Patent attributes
A method, system and computer program product for detecting software build errors. A classification system is created that identifies users' questions in crowdsource data pertaining to errors in computer programs that are associated with a log report. A model is built to classify log data as bug-related or not bug-related based on the classification system. Log reports from log data obtained from crowdsource data are identified as being bug-related based on the model. After vectorizing such log reports and storing the vectorized log reports, the language of a new build log report for a software product is vectorized upon completion of the build of the software product. If the vectorized log report is within a threshold amount of distance to a stored vectorized log report, then a copy of the log report (bug-related) and a source of the log report associated with the stored vectorized log report is provided.
