Nicky van Oorschot

Industry

Linking National Core Registries

In the Netherlands several core registries are available as open data. Several others are closed data. The 25 safety regions (Fire-, Police- and Health Organizations), decided to connect all administrative links between the registries in a new dataset. Instead of creating a new data silo, linked data is used to create the links between the registries. Other semantic web techniques like SHACL will be used to do the interpretation of subjects. For instance, if a building has a permit for LPG and a certain fuel company is located in this building it is a fuel station.  Registries used in this projects are for example the chamber of commerce,  cadaster data, national daycares and permit registries. In the current situation the safety regions do not use this data often, since it is not accessible and interlinked. Because of this interlinking using linked data, emergency responders from the safety regions are able to find critical objects nearby an incident or discover a incident is happing in a critical object because of certain chemicals for instance.

Relations within geo information are found by overlaying the geo layers and find all object which are on the same position. Finding related objects in such way results in lacking semantics on the relations. Objects are on the same location but it is unclear what their relation is and free for interpretation. Besides the relation is unclear, the necessary calculation for matching the objects from different layers is a costly job, performance wise.
Using linked data for creating administrative links is a new approach for the safety regions. It is a flexible approach and it prevents the safety regions for doing the calculation over and over if a new dataset is available (some are almost real-time).
Having a large amount of national datasets is very interesting for someone doing analytics. Besides that we are able to provide information about buildings in a certain area during a incident. Another way of using this data is for prevention activities, which critical objects are near each other and if something happens around them are the emergency responders able to solve it adequately?
One of the innovative techniques used in this project is SHACL. SHACL is a new proposed semantic web standard proposed by the W3C. Using SHACL during the project we are able to describe the per-conditions of objects and find automatically whether an object is compliant with a certain shape.  
During the project we learned that using a subset of data from the national registries results in a very large amount of data. Therefore, using just SPARQL and query all data is not feasible. To tackle this problem we have changed the concept and created a RDF dataset of the silo created by the safety regions. Due to the limited size of this dataset and the simple query we are able to retrieve the basic information relatively simple. More advanced and deeper information or analytics could be gathered using more complex queries but these should be on a limited filtered subset of the data. This does results in some form of redundancy.
 

CV

Nicky van Oorschot is a developer and researcher for Netage. Netage is a Dutch company specialised in building Linked Data application (closed and open data) for fire departments in Europe and the US.
Projects he has been working on are innovative projects using semantic web techniques together with web ontology extraction and machine learning techniques.
 

Interested in this talk?

Register for SEMANTiCS conference
Register