CEPtor is a high-performance event management software system that efficiently processes disparate actor-based data streams, triggering on user-driven temporal sequences and statistical anomalies of interest.
How Does It Work
CEPtor monitors various streams of time-based events (temporal data) driven by actors such as users, subscribers, or customers. A finite state machine compares the events to a set of user-defined or machine-learned rules to determine if there is a match. While the events are being streamed, CEPtor is maintaining various state and statistical information about the actor. CEPtor does this in order to pick up where the actor may have left off over time. If all criteria in a rule are met, a trigger event is initiated. The trigger can take many forms – an entry in a database, an email, an SMS message, or integration to a software application.
Detect Industrial Espionage
- CEPtor Event Streams: CEPtor tracks a user’s badge access, network login, and file accesses from three disparate systems
- Rule Set: Determine if badge access to the building occurs after 1am, a login to senior official’s account is detected, and files are then copied
- Trigger Event: Alert corporate security
Count Number of Under the Weather Shoppers
- CEPtor Event Streams: CEPtor monitors the data stream from a grocery store transaction tracking system.
- Rule Set: Maintain a weekly per shopper average of over-the-counter medical purchases for a one-year span.
- Trigger Event: Count a customer as potentially sick if their total medical purchases in a week is greater than 10 times the average medical purchase maintained by CEPtor
Detect IoT Sensor Data Anomaly
- CEPtor Event Streams: CEPtor monitors a temperature sensor reading every 100 milliseconds
- Rule Set: Determine when a sensor reading exceeds 120 degrees F for the first time in each hour
- Trigger Event: Alert machine operator
Elder Research works with clients to understand the insight they seek from their data, determine if CEPtor is a good fit for the problem, and assist the client with installation and rule development. Once accepted, the system can be maintained by Elder Research or the client.
Run Time Environment: Cloud (AWS, Azure, Google) and premised-based support, containerized for deployment.
Configuration: Web UI and XML rule definitions
Data Sources: SQL variants, Square, Stripe, Elasticsearch, Solr, and CSV or other delimited files
Data Types: Categorical, continuous
Operational Support: Continuous distribution, categorical distribution, string, transformation, combination, and date/time.