Method and system for detecting and preventing an imminent failure in a target system

Abstract

A system for detecting and preventing an imminent failure in a target system includes an interface, a processor in communication with the interface, and non-transitory computer readable media in communication with the processor. The interface receives training data items. Each item corresponds to either a service ticket or a machine-generated log that specifies an issue and a resolution to the issue. The instruction code is executed by the processor and causes the processor to group the training data items according to different categories. For each group, the processor trains a model to match the issue of each item of the group with the corresponding resolution and associates a model configuration of the trained model with the category of the group. The processor receives a sequence of new data items that include service tickets or machine generated logs for which a resolution is unknown. The processor groups the new data items according to one or more categories. For each category the processor determines one or more model configurations associated with similar categories; configures the model according to at least one of the model configurations; processes each data item of the group through the configured model; and communicates the resolution generated by the model to a user of the system to facilitate performance of remedial actions on the target system to thereby prevent an imminent failure of the target system.