Automated problem triage using artificial intelligence systems

Abstract

A search is conducted to identify database defect reports describing similar defects to a proposed defect description. The proposed defect description is analyzed to determine whether the proposed defect description includes an amount of information sufficient to describe a corresponding problem based on historical defect data. A synthetic updated defect description is generated based on the latest updated defect description or the proposed defect description in response to the latest updated defect description or the proposed defect description not including a sufficient amount of information. An evaluation to identify high-level content differences between the synthetic updated defect description and one of the latest updated defect description and the proposed defect description is performed. Prompts describing suggestions of additional information to provide in the final defect report are generated based on the high-level content differences. A user is prompted to update the proposed defect description or the latest updated defect description.