Getting To Grips With Semantics

Semantic technology is being used by data managers in a variety of industries, from biomedicine to social media, where its agility and ability to classify and integrate data is proving valuable. Advocates of the technology say the financial services industry can also benefit. However, they face a challenge to get their message across and convince often conservative, slow-changing financial firms that the new approach is worth adopting.

Semantic technology employs machine intelligence to form an underlying basis for describing concepts and data, using sophisticated schema called ontologies. The concepts are understandable to both humans and machines, and can be used to link, integrate and classify data. In the travel industry, semantic technology and natural language processing are being used to characterize the comments customers leave on websites about destinations, hotels and restaurants, so the websites can be personalized to improve their experience.

In the context of the financial services sector, experts say semantic technology can be used to classify financial instruments, such as interest rate swaps. The defining characteristic of an interest rate swap is the fact that it has legs that are based on an interest rate. If data managers embed this knowledge in an ontology, semantic technology can be used to classify data based on this attribute. An internal program called a reasoner can analyze data and make logical deductions and classifications. If the reasoner encounters a swap with one leg that is tied to a fixed rate and another leg tied to a floating rate, it can infer that the swap is a fixed-float swap. If a swap has two legs attached to a floating rate, the reasoner will classify it as a float-float swap. And so on.

The Concept
“With semantics, the key thing is to define conceptually what something is, not necessarily what we call it. When we understand a derivative by virtue of its attributes, we can call it many different things, and semantically we will discover what its true state is because we understand the concept surrounding the definition of that instrument,” says David Newman, Wells Fargo’s San Francisco-based, vice president, strategic planning manager, enterprise architecture, who is also chair of the semantic program of the Enterprise Data Management (EDM) Council, a trade association.

Newman says semantic technology will bring greater clarity to financial data standards. “It will be attractive to financial institutions and to regulatory institutions because it is a very effective way of defining data standards, with the concept represented in the ontology rather than the concept and operational concepts being described in application programs,” says Newman. “Semantic technology will allow us to move the logic that was hitherto in a cryptic application program into an enterprise model, so that it becomes much more efficient. It is unambiguous. There is a common view of the concept.”

Legal Entities
The EDM Council is working with the Object Management Group, a computer industry consortium with expertise in standardization, to develop a semantic representation for legal entities and interest rate swaps before moving onto other derivatives, such as credit default swaps. It also plans to cover other assets, such as mortgage-backed securities and loans.

Industry groups are not the only ones exploring the potential of semantic technology. For the past year, SunGard Consulting Services, the consultancy arm of the technology vendor, has been doing proof of concept (POC) work for data management systems that use semantic web technology. Yingxia Wang, New York-based senior manager at SunGard Consultancy Services, says the POC work is now going to be shown to both internal and external clients, and the company is then hoping to start implementing semantic solutions for its clients.

Semantic technology will allow us to move the logic that was hitherto in a cryptic application program into an enterprise model, so that it becomes much more efficient — David Newman, EDM Council

Wang says SunGard’s research has highlighted a number of advantages of semantic technology. She says it can improve agility and responsiveness, because the data model is extendable and less rigid than the model for traditional relational databases, allowing changes to be made to a data model without the need to redesign it from scratch.

Wang says semantic technology can give users a more comprehensive view by integrating large amounts of data at a semantic level, and provides a framework that can be used to remove redundancies in current data management processes by encouraging practitioners to share and reuse data. “Based on my experience, almost all the projects I have worked on have reference data,” says Wang. “Some of it is overlapping and some of it is unique to the project. If you can get the common component then make that a separate piece, you definitely reduce some of the duplication of work for each project.”

Brian Sentance, New York-based CEO of technology vendor Xenomorph, says advocates of semantic technology tend to focus on how it can improve agility. He says this is an important requirement for many practitioners today who want to be able to respond to changing business needs quickly, but companies such as his are already able to provide the necessary flexibility today without the use of semantic technology, by creating a separation between the end-users’ view of the data and the way the data is actually stored.

Complex Searches
Sentance believes there are other more interesting benefits of semantic technology that are worth further exploration and promotion, including the ability to conduct more complicated searches of databases and to establish relationships. He says Xenomorph is considering integrating semantic technology into its products.

Wang believes that at present, online companies such as Facebook and Google are leading the way in their use of semantic technology, but says she has seen tier-one firms in the financial sector doing POCs, starting with small-scale implementations. She hopes to see semantic technology become widely used in financial services in the coming years.

Wang expects semantic technology to complement existing technology, rather than replace it, in what she describes as a data management “evolution” rather “revolution.” “A lot of the technology in use today is there for a reason—because it is good at something,” she says. “For example, relational databases are solid, they are good at managing one type of data very well and they do a good job. So we don’t see the existing technology going away—we just believe the new technology will add to it.”

Implementation
Newman says the EDM Council is conscious of making the adoption of semantic technology as simple as possible, avoiding a “big bang” approach. He says that by using basic mapping, a common semantic platform can be “injected” into institutions and regulatory agencies without the need to change any of their legacy systems or legacy databases. This semantic system would act as a filter to ensure all the data is correctly aligned semantically with a common standard so everyone knows they are using the same terminology.

However, the financial sector is inherently conservative in its attitude to adopting new technology, with firms often preferring to leverage existing technology and take a wait-and-see approach before deciding to implement a new solution. Sentance says that if semantic technology is going to catch on, the academics and other experts in the field need to do a better job of explaining how semantic technology can benefit end-users and how it can help them to do things they cannot already do. He complains that when he attended a conference on the subject earlier this year, many of the speakers failed to explain clearly the role semantic technology played in achieving impressive results—a fact that he found particularly ironic, as semantics is supposed to be about providing greater clarity and meaning.

Sentance says end-user firms spend a large percentage of their budgets on just keeping their businesses going and they need to see clear, compelling business benefits before they will consider investing in semantic technology.

It would seem therefore that despite its potential, experts have some work to do before adoption of semantic technology becomes commonplace in the financial services sector.