2017-01-10 Meeting notes
- Dennis Wisnosky
Date
Attendees
- Elisa Kendall
- Former user (Deleted)
- John Gemski
- John Nowlin
- Maxwell Gillmore (Unlicensed)
- Tahoe Blue
- Wesley Moore
Proceedings: 20170111 FIBO CFTC PoC.docx
20170111 FIBO CFTC PoC
David: John Nowlin, what is the news on use cases? John: We are moving slowly because of red tape. Moving slowly. Have installed Graph-DB and Blaze Graph. Technical issues because we are new to ontologies and triple stores. Difficulty loading directly. Have installed AllegroGraph and Gruf. Thank you Dean, for sending us FIBO. We are working with James 140 fields data mart. These are comma delimited data files and difficult to work with. We can't yet share this. Need to go through internal hoops. Our intent is to map this data mart to FIBO. Need to upgrade Portege first. Now on 3.5 version. It will take 3-4 weeks to complete security process before we can proceed with Cambridge Semantics.
DN: No worries. Is there some process that we can follow to see how our model matches your fields. JN: CFTC issue is that the attorneys are concerned about revealing even meta data, at this time. CFTC working on this, but no date yet. There are 120 fields published in Fed Register that could be used. DN: Wes has identified these.
JN: That would be good enough. This is a PoC, not the real system. Change in Administration may make it easier to collaborate. Drafting an MoU saying how this PoC would work could help. An MoU to say how EDMC would sign NDAs and work with CFTC. Moving forward on the technical aspects of a PoC, but still can't formally have a collaboration agreement.
DN: Should we keep this engagement confidential for now? JN: Yes, until the new Chairman, John Carlisle is in place. This would be better viewed as a new initiative under the new Chairman. John Carlisle will be acting until at least the new President makes an appointment. According to public knowledge, John. Carlisle is supportive of new technology and the mission of the CFTC.
DN: Wes can we have a look at the 120 public data fields next week. Wes: I have looked at 50 of the main terms and can share with this Team.
DN: Our questions are what do we have modeled and what do we need to further model?
JN: Can't buy Cambridge Semantics tool, but it could be used on the PoC. Chose Blaze Graph because CS told us that they use this. DN: But, they also bought SPARQL City and use that. JN: We don't care what the underlying technology is. We are evaluating FIBO, not the technology. Need to prove that FIBO Triple Stores make it easier for us to find the bad guys.
DN: It would make sense for us to have a session with Marty L of Cambridge to see what tools we really need to do this PoC with them.
JN: Our point is that the driving force is FIBO, not the platform.
DN walks through current state of swaps in FIBO. Question to JN, is this useful for your use case?
JN: We have 5 use cases under one major use case. We will make sure that EDMC gets all 5, but can't share them now. Start with vanilla interest rate swap. We will be asking. "what if" questions. For example, what if the price of oil goes up dramatically?
DN: Standing by to work on this. JN: Use Cases heavy on exposure and risk. Want to be able to do dynamic analysis. Analysis that can't be done, with SQL, or least not done easily.,
JN: I will be bringing in our business people and they will be interested in "exotics". The risks go up with exotics that cannot be managed with SQL queries.
DN: We will do simple use cases first and then move on to exotics. JN: Yes!
DN: A question we will be investigating is how we use machine learning to enhance our understanding of the data.
DN: Next week we can look at the model - the 120 elements. And, look at the Abox to validate the individuals for constructing test cases.
There is agreement on this.
Decisions:
Action items"
- Wesley Moore Be prepared at next weeks meeting to discuss the public fields he has investigated
- David Newman Set up a Marty Loughtlin and John Nowlin meeting