Enhanced data aggregation techniques for anomaly detection and analysis
US10764312B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Dec 28, 2017 |
| Grant date | Sep 1, 2020 |
| Priority date | — |
| Expiry date | Mar 1, 2039 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06F16/23
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
An anomaly detection system that performs data aggregation against user-defined subsets of multiple variable columns within an aggregation table. Various embodiments enable users to generate rules by defining values for user-defined subsets of variables of an aggregation table that is used to deploy numerous other rules. Thus, if an aggregation table includes hundreds of columns each associated with discrete variables, a user may generate rules for aggregating data into specific variable columns of the aggregation table rather than into all of the hundreds of variable columns. The user may also define a sliding time window for the rule to be applied against and to set an occurrence threshold for the user-defined sliding time window. When a rule is triggered, the anomaly detection system may automatically generate a post-analysis table by preemptively and selectively populating the post-analysis table with values of event data indicated by a cardinality threshold and/or filter parameters.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.