Building a failure-predictive model from message sequences

A method to build a failure predictive model, the method comprising:

• receiving an input of a set of event sequences, where each sequence is labeled as representing a failure or not representing a failure;
• extracting a single predictive closed pattern from among the input sequences;
• creating a root node with the single closed pattern;
• splitting the set of event sequences into a first set that includes the single closed pattern and a second set that excludes the single pattern;
• and
• recursively processing each of the first and second sets until each process terminates with a child node labeled as representing a failure or not representing a failure.

A computer readable storage medium embodying instructions executable by a processor to perform method steps for predicting a failure, the method steps comprising instructions for:

• incorporating class information indicating whether each of a plurality of example message sequences represents a failure into a Bi-directional extension (BIDE) algorithm to generate a discriminative BIDE (BIDE-discriminative) algorithm;
• using the BIDE-discriminative algorithm to output a single pattern from the message sequences;
• building a decision tree to predict a failure based on the single pattern;
• and
• determining whether a new message sequence will cause failure using the decision tree.