Logs track changes in system behaviour. They typically store information about the application that generated them, the time and the user that triggered the event, and the task the system was requested to perform. Some of the system changes stored in logs are tagged as errors. Errors act as alerts for system failures. They can have originated from a series of changes preceding them.
System failures can be crashes that immediately stop the system and are easily identifiable as all as deviations from the expected output that let the system run and reveal only at completion of system tasks.
This talk presents how to effectively mining sequences of logs and provide correct predictions. The approach integrates different machine learning techniques to control for data brittleness, provide accuracy of model selection and validation, and increase robustness of classification results. We apply the proposed approach to log sequences of 25 different applications of a software system for telemetry and performance of cars.

Barbara Russo is Associate Professor at the Faculty of Computer Science of the Free University of Bozen-Bolzano, Italy. She got her PhD in Mathematics from the University of Trento, Italy.  She was visiting researcher at the Max-Planck Institut für Mathematik in Bonn, Germany. She is invited member of the International Software Engineering Network (ISERN) whose annual meetings she co-chaired in 2013 and 2014. She has more than 60 publications in international journals in mathematics and computer science including, Journal of Empirical Software Engineering and Journal of Software Systems. She has co-authored the book Adopting Open Source Software, A Practical Guide (with B. Fitzgerald, J. Kesan, M. Shaikh and G. Succi), MIT Press, 2011. She serves and has served in program committee of international conferences in empirical software engineering and software quality, like ESEM, FSE, ICSME, MSR. She is reviewer of the major international journals in the filed (e.g., Transaction of Software Engineering, journal of Empirical Software Engineering, and Journal of Systems and Software).


Download Slides