websec-score-bola-batch
Score each input row, aggregated per subject, time window, URL template and object ID parameter, against a baseline behavior profile.
The objects field holds the number of distinct objects the subject touched in the window. The threshold is exp(mean + sigma x standard deviation) - 1 over log(1 + objects) in the baseline, at least min-objects. The input row must carry the key fields of the profile, and the profile rows samples, log_objects_avg and log_objects_stddev. A row the baseline does not know, or knows from fewer than min-samples samples, is kept and judged against min-objects.
websec-score-bola-batch baseline=VALUE [min-objects=VALUE] [sigma=VALUE] [min-samples=VALUE]
- baseline=VALUE
- Required. Name of the baseline behavior profile, e.g. websec-bola-baseline-5m.
- min-objects=VALUE
- Optional. Fewest distinct objects in a window that can be judged bola. Default is 10.
- sigma=VALUE
- Optional. Standard deviations above the baseline mean, on the log scale, where the threshold lies. Default is 3.
- min-samples=VALUE
- Optional. Fewest samples a baseline row needs before its statistics are used, one per subject and time window. Default is 30.
Output fields
| Field | Type | Name | Description |
|---|---|---|---|
| objects_p99 | Double | Baseline p99 | 99th percentile of objects per window in the baseline, null when unknown |
| objects_max | Long | Baseline maximum | Largest objects per window in the baseline, null when unknown |
| samples | Long | Baseline samples | Number of samples the baseline was built from, one per subject and time window |
| log_objects_avg | Double | Baseline log mean | Mean of log(1 + objects) in the baseline, null when unknown |
| log_objects_stddev | Double | Baseline log deviation | Standard deviation of log(1 + objects) in the baseline, null when unknown |
| threshold | Long | Threshold | exp(log_objects_avg + sigma x log_objects_stddev) - 1 rounded up, at least min-objects; min-objects when the baseline does not know the row or has fewer than min-samples samples |
| z_score | Double | Z-score | (log(1 + objects) - log_objects_avg) / log_objects_stddev, null when unknown or the deviation is 0, e.g. 8.4 |
| verdict | String | Verdict | bola when objects exceeds threshold, normal otherwise |