Skip to topic | Skip to bottom
PASOA
Home




PASOA.Ontologies

Start of topic | Skip to actions

Provenance-Related Ontologies

On this page, we describe the provenance-related ontologies attached to the page. Concepts in these ontologies are used by reference within process documentation.

Temporal Relationships Ontology

The temporal relationships ontology, temporal.owl, aims to define relations between instants in time, so that we can describe when one thing happened before or after another. It is not intended to be a complete conception of time. As we are dealing with a distributed system in which physical clocks cannot realistically synchronized, we cannot specify instants by absolute clock times. Instead, the following concepts are defined:

  • An Instant class, representing an instant in time.
  • Sub-classes of Instant that represent the three ways in which instants are defined within process documentation:
    • InstanceOfSend is the instant that a message was sent by an actor.
    • InstanceOfReceive is the instant that a message was received by an actor.
    • InstanceOfActorState is the instant that an actor had an asserted state.
  • Two relations (object properties), after and before, whose domain and range are Instant, and represent the fact that one Instant came after or before another in time.

Using the temporal ontology, we can specify temporal relationships between system actions or states. We can also use it to refer to application entities in given ranges of time, e.g. patient X before their transplant operation. Such references can be used in asking for asserted properties of entities and in determining the provenance of the entity.

Causal Relationships Ontology

The causal relationships ontology, causal.owl, aims to define causal relations between properties of the system at different instants. We define "C caused E according to A" as A's belief/assertion that E would not have become true if C had not been true, all else being equal. The causal ontology uses concepts from the temporal ontology, and adds the following:

  • A EntityAtInstant class that represents something being true at an instant of time.
  • An atInstant relation (object property) whose domain is a FactAtInstant and whose range is an Instant, used to assert the instant at which the fact is true.
  • Two relations (object properties), caused and wasCausedBy, whose domain and range are EntityAtInstant, used to assert that one fact being true at one instant caused or was caused by another fact being true at another instant.
  • Two sub-classes of EntityAtInstant?, Cause and Effect that represent the cause and the effect in a causal relationship.

Additionally, we specify two rules which we know must hold: that if fact F0 being true at instant T0 caused fact F1 to be true at instant T1, then T0 must have been before T1, and correspondingly for wasCausedBy/after. These rules can be expressed as SeRQL? rules as here.

Using the causal ontology, we can specify causal relations between the contents of interactions and between actor states. For example, we may specify that an argument output in a message by service S was "5" because the two arguments given to the service were "10" and "2" and the function it was performing was "division".

-- SimonMiles - 17 Nov 2005
to top

I Attachment sort Action Size Date Who Comment
temporal.owl manage 1.4 K 16 Nov 2005 - 10:33 SimonMiles Temporal relations ontology
causal.owl manage 1.7 K 11 Jan 2006 - 11:38 SimonMiles Causal relations ontology
causalserql.txt manage 0.7 K 17 Nov 2005 - 16:04 SimonMiles Rules of causality expressed as SeRQL? queries

You are here: PASOA > Ontologies

to top

Copyright © 2004, 2005 by the PASOA consortium.