FIBO Foundations Content Team Mike Bennett EDM Council
FIBO™ Foundations Content Team Mike Bennett EDM Council ● April 23 2019 Confidential
Agenda 1) News 2) Open actions / housekeeping etc. 3) Temporality Discussion next steps – Open issues / discussion – Requirements: Use cases / example data – Questions for FLT to address 4) Jira Status 5) FND Roadmap for Q 1. 5 Release 6) Ao. B EDM-Council/FIBO Foundations Content Team 2
NEWS • No news • Questions of timing for this meeting EDM-Council/FIBO Foundations Content Team 3
Open Actions Review • https: //wiki. edmcouncil. org/display/FND/ Actions EDM-Council/FIBO Foundations Content Team 4
Pull Requests Review Since 16 April • FND-240 – loosen restrictions on certain date / time – Definition of ‘Asset’ has also been made broader (covers more uses of the word than the single original concept) – This change was reversed out – New FND-242 raised for datatype issues • BE-202 – integrate informative material, deprecate remaining Ext ontologies – Comment: use of arrxx-math dependency remains – Changes pushed into CIV need more work • SEC-82 Bonds simplification – FND Impact: Eliminates use of Values ontology • Thought we had adopted this in place of Business. Facing. Types • Though the reserved term overload was addressed? • Do we now have policy to NOT use Values? New if so • FBC-103 Exempt Security EDM-Council/FIBO Foundations Content Team 5
Temporality • Discussion initiated at FLT Call and continued last week – Last week: FIBO concept intended to be atemporal (e. g. IR Swaps interest terms) • Requirements: Use cases and Context – Loans – phase, status, temporal information – Market data (time series etc. ) – Derivatives data (greeks etc. ) • Discussion – Questions to take to the FLT EDM-Council/FIBO Foundations Content Team 6
Reminder: Concepts Extraction = Context-specific concept ontologies N-dimensional content Shown as 4 D for simplicity © The Semantic Shed 2019 Various extracts from that hypercube in lower dimensionality
FIBO Deliverable Ontologies = Context The FIBO Sweet Spot: Multi. Context concept ontologies Extract from that hypercube in lower dimensionality N-dimensional content Shown as 4 D for simplicity © The Semantic Shed 2019 Extraction?
FIBO IR Swaps Interest Terms in FDTF Po. C • Action (16 April): Check parentage for consistency
Checks: Superclasses Usage • Consistent use of Occurrence Kind • Use for all such atemporal occurrence concepts 10
Conceptual Thinking • Understand the problem space • Design against that understanding, for a given use case • Use and validate against examples data EDM-Council/FIBO Foundations Content Team 11
Conceptual Time • Time versus Temporality • Time: – Date as a thing – Data and time datatypes • Temporality – Past present and future values (‘tense’) – Commitments / prescriptions v descriptions EDM-Council/FIBO Foundations Content Team 12
Time Ontologies • W 3 C Time – Instant and Interval – Extended in legacy Time ontology • Parametric versus absolute time cncepts • OMG DTV – Based on Allen Relations – Concept focused • FIBO Financial. Dates – Implements the Fp. ML temporal concepts within pragmatic OWL – Did not use DTV in the end EDM-Council/FIBO Foundations Content Team 13
Conceptual: Prescriptive and Descriptive Occurrents Control flow Process occurrence is the ‘instance’ of the prescriptive occurrent being carried out • From OMG Blockchain Po. C • Applied to process occurrences
Conceptual: Three Perspectives in Time © The Semantic Shed 2018 • Specification: How to make a bicycle • Plan: Make 100 bicycles • Account: Used 201 wheels EDM-Council/FIBO Foundations Content Team 15
Pragmatic Ontologies: Temporal Options • Option 1: Atemporal Ontology – Does not treat for time at all – All concepts in the ontology are used for whatever time is relevant to the application – Commitments • Commitment: binds party to future activities • Not a temporal matter but a commitment to act • Option 2: Ontology treats for different temporalities – History • Extract sub-set for moment in time • Separate sub-set for reporting multiple values (time series) in the past – Predictions • • Time = future Predicted by some means, not absolute Not date/time in many cases Prediction for end of month, period etc. (debt analytics, yields EDM-Council/FIBO Foundations Content Team 16
Pragmatic: Commitment versus happening • Schedules: Specifies a set of events we commit to – Uses date roll rules, relative dates etc. – Business calendars = Commitment • Reports: Records what actually happened – It happened on a given date • Risk / projections: Future time – Not the same as Commitment – Descriptive future, not prescriptive • Conclusion: Don’t mix contractual terms with reporting concepts – These are not simply atemporal since one is a commitment not a prediction EDM-Council/FIBO Foundations Content Team 17
ACTUS EDM-Council/FIBO Foundations Content Team 18
ACTUS • ACTUS part 1: Standard contract terms – Commitment not prediction • ACTUS Part 2: What happens to the instrument during its life (environment) – – Changes in market variables e. g. interest rates Behaviors, sentiment etc. Happy v unhappy paths This is all about the life of the contract in the world EDM-Council/FIBO Foundations Content Team 19
ACTUS As of Now (temporal) FIBO (and ACTUS) Contract Terms EDM-Council/FIBO Foundations Content Team 20
Practical Ontologies: What does the Data Look Like? • As of now data – Real time pricing, analytics • Time series data – Past prices, yields, analytics • Trade Execution – Clean v Dirty Price - accrued interest amount at trade date – Status (CD/XD) • Securities trading status • Credit ratings and status • Loan phase and status • Proposed Rule: whatever the data looks like, include the semantics of that data EDM-Council/FIBO Foundations Content Team 21
Some Fun Stuff: Debt Analytics and NPV • Discounted cash flow (DCF) – Based on current Value of future cashflows • Variable interest: – Use period end rates as basis for DCF • Use the current value (%) of the base rate as basis for estimation – That future dated value yields a present value – That present value yields current analytics figures as of now • Tomorrow: the month end estimate changes again – Today’s estimate of NPV differs from yesterday’s – Based on the SAME future event (this period end interest amount) • So you have a past time series of future value projections EDM-Council/FIBO Foundations Content Team 22
Ongoing discussion on these • Date rules (date roll rules, holiday calendars) • Rules for determining dates and times – Schedules – Relative dates • Past dates / times that a thing actually happened • Future dates / times where these are known • Questions on these – When to use date related types – When to use ontological class of ‘Date’ or similar EDM-Council/FIBO Foundations Content Team 23
Discussion EDM-Council/FIBO Foundations Content Team 24
JIRAs Triage • Categories – – Trivial, textual fix Complex but textual fix Definitions / metadata only Needing CCM or Protégé modeling and validation • Who / When – As defined in Jira ‘assigned to’ EDM-Council/FIBO Foundations Content Team 25
JIRAs Triage • See Spreadsheet EDM-Council/FIBO Foundations Content Team 26
FND Roadmap • Planning spreadsheet; for 2019 Q 1. 5 • See Wiki page at: – https: //wiki. edmcouncil. org/display/FND/Roadmap+Spreadsh eets EDM-Council/FIBO Foundations Content Team 27
Next Meeting • We meet weekly – – Tuesday at 3 pm Eastern Time NEW DIAL-IN DETAILS FCT leads invited Invite is via Wiki page creation only – use Watch! • Next call – Next call: 30 April 2019 EDM-Council/FIBO Foundations Content Team 28
Annex Recent JIRAs • Working backwards from most recent EDM-Council/FIBO Foundations Content Team 29
Recent Jiras • FBC-219: loosen constraints on has. Name properties to change the range from xsd: string to rdfs: Literal • • • Raise corresponding FND Jira for other comparable properties FND-239: Allow modeling of composite identifiers FND-238: Support requirement for time zones in date related datatype(s) – Raised wrt Feb 12 notes • FND-237: Date time datatype requirement for nanoseconds – Raised wrt Feb 12 notes • • FND-234: Shortfalls on Independent. Party FND-233: Need a separate identifier for party-in-role – See below • FND-231: Cannot distinguish locations that are points vs those that areas – See below • FND-229: Invalid definition of Currency. Identifier – Simple change in definition EDM-Council/FIBO Foundations Content Team 30
FBC-219 • Loosen constraints on has. Name properties to change the range from xsd: string to rdfs: Literal. • rdfs: Literal is the default in OWL so action was to delete the range restrictions • FBC-219 did not cover the People ontology, only Agents and Relations, since the issue was about LEI content. • Agreed action last week: – Raise a new Jira to do the same fix in People; – Check all other cases where type is a string but legitimately may have variants in other languages (like has. Name) EDM-Council/FIBO Foundations Content Team 31
FND-233 • Need a separate identifier for party-in-role – Currently we have identifiers for organizations, but not for parties-in -role at the highest level. Examples include identifiers for customers / clients, borrowers, lenders (in the role of lender rather than for the financial institution per se), etc. • Follows discussions on Pull. Req for FBC-212; per Dean: – I don't feel that this discussion has really socialized this change fully - I'd particularly like to hear from @Cory. Casanave about whether this particular change makes any commitments that he thinks will get us in trouble in the future. – If not, then I say we merge this in and continue to tackle the difficult issues around identification as we continue to develop FIBO. • Discuss… EDM-Council/FIBO Foundations Content Team 32
FND-233 MB comments on FBC-212 Thread – The issue of identifiers for things-in-functions (as for things-in-roles) is a complex one. – In most cases one would expect that an identifier issued to an entity for the purpose of some function, is still issued to the entity as an entity. – There are likely few data sources out there that define thing-infunction or thing-in-role concept directly, so unless the identifier is in an isolated data source that belongs only to the content of that function (as an implicit context) it's likely it is assumed to belong to the independent entity. – A simple way forward would be what we did with LEI which was to recognize that eligibility was relative but identification remained that of the independent thing. – The litmus test is this: would the same bank, performing two such functions, have a separate identifier for both functions? • And if so, would these always be from entirely separate schemas (one schema per function)? – If this is the case, then the identifier could indeed identify the relative thing. EDM-Council/FIBO Foundations Content Team 33
FND-233 PR Comments in Pull. Req Thread • What about the example I gave of Customer. Identifier which is arguably associated with a Thing(? )In. Role; or an Employee identifier? • Even a SSN is arguably associated with a Person in Role of Citizen of one Country (I have one for US and another for UK). • So this needs more consideration IMO EDM-Council/FIBO Foundations Content Team 34
FND-233 EK Comments in Pull. Req Thread • That's a different challenge - service providers are functional entities, and we have not made them disjoint from independent parties, though I think we should. • It was the fact that the organization identifier applied to organizations, which are disjoint from agent-in-role that caused the issue, but only because there was another inference for the individuals in question that inferred that they were also parties (from has. Domestic. Ultimate. Parent). • So, to address your issue I will raise another issue [FND-233] to add an identifier for a party in role – but that will be separate from an identifier for a functional entity regardless – unless we make all functional entities parties in role, which we have not done to date. EDM-Council/FIBO Foundations Content Team 35
Role v Functional Relativity • Relative Thing = contextually defined thing • Contexts: – – – Role Function Actor in process Underlying of a derivative Examples elsewhere: • Catalyst • Thing. In. Role is just one example, where the context is the role • Functional. Entity (in BE) is separate EDM-Council/FIBO Foundations Content Team 36
Alternative Pattern (Cory) • We could also have a supertype of legal entity and legal entity in role – I have needed such patterns before. EDM-Council/FIBO Foundations Content Team 37
Disjoint Relation(s) • PR: A short term fix for Elisa's inconsistency issue would be to remove the disjointness I have always argued was wrong - this adds weight to that. – CC: Until we have factored out (or decided to ignore) differences between a role inhering in the entity (Dennis is a pilot) or a role with respect to something else (Dennis is a pilot of [with respect to] flight UA 101), I agree the disjointness is over-committed to a particular implementation pattern. • EK: it's up to the Foundations FCT whether or not we eliminate the disjointness, but that does not change the fix that I have made, which would be correct in either case. EDM-Council/FIBO Foundations Content Team 38
FND-233 Theory • Case 1: Relative Need, Independent Thing Identified – Example 1: LEI – Example 2: SSN etc. (Resident is relative) – In Example 1, what is identified is the independent thing • For example, one identifier for being ‘an employer’ does not refer to any specific employer / employee relation, is is ‘about’ the Independent Thing – In Example 2, we scope the model to treat certain relative things as being de facto independent for our usages • Make sure this is appropriate for all usages • Case 2: Relative Need, Relative Thing identified – What is identified must itself be thing in the role – This would be at whatever level is the Relative Thing – Indication: there is a class specifically for one thing in one role in a specified context • Not compatible with multiple classification of individuals! EDM-Council/FIBO Foundations Content Team 39
FND-231 • “Cannot distinguish locations that are points vs those that areas” • Response (posted today) – These are conceptual abstractions (good ones) • In general: Introduce the need for new conceptual abstractions by way of a use case or formal requirement for one of the other FCTs or from a Po. C • Requirement has now been confirmed – This Jira to be actioned as stated EDM-Council/FIBO Foundations Content Team 40
FND-229 • Incorrect usage of ‘trigraph’ in definition for Currency. Identifier – Sounds fair. – Implement as seen EDM-Council/FIBO Foundations Content Team 41
FND-228 • An ad hoc schedule entry should be a child of collection constituent – Thought it was – Make it so! EDM-Council/FIBO Foundations Content Team 42
FND-226 • Definition of Agreement • Follows from changes in FND-224 – Requirement already agreed as part of that work – Distinctions are in the wording – To be done as seen EDM-Council/FIBO Foundations Content Team 43
Address Po. C Next Steps • Pick a time • Inputs – Interesting address challenges – Pete Rivett – Other address challenges – MB • Westward Ho! – Usage contexts to support • People – PR, MG, MB, JG • MB to circulate note to attendees – Focus of Po. C is on physical addresses – FIBO Address pattern also covers virtual EDM-Council/FIBO Foundations Content Team 44
- Slides: 44