Unknown User (victoriastarr)
Unknown User (pablomiranda)
Unknown User (stefaniiestuarts)
1) Use Case reminder
2) Where we are on our road map.
3) Open Action Items
4) JIRA Issues Review
5) Todays content discussion.
6) For next week.
Michael Atkin (to Everyone): 11:11 AM: CC - goal for today - this is a collaborative process, there is a starting point but want adjustment
11:12 AM: CC - project initiated by FNMA - seeking feedback and alternative approaches to develop a set of ontologies around ratings
11:13 AM: CC - assumption that folks understand the basics about formal ontology and FIBO. Most of the discussion is designed to be at a concept level
11:16 AM: CC - FIBO system of record is OWL (also using UML diagrams) - objective is to harmonize data both internal and external for integration into applications. Using OWL also for inference capability (flexible analysis).
11:19 AM: CC - multiple level of ratings (and things to be rated) - but the core concepts are shared across uses. First step is generic concepts --> then more specific ratings ontologies for instruments and entities
11:20 AM: CC - essential concepts slide (five core concepts) - ratings scores (i.e. AAA, B+) and scales (assign meaning and ranking to scores) - these are reference information
11:20 AM: Rating scales are published by some rating agency
11:22 AM: CC - the "rating (fact)" is dynamic and created by the rating party to rate something
11:23 AM: SR - this is the highest level (makes sense) - think about the context of the rating and understand there are multiple parties involved in the rating process
11:26 AM: CC - model of the essential concepts (and the correspondence within the model) - and links to predefined concepts in FIBO
Ron (to Everyone): 11:28 AM: Does the modeling technique allow for subtypes?
11:28 AM: CC - (next level of detail) - rating has an "effective date" and is for a period. Concept of "rating issuer" (which may or may not be hte same as the rating party
11:29 AM: SR - word "rating issuer" is confusing. Keep in mind that what is being rated is different from the agency doing the rating
11:30 AM: CC - agreed - we need to get our language clear on the concepts associated with ratings
11:32 AM: SR - think about including the rating analyst (as one of hte "ratings party") - the individual performing the rating. Rating party could include the analyst as well as the ratings agency
11:33 AM: CC - will incorporate the concept of "party in role" to distinguish the actors in the process
11:34 AM: CC - model accounts for "ratings codes" (allowing for numeric values). Ratings score code is proposed for ordering ratings schemes
11:35 AM: SM - what's included in "ratings score"
11:36 AM: CC - scores = definition of code (issuer defines the semantics of hte ratings scale)
John Gemski (to Everyone): 11:37 AM: We need to clarify the difference between rating party and rating agency. To me they are the same. Maybe give an example of each?
Michael Atkin (to Everyone): 11:37 AM: PR - two dates are important (the day the rating was made as well as the date of coverage) - caution urged on the word "effective date"
John Gemski (to Everyone): 11:37 AM: Rating agency could itself be a role of party in role
Michael Atkin (to Everyone): 11:37 AM: CC - agreed - must capture these differences
11:40 AM: CC - example from Moody's (knowledge graph)
11:40 AM: JG - in diagram is Moody's a ratings agency or ratings party
11:40 AM: CC - in this case they are both
11:42 AM: CC - scope of activity is financial ratings. This conversation is starting with generic concepts that should be applicable to use cases beyond financial
11:44 AM: PR - credit ratings is another example - BoA could issue a FICO score (FICO is not a BoA concept)
11:45 AM: SR - possibly of different definitions for the same ratings concept (AAA from Fitch and Moody's might slightly differ based on the process of the ratings agency)
11:45 AM: CC - In this model - it is rating code as defined by rating agency
11:48 AM: CC - concepts of party rating an event and producing a report
John Gemski (to Everyone): 11:48 AM: Another thing we should show is the relationship between different rating scales. For example - Fitch Aa might equate to Moody's B.
Michael Atkin (to Everyone): 11:48 AM: CC - assessment activity (the rules for the rating)
11:51 AM: CC - definitions of relevant concepts for review (classes as well as the properties) - request is for review for business acceptance
11:52 AM: CC - introducing the concept of "notional" partitions - how many of these concepts can be defined at an industry level (shared) vs. those that are specific to the individual rating scale
11:55 AM: CC - Process (Wiki for documentation) - meetings schedule (weekly, Wednesday at 11:00) -- generic ontology (one month) -- specific areas to be determined
John Gemski (to Everyone): 11:59 AM: We need to specify whether we're modeling rating consumers or the rating process or both.