DeepParse mines reusable regex patterns with an LLM from few log samples and applies them via Drain to achieve 97.6% average parsing accuracy on 16 datasets, outperforming baselines and cutting anomaly detection false alarms by over 30%.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
DeepParse: Hybrid Log Parsing with LLM-Synthesized Regex Masks
DeepParse mines reusable regex patterns with an LLM from few log samples and applies them via Drain to achieve 97.6% average parsing accuracy on 16 datasets, outperforming baselines and cutting anomaly detection false alarms by over 30%.