CESN home
HOME > RESEARCH > Past Projects > CESN Reasoner
Semantic Data Reasoner
Machine Reasoning about Anomalous Sensor Data
We have developed a semantic reasoning system which is able to make use of domain science rules to validate and make inferences on data that is generated by a sensor newtork. This software has a variety of responses including notification, data augmentation and creation of inferred data.

The software is driven by an OWL ontology for sensors developed by CESN and is made domain-aware by an additional ontology which describes what sensor data means for concepts in that specific domain. For example, CESN has modeled ocean events as domain ontology. The software runs on top of the Jena semantic web framework. Rules are in the language of first order logic and are expressed using the Jena rule language. An example of a rule is

[model37 range check:

(?s type Model37_TemperatureSensor),
(?s hasTakenMeasurement ?m),
(?m value ?v),
lessThan(?v, -30) ->
(?s hasOutOfRangeViolation ?m) ]

This rule specifies that the model37 temperature sensor should never take a temperature measurement less that -30. This type of range validation is an easy sanity check that sensors are not malfunctioning.

An example of a domain specific rule is


(?m type TemperatureMeasurement),
(?m value ?value),
lessThan(?value, 0) ->
email('scientist@domain.edu', 'water is freezing! check the pond!') ]

This over simplified rule says that if the temperature of the pond reaches less than 0 then the pond is freezing. The implication of this rule is that an email notification is generated when the event happens.

Software Architecture of the Semantic Reasoner System

Knowledge Representation and Sensor Ontology