The Expert Reasoning and Data Exploitation component, developed by the INGENIOUS partners from CERTH, aims at modelling the information about resources, entities and relations related with the INGENIOUS application domain as ontology patterns for the purpose of enabling a system generating early alerts.

The main software components composing the Expert Reasoning are:

  • The Semantic Knowledge Base (KB) – A semantic repository (RDF triple store) to accommodate the INGENIOUS’s knowledge. The KB should be compliant with the W3C standards, and especially with RDF, OWL 2, and SPARQL. It should provide the core infrastructure mainly for modelling agility, data publishing and integration.
  • The Knowledge Base population service – It contains the mapping algorithms that populate the INGENIOUS Knowledge Base (KB) with the main entities and their relations in the domain that are saved as Knowledge Graphs. This sub-component is responsible for semantically integrating output from other modules in the ontology. More specifically, the Fusion Engine collects this output from other modules, checks which are the available resources, gathers data from IoT systems, for various vital signs of a First Responder, such as Helmet, Uniform, Boots, Gas sensors, and forwards them to the Expert Reasoning module. Subsequently, the Expert Reasoning parses this information, namely the resources, and their corresponding measurements, and saves them in the KB as ontologies. Periodically, the Expert Reasoning system collects information by the Fusion engine in order to detect changes and update the Knowledge Graphs.
  • The Expert Reasoning framework – A middleware that processes the data sent from the Fusion Engine and produces alerts for early detection of risky situations for a First Responder. This framework provides a reasoning infrastructure for custom rule-based reasoning by running SPARQL rules upon the Knowledge Base in order to produce alerts. Rules can be as complex as combining knowledge from various IoT systems and using inference for uncovering new information out of the existing relations. For instance, a complex rule could use information from both the Uniform (vital sign: heart rate) and the Boots (immobility) for detecting an alarm of an extreme severity. Additionally, the Expert Reasoning produces alerts of different severities and urgencies, contributing, in such a way, to an alert-system offering various priority scales with the aim to also warn in less dangerous situations for better prevention.

In Figure 1, the workflow that is followed by the Expert Reasoning is depicted. The Fusion Engine collects data from IoT resources, sends them to the Expert Reasoning component which processes the data, saves them in the Knowledge Base, and performs rule-based reasoning in order to detect hazardous situations and produce alerts.

Expert Reasoning Workflow

In Figure 2, an excerpt of a knowledge graph is depicted representing the analysis of a complex alert. This alert is generated by a complex rule handling data from two components, the Uniform (Equipment_UN-128) and the Boots (BootsAlert_xxx), in order to detect that an FR is in serious danger and he needs immediate attention. As we see in the figure, the analysis for the detected alert is associated with the corresponding First Responder.

Example of a Knowledge Graph
Share this post!