Influencing anomaly detection with lexical keywords

  • Release version: Yokohama
  • Updated January 30, 2025
  • 1 minute to read

    • Influence how Health Log Analytics finds anomalies by managing keywords it looks for in your log data. When text in log data for a source matches a lexical keyword that exceeds a specified count threshold, the system identifies an anomaly and generates an alert.

    Health Log Analytics scans your logs for words that can indicate important issues. Lexical keywords such as 'crashed' or 'failed' signal conditions that may require attention. The application sets a threshold for each lexical keyword based on its normal occurrence pattern and frequency in your logs.

    When Health Log Analytics scans your logs, it identifies all occurrences of the keyword. If the frequency of a lexical keyword in the log data for a source exceeds the specified threshold, the system identifies an anomaly and generates an alert.

    Health Log Analytics comes with many default global keywords. You can add, edit, and delete these global keywords or phrases, which apply to all source types.
    Note:
    To add a specified keyword associated with a specific source type, see Configure source type capabilities.

    You can manage lexical keywords as follows: