A method, system and computer program product for detecting business anomalies. Enterprise data as well as compliance rules (e.g., business operational rules) are analyzed to identify entities and concepts to be placed in an ontology. The identified entities and concepts are represented as multi-dimensional vectors. The updates, movements and access to the enterprise data and compliance rules are tracked to determine parameters, such as information velocity, associated with at least some of the elements of the multi-dimensional vectors. The meaning from the enterprise data as well as from the data parameters is discerned. The discerned enterprise data and discerned data parameters are compared with the historical and/or projected utilization of the discerned enterprise data and discerned data parameters to identify any differences. Statistical analysis is then applied based on the enterprise data and the identified differences to generate a value corresponding to a prediction of a business anomaly.