Semantics’ Seminal Moment

Financial services is an industry that reveres its watershed moments. From the Great Depression to the dot-com bubble to the financial crisis of 2008, turning points are endlessly analyzed, exploited for value and eventually, the aftermath is put under the same microscope. The scrutiny is for good reason—financial watersheds don’t just change the marketplace, they change the world. 

However, some pivotal turning points are quieter, subtle shifts that might not seem momentous, at first, but have the power to transform how the market operates. Right now, semantics are having a seminal moment.  

“I’ve never seen anything like it,” says Mike Atkin, strategic advisor for the EDM Council, when asked whether an industry-wide semantic ontology is within reach. 

David Saul, senior vice president and chief scientist at State Street, has been working on semantics, standards and ontologies for six years. He says the industry is finally at a point where the need for data harmonization is understood and managed by custodians who grasp both the risks of fractured data and the potential for semantic data standards and ontologies to prevent another financial crisis. 

“The technology’s matured, the standards organizations are working together, collaborating at a global level,” Saul says. “This convergence is really all coming together and I’m very optimistic that we’re going to see an accelerated update in adoption of semantic data standards that are going to benefit everybody.” 

Saul says the financial industry continually creates increasingly complex ways of dealing with itself and the only way things will get better is through ontologies. 

“The problem is real and getting realer,” Atkin says. “The ontologies are maturing and approaching material completion. The tools, the vendors that support this are expanding and growing up in capability. Initial use cases have all worked as advertised. Put that together and you’ve got a perfect storm: requirement—both regulatory and business—capability and maturity all coming together. And it fixes the problem.”

The Data Problem

Specifically, that problem is what Atkin calls “the fragmented IT reality,” wherein the underlying technology that drives the financial industry has been built up and is managed in silos, which results in repositories of data aligned vertically to their applications. 

“So things from data get transformed to meet the data models and software objectives of the systems—a lot of that proprietary—and all of that is designed to drive applications and we manage that vertically. If you multiply that by hundreds, thousands, ten thousands and then across the industry, hundreds of thousands of those things, you have this big transformation problem, where data that represents something precise and real, like an obligation of contract, has been transformed and renamed. Same word, different thing; same thing, different word. We have front-office systems that are fairly straightforward, that don’t contain the nuance needed for back, and that is rampant across our industry,” he says. “And that is the data problem.”  

According to Atkin, the solution is a model of reality where data’s labels are aligned to their precise meaning via standards. 

“The model reality, that’s an ontology, the reality of financial instruments and how pricing and processes work,” he says. “The standards allow us to assign meaning to the data and process it so that it’s not restricted by tables and rows and columns and relational environments that are location-based problems. It’s hard to unravel all those joins and builds and locations of all the data because it’s all over the place.”

Georgia Prothero, principal data modeler at Schroders, says semantics define the real world at a conceptual level and “an ontology is a statement about sets of things and how they interrelate.” 

According to Saul, a data ontology builds on things that have been around for a while, like taxonomies or data dictionaries, but as a complete representation of not only what the data means, but how it relates to other data. 

“For example, we can look at a financial instrument and see the shared characteristics and shared identifiers with the various kinds of financial instruments, whether they be equities, bonds or whatever, and having this degree of mathematical precision describing the data and carrying that along with the data means that when I move data internally within our company from one business area to another, or to a client, or from a client or to a regulator, they know exactly what the data means, what it represents,” he says. “It increases the level of data quality to the point that trust increases, and so you can now be sure that this particular transaction, you know exactly what’s going to happen with it.” 

Cross-Standards Collaboration 

There are a number of organizations working on developing various standards models, and harmonizing them with others: International Swaps and Derivatives Association (ISDA) is preparing a common domain model for derivatives, FIX Trading Community promotes the use of standards—including the FIX Protocol, a messaging standard for securities—to address market and regulatory issues, XBRL communicates between business systems, and Inter-bank Market Information Exchange (IMIX) is a financial standard for Chinese foreign exchange (FX) trades, just to name a few. 

Banks are getting in on the progress, working on solving the data problem through development of internal standards. Schroders’ Prothero says the asset management company recognized the importance of a semantic ontology and established an internal ontology independent of any of their physical system implementations. 

“Our semantic ontology is represented by a group-wide logical data model and an accompanying glossary,” she says. “Having an agreed internal language has made it much easier for people to communicate business scenarios, and for IT systems to ‘speak’ to each other.” 

Among the best and most prominent examples of breakthroughs and collaboration in terms of advancing standards and ontologies, however, are Financial Industry Bus­iness Ontology (FIBO) and ISO 20022. 

FIBO is a data harmonization standard developed by the EDM Council in partnership with the Object Management Group (OMG) using RDF/OWL (Resource Description Framework/Web On­tology Lan­guage). 

“We recognize in our industry that the thing that we were focused in on was the reference data about financial instruments and financial process, corporate actions and issuance and transaction pricing and things of that nature. So FIBO is the ontology for that. It’s the ontology of contracts,” Atkin says, adding that contractual obligation is a “big section” of the industry. FIBO uses XML schemas, and “much of the reference data contractual world has been defined already as part of those XML schemas. So what we’re doing is taking the knowledge that exists in those schemas and expressing it as both an ontology in the web ontology language standards. So it is aligned with all of these existing things because all of them are based on contractual requirement.”

He calls the work of expressing all of the “very difficult contractual obligations” in an ontology the “underlying foundational segment,” and says it is complete, as in, fully aligned with XML schemas. 

“[The ontology] is all standardized, it’s all been released as standards though our partnership with OMG, so all reviewed, validated, tested and we have a high degree of confidence. Then we have domains. Equities, bonds, mortgages, interest rate swaps and derivatives to some degree—those are all complete, tested and we have a high degree of confidence and they’re being used,” Atkin says. “It is not a complete ontology, it’s not finished, but it is pretty damn good. And if you look at what we currently do, which is mostly communication types of things, trade confirmation types of things, and we use all these XML schemas for that, much of the reference data contractual world has been defined already as part of those XML schemas. So what we’re doing is taking the knowledge that exists in those schemas and expressing it as an ontology in the web ontology language standards. So it is aligned with all of these existing things because all of them are based on contractual requirement.”

Now, he says, the work is on development projects in RDF/OWL format, so far untested for conformance and in need of further small and medium-sized enterprises (SME) reviews.

“Some of them are really good. Some of them are fairly good. Some of them are just conceptually good. And that’s the work that has to be done to materially complete,” Atkin says. So now that the contractual concepts are defined, the debate is focused on how best to structure them, and how they relate to each other. “We’re a consensus organization. If anyone finds anything that they think is wrong with the way we’ve captured the ontology, they are encouraged to challenge what’s been done. We’ll open up a ticket audit, we’ll convene a discussion, we’ll explore. We get things wrong all the time.”

The Big Breakthrough

As part of this collaborative process, the EDM Council is an active voting member and participant in the TC 68 working group “designed to bring all of these niche standards together to create a semantic model for the industry,” Atkin says. 

TC 68 is one of several hundred technical committees formed under the International Organization for Standardization. The first committee, TC 1, was formed in 1947 to standardize screw threads, and active committees focus on developing standards for everything from toy safety to fine ceramics to internal combustion engines. TC 68 covers financial services through the work of three subcommittees: reference data standards, information exchange and information security. ISO 20022 is one of the standards authored, supported and maintained by TC 68. 

ISO 20022 was first published in 2004 as a mixture of international standards and went back into revision “to progress the work on technical specifications so that they could be reissued as international standards,” says Karla McKenna, director of market practice and standards for Citi Markets and Securities Services, and chair of ISO/TC 68. “Since then, it’s come up for revision again. The current revision that’s going on is adding semantic capability to ISO 20022.” 

ISO 20022’s core messaging syntax is XML, so it’s extensible, as opposed to fixed-length code qualifier tag pairs found in past transactional and messaging standards. “It takes it up a notch as far as being able to be adaptable. It is a recipe for putting concepts into a common repository. You can use the same recipe for more than one business domain,” McKenna says. 

McKenna describes a “shift in focus” among the committee as it “worked with the standard more and observed the industry around us. This is where the motivation for the introduction of semantic capability into ISO 20022 came from, where the emphasis is now on interoperability with other standards. There are other standards that have grown up within financial services, and we were looking through semantics to be able to make sure that they interoperated or related to or could leverage in some way ISO 20022. This is where the kernel of the idea and motivation came from.” 

Initially, the committee attempted to harmonize and incorporate different standards’ concepts into ISO 20022, but now they are building a semantic representation of the ISO 20022 metamodel. 

“The current effort is not focusing on trying to be able to map at the business domain level, the business concept level, but more at the semantic and format level,” McKenna says. “The ISO working group has created a tool to normalize other standards into a common semantic format that is derived from the ISO 20022 metamodel itself. So it shows us the equivalent concepts of ISO 20022 in other standards, but also, because these standards have grown up to be able to execute and serve different purposes, it also shows us enriched concepts that they carry in the standard.”

This cross-standards work is not a mere kumbaya moment for data professionals: It is a core component of TC 68’s strategy to advance and enrich the standard. 

Jim Northey is a technical committee co-chair for the FIX Trading Community and will replace McKenna as chair of TC 68 this coming January, once her term expires. He says TC 68 “had a bit of a breakthrough” when investigating whether semantics would help map ISO 20022 across different standards, resulting in the shift in focus McKenna described, which led to work toward a semantic version of the ISO 20022 structure, executed so that other standards could be represented—as they currently exist—in that format. 

“We’re not expecting any of these other standards to change their formats,” Northey says. “All we’re asking is give us a translation into this independent semantic format and then we can start to use all of the semantic tools for querying, knowledge discovery, automated machine learning, against the standard itself to start to identify new facts and new knowledge.”

He calls it “comparing apples to apples” and says if the plan sounds simple, that’s because it is. “It wasn’t so obvious until we thought of it and then, like many good ideas, seemed obvious.” 

Before the breakthrough, Northey says, FIX spent years and “quite a bit of money” attempting to map the FIX standard into the ISO 20022 model, a process he says was “very expensive and frustrating because things didn’t quite align, and once it was mapped, it wasn’t really useable by any practitioner.” 

The committee found that when taking a semantic approach, the representations are “more extensible and powerful tools where we can actually start to look at and compare and start to use them,” he says, adding that TC 68 plans to bring in other reference data standards. “We have a number of other industry standards we want to get represented in this format so that then we can start to do work in terms of interoperability, improving processing.” 

No VHS vs. Beta 

A combination of bank needs, regulatory requirements, industry collaboration and ontology creation and implementation is what will move semantics toward becoming a workable, robust solution, says EDM Council’s Atkin. 

“A lot of people are working on this,” he says. “If you look at any bank, most of them participate in some form of conceptual modeling as part of their processes. Conceptual modeling just means, do we understand what these things are, how they work, can we model them so we can implement them in our environments? [FIBO] is a standard conceptual model. A lot of people are working on the same problem, and now we’re just trying to collaborate and bring it together into a single view.”

Schroder’s Prothero underscores Atkin’s view of how the industry will collaborate, noting that Schroders’ glossary “has been cross-checked to FIBO and wherever there is a match the FIBO definition is used and a citation is made to the FIBO term.” 

She predicts the first adopters of an industry-wide semantic ontology will be vendors, “ahead of take-up by bigger organizations.” 

The way Atkin sees it, current initiatives fall roughly into three camps. First, regulators collaborating with each other and issuing requirements documents, essentially stating, “here’s what we regulators need from the industry so we can do our job, and we need it to be consistent, so we’re going to specify what that is.” The second camp is a charge led by the Financial Conduct Authority (FCA) toward smarter regulatory reporting through technology, and their “better way” matches the views of the EDM Council: an ontology “with regulatory business rules, specified in machine-executable language, so that there’s no confusion about meeting, and we can all adopt standards and move to the next stage,” he says. The third camp is banks saddled with “environments that do not allow them to integrate, link and leverage” their own capabilities, and they want to fix that. Both the regulator and bank problems can be solved the same way, Atkin says, with “ontology plus standards plus executable business rules, and that’s the direction we’re all covering, moving in.” 

In the meantime, he adds, “it doesn’t have to be perfect to be functional.” 

It’s an unprecedented level of collaboration, and State Street’s Saul says it’s just the beginning. 

“I’m very optimistic that we’re going to see more collaboration rather than splintering into multiple competing standards,” he says. “This is not going to be, in my opinion, VHS and Beta. It’s going to be more like the DVD standard where all of the industry converged on one.”