequences of user-specified frequency within a supplied runtime budget or time threshold. The engine can achieve desired levels of completeness in the results by mining the input log file in multiple stages or steps, each having successively longer lengths of event sequences. After completion of each stage, the engine calculates a remaining amount of runtime budget, and updates the amount of time to be allocated for each of the remaining stages up to a generated maximum stage (or sequence length). The engine thus corrects the estimated remaining time in the runtime budget (or threshold) after each stage, and continues to the next stage until the runtime budget is consumed.

First claim

What is claimed is:

1. A method of mining data, comprising:
accessing a log file comprising a set of candidate episodes each comprising a sequence of events;
receiving a runtime budget from a user for detection of a set of detected episodes from the set of candidate episodes using a set of pattern criteria;
generating, based on the runtime budget and the set of pattern criteria, the set of detected episodes from the set of candidate episodes;
generating a set of episode stag…

Assignees

• Jung Gueyoung
• Gnanasambandam Shanmuga-Nathan
• Xerox Corp

Inventors

• Jung Gueyoung
• Gnanasambandam Shanmuga-Nathan

Classifications

• G06F16/2465 Primary

  Physics · mapped topic