Log message template identification aims to convert raw logs containing free-formed log messages into structured logs to be processed by automated log-based analysis, such as anomaly detection and model inference. While many techniques have been proposed in the literature, only two recent studies provide a comprehensive evaluation and comparison of the techniques using an established benchmark composed of real-world logs. Nevertheless, we argue that both studies have the following issues: (1) they used different accuracy metrics without comparison between them, (2) some ground-truth (oracle) templates are incorrect, and (3) the accuracy evaluation results do not provide any information regarding incorrectly identified templates.

In this paper, we address the above issues by providing three guidelines for assessing the accuracy of log template identification techniques: (1) use appropriate accuracy metrics, (2) perform oracle template correction, and (3) perform analysis of incorrect templates. We then assess the application of such guidelines through a comprehensive evaluation of 14 existing template identification techniques on the established benchmark logs. Results show very different insights than existing studies and in particular a much less optimistic outlook on existing techniques.