Domain aware explainable anomaly and drift detection for multi-variate raw data using a constraint repository

This invention describes a computer method for finding unusual patterns or changes in raw, untransformed data. It works by first figuring out what rules or expectations (constraints) apply to that data, partly by inferring them directly from the data itself and partly by looking them up in a knowledge base organized by domain. When data breaks these rules, the system flags it as an anomaly and explains why, pointing to the specific rule that was violated.