Events and Situations

From Spatial Data on the Web Working Group
Jump to: navigation, search

Events, Actions, and Situations

In a nutshell, old-SSN observations are DUL situations while SOSA (and thus new-SSN) observations are acts (activities) and thus potentially DUL events; See [1] for more details.

  • DUL situations are social objects which are objects in turn, and these are disjoint from DUL events.
  • Hence, if we decide to align SOSA Observation as a subclass of DUL Event, we cannot define old-SSN Observation to be a subclass, superclass, or equivalentclass of SOSA Observation.
  • On a minor note, IMHO, SOSA Observation should be actions (a specific type of DUL events that have at least one participant and are executed based on some plan or workflow (which we call procedure). According to DUL an action is '[a]n Event with at least one Agent that isParticipantIn it, and that executes a Task that typically isDefinedIn a Plan, Workflow, Project, etc.', while an Agent can be "a computational agent can be considered as a PhysicalAgent that realizes a certain class of algorithms (that can be considered as instances of InformationObject) that allow to obtain some behaviors that are considered typical of agents in general. "
  • We will have to either compromise on the side of old-SSN or SOSA (and new-SSN). It looks like we favor getting the new alignment in the best possible shape at the cost of partially breaking old-SSN to new-SSN (sosa) interoperability, realizing that old-SSN usage will quickly fade away.
  • We have multiple similar situations such as the complex alignment sosa:Deployment subclass of [ owl:unionOf ( dul:Situation dul:Process ) ] which also runs into the fact that processes are events and hence disjoint from situations. This does not prevent us from having a unionOf axiom per se, but it would imply that any two deployments can be entirely different things/ideas.
  • Similarily, wrt properties, we have many alignments to the catch-all property associatedWith which does not add anything over using the OWL's universal property aside of turning everything into DUL entities (e.g., for alignments).
  • It could turn out to be very helpful that we decided not to have global (unguarded) domain and range restrictions in old-ssn as we could use something like \exists old-ssn: observationResult.T \sqcap \not old-ssn:Observation \sqsubseteq SOSA:Observation.