Date

06 Jun 2017

Attendees

Agenda

1) Use Case reminder

2) Where we are on our road map.

3) Open Action Items

4) JIRA Issues Review - https://jira.edmcouncil.org/projects/FCP/issues/FCP-4?filter=allopenissues

5) Todays content discussion.

CFTC Ontology and Triple Store Pilot

The purpose of this pilot is to demonstrate the following:

Map to Financial Industry Business Ontology (FIBO). The CFTC has mapped Interest Rate Swaps (IRS) fields received (via comma and tab delimited) from all four Swap Data Repositories (SDRs) in a single spreadsheet (to be provided). We will determine which fields can 1) map easily to FIBO and 2) provide a viable pilot. Note: There is no need to map all the fields at this time. In fact, we should use as few fields as possible to expedite the pilot.
Use FIBO to create a Triple Store and import a month’s worth of real IRS data from the SDRs.
Produce Conceptual and Logical data models of the IRS fields.
Provide analytics similar to those completed as part of the previous Proof of Concept (POC).
Export data from the triple store for consumption by an RDBMS. Note: CFTC continues to experience loading issues.
Populate an RDBMS using the source files (comma and tab delimited) and FIBO. Note: CFTC continues to experience loading issues.

Proceedings:

20170606 FIBO CFTC Prototype

JN: Prototype is a go. James will be a resource. Agreement from Shrini to get $$$. Now need to work with procurement people to get the $$$ under contract. But, would like to keep going in the meantime.

DN: Need to focus on what data CFTC wants to consume.

Jeff: Should have more data, not less.

DW to JN: Time from for $$$$. JN: Maybe 4 weeks. If we could have a demo in 4 weeks that corresponds to getting funding, that would be great.

DN: If we can use existing model, with CFTC data, this can be done. JN: Can't spend weeks talking about one field. If this will happen, then wait till we get the $$$$.

JN: Displays purpose for Pilot. Question about why doing RDBS and how and what it would accomplish in this pilot?

DA: Would populate RDBMS from the triple store. Easy! CSV to FIBO to RDBMS.

DN: Can ANZO do this. Unknown. DA: Can build little Schema's with SPARQL queries. But, not at scale. DN: There is a semantic mismatch on purpose between RDBMS and triple stores. DN: Requirement is the most interesting.

JN: Puts up questions we should answer. DN: We focused on Swap Data Confirmations. JN: So, we should stick with that. Max: We could easily add more fields.

JN: Should we use an existing relational database, or build a new one. Need to ask James.

Question : Could bring each SDR into a named graph. and then use a default graph to aggregate them. DA: ANZO can do quads also. JB: Does FIBO have the notion of a data source property. EK: Some are in indicators. DN: Would use PROVO in this process. Easy!

ACTION: Schedule working sessions with James for work on data.

Decisions:

Action items

David Newman Schedule working sessions with James for work on data.

Browser not supported