LogVM: Variable semantics miner for log messages
Publication Type
Conference Proceeding Article
Publication Date
11-2022
Abstract
Modern automated log analytics rely on log events without paying attention to variables. However, variables, such as the return code (e.g., “404”) in logs, are noteworthy for their specific semantics of system running status. To unlock the critical bottleneck of mining such semantics from log messages, this study proposes LogVM with three components: (1) an encoder to capture the context information; (2) a pair matcher to resolve variable semantics; and (3) a word scorer to disambiguate different semantic roles. The experiments over seven widely-used software systems demonstrate that Log Vm can derive rich semantics from log messages. We believe such uncovered variable semantics can facilitate downstream applications for system maintainers.
Discipline
Software Engineering
Research Areas
Intelligent Systems and Optimization
Publication
Proceedings of the 2022 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), Charlotte, NC, USA, October 31 - November 3
First Page
124
Last Page
125
Identifier
10.1109/ISSREW55968.2022.00053
Publisher
IEEE Computer Society
City or Country
Los Alamitos, CA
Citation
HUO, Yintong; SU, Yuxin; and LYU, Michael.
LogVM: Variable semantics miner for log messages. (2022). Proceedings of the 2022 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), Charlotte, NC, USA, October 31 - November 3. 124-125.
Available at: https://ink.library.smu.edu.sg/sis_research/10822
Additional URL
https://doi.org/10.1109/ISSREW55968.2022.00053