Data Gizmos in the Loans Industry MISMO and

Data Gizmos in the Loans Industry MISMO and FIBO Conference Tuesday April 19, 2016 Lynn Calahan, Wells Fargo Michael Uschold, Semantic Arts

Agenda § Introduction to MISMO and FIBO § Semantics and FIBO § Building FIBO Loans – Objectives & Approach – The HMDA Use Case – Live Demos: HMDA, MISMO, FIBO § Conclusions & Lessons Learned 2

MISMO: What is it? Mortgage Industry Standards Maintenance Organization § Standard pertains to data transfer or data exchange between business partners to conduct mortgage related transactions § XML Schema based § Structure is based on traditional data modeling concepts § Messages are structured explicitly and integrate with interfaces built to receive the messages § Model content reflects needs of industry partners in transmitting data to one another for transactions, reporting, and other business functions § Data terms and concepts are specific to the US mortgage industry only 3

FIBO: What is it? Financial Industry Business Ontology § § § Standard set of financial concepts and terms with definitions RDF/OWL based– semantic technology Structure is not based on traditional data modeling concepts, but on “Triples” that are used to make assertions Data content is expressed in web pages, published ontologies, and other content on the web Model content reflects business concepts used by industry partners in conducting business but is agnostic to transaction, platform, domain, and intent Content represents international financial concepts, and draws from several vertical standards 4

MISMO & FIBO: Scope Differences MISMO FIBO Mortgage Industry Finance Industry US Domestic Market VS International Data needed for transactions All financial market concepts A defined sender and receiver Used to interpret disparate info 5

MISMO & FIBO: Tech Differences MISMO FIBO XML Schema Based RDF/OWL Based Used to create structured messages VS Used to interpret unstructured data Uses classic data modeling concepts Uses a series of assertions Transmits a transaction Conducts reasoning 6

Where Do They Shine? o Requesting a service from a specific party o Submittal of data to a specific party MISMO o Transaction details and response o Rigorous specification of meaning based on logic o Machine inference o Integration of multiple data sources FIBO 7

Semantics and FIBO Most meaning in § XML is about structure terms and text § FIBO is about machine-sensible meaning What does this mean? “Commitment” 8

How Does Semantic Technology Work? Ontology (metadata) Deploying ontology as a data model Triples Possibly from a different database! URIs The key to self-assembling data National Information Center Registry . 9 http: //www. ffiec. gov/nicpubweb/nicweb/Nic. Home. aspx

Triple as a Common Denominator XML : predicate : subject Social Media RDB : object HTML 10

Example: Loans Data as Triples Loan Contract Security Agreement Collateral Appraisal Borrower & Lender 11

Formal Logic and Inference § Formal Logic: a way to represent meaning that is otherwise in code, specifications or people’s heads. To improve (re)usability & reliability. § Inference: drawing conclusions from existing information – detect subtle inconsistencies in ontology – automatically categorize e. g. a loan secured by real estate is a mortgage – data integrity 12

The Role of Inference § § Helpful in finding errors in the ontology Helpful in finding errors in the data Goes beyond XML validation capabilities Automatic categorization 13

Building FIBO Loans § Objectives & Approach § HMDA Use case § Live Demos 14

Objectives: Loan Content Team Create and maintain FIBO Loan content for use by the industry § Define concepts and relationships required by industry to represent loan data § Leverage & align with ─ established industry data standards (MISMO) ─ existing FIBO ontologies § Ongoing maintenance § Promote and support use by the industry 15

Approach: Loan Content Team § Work team contributions § Influence from vertical standards Discuss with Team Define Scope Establish Core Model Scope – Author – Review Educate Team 16

Scope – Author – Review Cycle Author Ontology Update Ontology Define Working Scope select key subset Extend Working Create review materials Review with SMEs Feedback When current scope is complete Keep going until full scope is covered. 17

Core Loan Concepts Mortgage: A Loan. Contract that has a Security. Agreement where the collateral is Real. Estate. Can infer into this class. 18

Foundational Ontologies & Modularity date & time contract currency person corporation commerce trust subsidiary underwriter exchange bank Foundations ownership occurrence party identifier Business Entities registration authority beneficial owner legal person receiver regulatory agency Financial Business & Commerce clearing service brokerage firm 19

Connecting to Existing Ontologies Contract Occurrence Written. Contract has. Part Legal. Entity Financial. Instrument Schedule Foundations identifies owns Business Entities Commitment Identifier has. Effective. Date guarantees Financial. Instrument. Identifier Financial Business & Commerce § Do not reinvent the wheel § Connect to existing FIBO ontologies Loan. Contract Credit. Check Universal. Loan. Identifier Loan Obligation. To. Pay has. Closing. Date 20

Connect to Other FIBO Modules Outcomes: § Acceptable connection found § No connection found – suggest class or property to be added to FIBO – inform other teams – perhaps create temporary version and wait to hear about the official change. § Identify issues with existing FIBO modules § Coordinate with other teams 21

A FIBO Loan Use Case: Newly Revised HMDA Rule § § Why HMDA? The HMDA Use Case The Analysis Process The Expected Results 22

Why Would FIBO Look at HMDA? § New reporting required for lenders and originators in US lending § New reporting data requirement is significantly enlarged § Reporting includes integration across business lines and revised scope § New rule adds quarterly reporting for large filers 23

Data Integration for HMDA Reporting 24

Live Demonstrations § § HMDA mapping document MISMO XML in XMLSpy FIBO Loan in OWL (Protege) Inference (time permitting) 25

The Role of Inference (live demo time permitting) § Just assert the loan to be a Loan. Contract § If it has a Security. Agreement with Real. Estate as collateral, then it is inferred to be a Mortgage. 26

The Role of Inference (live demo time permitting) Mortgage. Loan. Contract: A loan contract that is secured by real property. Security. Agreement: Agreement between two trading parties that contains information about their relative duties and rights regarding the disposition of a specified asset that has been used as collateral. 27

Wrapping Up § Extending FIBO to specific sub-domain § Progress made towards tightening up: – core FIBO ontologies – the coordination process for extending FIBO § Importance of a stable core to build on – Foundations, Business Entities, Financial Business & Commerce § Designed to be directly usable as a data model 28

Key Takeaways § MISMO and FIBO Loans can co-exist § Importance of SMEs for quality result § Value of having a good process that is continually being improved § Future is bright for pushing out more and more FIBO ontologies in specific areas 29

Questions ? For more information about the FIBO Loan Content Team: https: //wiki. edmcouncil. org/display/LOAN/FCT-LOAN Come Join Us! 30