analyzing-text-patterns

Extract recurring log message patterns by removing variable content like IDs and preserving punctuation-based structural markers. This classifies heterogeneous telemetry data into stable templates to map log diversity and guide parser development.

Group log templates across multiple services by applying punctuation-based extraction modes to fields like body, span_name, or event_name. This maps structural patterns to their originating sources for standardized parsing.

Identify log diversity by stripping variable data such as timestamps and IDs while keeping structural punctuation intact. This approach isolates the underlying message templates, enabling faster anomaly detection and parser creation.

You can preserve digits when extracting structural templates from observability logs by using the numeric-preserving mode. A hybrid mode is also available to combine punctuation-based extraction with digit preservation for higher precision.

The limitation when analyzing text patterns in large log datasets is a processing cap of 100000 items. Pattern extraction targets specific fields such as body, span_name, or event_name to classify messages within this boundary.