Page tree
Skip to end of metadata
Go to start of metadata

Date

Attendees

 

Dennis Wisnosky

Agenda

  1. JIRA Issues Review and Actions
  2. Agree definitions for the agreed occurrent facets
  3. Consider how to link the legacy occurrent concepts (e.g. Event, Activity, Process, Lifecycle, Milestone) to sets of these
  4. Consider how to model interest accrual and its building blocks

We did not get to (3). We covered (4) ahead of this, so as to get an idea of how this will work.

Decisions: 

Made in UML which will result in RDF/OWL (including definitions and editorial / explanatory notes added to the model content in this prototype model)

Note that all decision made during this session were implemented on the prototype model during this session. The only actions are to re-create this model in RDF/OWL and to continue with the crafting of definitions and axioms for the classes we did not get to today.

 

2. Create and Review Definitions for the Occurrent Facets

 

Background

 

We have created a set of pairwise disjoint sub-classes of Occurrent Thing, based on the presence or absence of one aspect each. These aspects will be formalized by OWL properties where possible, but in any case the meaning of these classes is intended to be clearly understood and axiomatic.

 

The reason for these facets is that the primitives given in DOLCE and described in the Stanford Encyclopedia of Philosophy site are themselves composites of more than one aspect of what it means to be a given kind of occurrent thing. We have now agreed upon a set of facets with one aspect each.

 

Principle: Home in on a concept, agree or generate a definition which precisely corresponds to the intended concept, and then consider what might be a good word for it.

 

This means we are not going to look at dictionaries with the expectation that they will have a definition corresponding to precisely that concept. Instead we will agree definitions today, and look to dictionaries (and standards like DODAF) for help as and when we need it.

 

Definitions

 

For each class, these are the definitions and editorial notes or explanatory notes that we agreed on in the call.

 

Culminating and Non Culminating

 

This is basically the distinction between something like a person walking up a hill versus having reached the top of the hill. These form part of the DOLCE / Stanford concepts of Accomplishment and Achievement (themselves synonyms in natural English), while those concepts also include meanings from other facets.

 

Here we made reference to a dictionary source (the Free Online dictionary) for a suitable definition of Culminating that we could adapt for our intended concept.

 

Culminating Occurrent

 

Definition

 

An occurrent which has or can come to completion.

 

Non Culminating Occurrent

 

An occurrent which has not or will come to completion.

 

 

 

Explanatory note:

 

This includes things like maximizing profits, orbiting the sun etc.

 

Properties

 

We did not define any properties for these.

 

Suggestion for next review: Culminating could be defined with reference to some state (not however some goal, which is a desired state).

 

 

 

Discrete and Continuous

 

These are distinguished by whether or not they are described as going on over a period of time, or as happening at an instant.

 

Discussion

 

From Chat Log:

 

Michael Bennett (to Everyone): 18:39: Precision: the precision will therefore be framed as the definition as a unit.

 

Michael Bennett (to Everyone): 19:04: Whenever we call something an instant we say so with reference to some unit of measure

 

 

 

In addition to what is in the formal notes, there was some discussion on this, resulting in modeling decisions.

 

There are two seemingly incompatible ways of modeling time:

 

  1. Instant and Interval are disjoint children of Temporal Thing (W3C Time and Legacy FIBO)
  2. There is no such thing as an instant, since all instance can be decomposed into intervals at a finer granularity on the time axis (OMG DTV)

 

Legacy FIBO was modeled around W3C Time, but we had intended to formally commit to OMG DTV. In the event, instead of getting OMG DTV we got a quasi-operational ontology for time (meeting the requirements of the current use cases for schedules etc.), called FIBO DateTime, crafted from whole cloth without reference to OMG DTV.

 

Meanwhile DTV does have a way of framing the idea of an instant, with reference to some temporal unit or scale.

 

The Need for Instants

 

The issue is this: while it is correct physics to say that there are no instants, it is a feature of language that many events are defined as occurring instantly. This feature of language is also seen widely in financial data.

 

Example: when we specify a payment date, this is specified in the language of an instant, where that instant is a day. So such and such debt matures on 31 July. There is no decomposition of this, as in physics, because this is a legal statement not a physical measurement – it is not meaningful to ask at what time on 31 July the instantaneous event actually happens. It happens on that day as a day.

 

The same will be true for things that happen or are specified to happen at a given hour, minute or even year.

 

So we need the language of instants.

 

At first we modeled Continuous Occurrent as beign an occurrent with a duration. In natural language this is is right, however “Duration” in legacy FIBO Time is a kind of measure, or the statement (like length) which would be expressed as a measure and is about some actual temporal interval.

 

So we changed this so that the property that makes a continuous occurrent continuous is now “has duration” but “occurs over interval” and made the range of this the class Interval from W3C Time.

 

Then we were also able to make the corresponding assertion for Discrete Occurrent, that it occurs at some instant, with reference to W3C Time “Instant”.

 

Precision

 

So: in physics any “instant” is itself decomposable into some duration at a different time scale, and given that we have terms which specify certain events or requirements only with reference to some time scale (without decomposition into a more granular time scale)…

 

Proposal: specify Discrete Occurrent itself, with reference to some unit of time (e.g. a day in the debt maturity example), that unit being the scale at which the language of “instant” is applied.

 

All agreed this will work and will bridge the W3C and DTV / physics views of time.

 

We made “has temporal granularity” a property of Discrete Occurrent, with a range of Temporal Unit.

 

Then:

 

Units / Dimensions

 

The things we are talking about here are not unique to time. There is a 3 way set of relations between our Units of Measure work, the Time ontology requirements, and these Occurrent Things. So Measures are given in units, have granularity and so on whether they are temporal, spatial or in other dimensions such as charge or mass.

 

We looked at the diagram from the Quantities and Units of Measure work. The material brought in from QUDV includes Dimension, with types of dimension including Time, and it has Units, with the potential for us to re-frame kinds of units (currently in legacy FIBO), with reference to those dimensions. Those would include the units of time we need now.

 

Since this is a prototype model and not the live edit, we used the legacy concepts for Unit and Temporal Unit, with the expectation that when this is done as an OWL edit we will pick up the latest concepts from the units of measure work here.

 

Non Occurrent Model Change

 

Finally, we made “has granularity” a property of “Thing”, so it may be applied both to occurrent and to continuant things.

 

Then the property added to Discrete Occurrent above, would become a restriction on that property.

 

In this prototyping environment (the original EA tool with up-issued model), we do not have the means to specify a Restriction as an Association Class. In experiments on visual presentation for restrictions this has been a useful base class to use – so in future we would want to add a stereotype for Restriction with AssClass as a base class (this is not compatible with ODM).

 

In the mean time we color this AssClass purpose, as a visual cue to the OWL modeler that this must be modeled as a Restriction. The RDFS ‘sub property of’ assertion would become an onProperty assertion in RDF/OWL.

 

With this, the model is complete; definitions and editorial or explanatory notes are given under each class and property below.

 

 

 

Discrete Occurrent

 

Definition:

 

An occurrent with no interval

 

 

 

Explanatory Note:

 

This is an occurrent which is described in terms of it being instant; however, in reality such a description depends upon the definition of some temporal unit in terms of which this is considered a unit, such as a day or a second/

 

 

 

Editorial note:

 

This allows us to talk about concepts which are defined with reference to an instant, where that instant may be a day or an hour or some other unit. The nature of many concepts in financial services is that they frame something as an instant at one or another scale, for example the day on which some instrument.

 

 

 

Explanatory Note:

 

Whenever we call something an instant we say so with reference to some unit of measure.

 

 

 

Continuous Occurrent

 

Definition:

 

An occurrent with an interval

 

Property: ‘occurs at instant’

 

Domain: Discrete Occurrent

 

Range: w3c:Instant (in local snapshot namespace copy of w3c:time)

 

Definition:

 

The instant in time at which the occurrent occurs

 

Property: ‘occurs over interval’

 

Domain: Continuous Occurrent

 

Range: w3c:Interval (in local snapshot namespace copy of w3c:time)

 

Definition:

 

The interval of time over which the occurrent occurs

 

Property: ‘has temporal granularity’

 

Domain: Discrete Occurrent

 

Range: Temporal Unit (in legacy FIBO; to be replaced with Temporal Unit as child of Unit from QUDV snapshot ontology, in the Quantities and Units of Measure work)

 

Definition:

 

The temporal unit in which the instant in time at which the occurrent occurs is defined.

 

 

 

Functional and Non functional

 

These correspond directly to what were labeled as Telic and Atelic in Stanford. Unlike Accomplishment and Achievement, for these concepts the meanings are quite clear from the names, however these names are obscure philosophical terms. These represent the same root as “teleological”

 

An additional wrinkle: to some philosophers, teleology implies that something has some intention, with a corresponding agent intending the function or purpose of the thing. To others, teleology extends to things which have function but not some intending agent, for example evolutionary processes.

 

So Telic and Atelic are labeled Functional and Non Functional Occurrent, and are defined with reference to some function. A separate pair of facets are given as children of these, namely occurrents which have some purpose, as pursued by some intending agent. These are labeled Intended Occurrent and Non Intended Occurrent.

 

Functional Occurrent

 

Definition:

 

Occurrent which is defined in relation to some function

 

 

 

Non Functional Occurrent

 

Definition:

 

An occurrent which is not defined in relation to any function.

 

 

 

Explanatory note:

 

An accident would be a non functional occurrent. These are things which happen or are happening, but have not purpose.

 

Intended Occurrent

 

Definition:

 

A functional occurrent which is intended by some agent.

 

Non Intended Occurrent

 

[No definition]

 

Property: is intended by

 

Domain: Intended Occurrent

 

Range: Autonomous Agent

 

Definition:

 

The entity which intends the occurrent

 

Prescriptive and Descriptive

 

This facet is needed to distinguish between process models, which are descriptive, and models of things which happen (whether these have happened or will happen – further sub-classing the Descriptive facet).

 

Prescriptive Occurrent

 

[No definition]

 

We did not get to this term; this is the basis for process definitions and plans.

 

Descriptive Occurrent

 

[No definition]

 

We did not get to this term

 

Stative

 

So far, there is no pairwise disjointness for State, and we have not come up with properties that would formally define what one of these – so it remains simply an axiom at this point.

 

We did discuss the idea of dynamic versus non dynamic, but could not come up with a formalism whereby dynamic is anything other than an occurrent (as distinction from “complex dynamic” which is one specific kind of dynamic system). We may revisit this. It may be that there is some relationship or exclusion between this and stative or state but this is not clear yet (the obvious disjoint of dynamic is static, and static is not occurrent; these may be relevant facets of continuant thing). The Dynamic facet has not been added at this time. We should review other literature for possible ideas on this.

 

State

 

[No definition]

 

We did not get to this term

 

4. Interest Accrual

 

We previously defined a class which we labeled (self-definingly) as “Ongoing Purposeful Occurrent”.

 

Ongoing Purposeful Occurrent

 

There was an error, in which this was modeled last week as a child of Functional Occurrent, where it should have been a child of Intended Occurrent. Changed this.

 

[No definition]

 

Interest Accrual

 

Added the assertion that this is a child of Culminating Occurrent.

 

[No definition]

 

Proceedings:

Three issues were resolved and closed as identified below.  Please go to FIBO JIRA-FND for details.

 

  1. New Feature FND-13
    1. Come up with more generally acceptable terminology to label the 4 sub classes of perdurant.  Mike will add the interposing DOLCE classes and compare with the 4 cell table we saw today.  This would be a polyhierarchy of telic / atlic and instantaneous …
  2. Bug FND-4
    Update cardinality of Party in Role in Agreement
  3. Improvement FND-2
    Keep external dependent files in git with appropri
An issue was transferred to FIBO-INFRA and assigned to Dean INFRA-92 Keep external dependent files in git with appropriate versions of FIBO
 An issue was transferred to FIBO-FLT for the FLT to consider FLT-38 ISO-3166 and Language Codes results in unacceptable reasoning performance

Ten issues remain in FIBO-FND.

Action items

  • Mike Bennett Re-model the concepts, definitions and annotations that in the EA prototype, in RDF/OWL (after completing the remaining classes)

  • Mike Bennett and Team: Complete the definitions and axioms for the remaining terms we did not get to on 24 July.