Automated Knowledge Graph Construction From Raw Log Data

Logs are a crucial source of information to diagnose the health and status of systems, but their manual investigation typically does not scale well and often leads to a lack of awareness and incom- plete transparency about issues. To tackle this challenge, we introduce SLOGERT, a flexible framework and workflow for automated construc- tion of knowledge graphs from arbitrary raw log messages. To this end, we combine a variety of techniques to facilitate a knowledge-based ap- proach to log analysis.