The Business Value of Semantic Technology Chris Moran CTO, Information Management Solutions Consultants, Inc. [email protected]

© 2013 Information Management Solutions Consultants, Inc. Distributed by: SemanticWeb.com

Introduction If you don’t understand what your software engineers are talking about, perhaps it’s because they are using a vocabulary they invented for the problem they are solving. Engineers invent a vocabulary and data structure for each system they build and each problem they solve, and only the engineers who built the system understand this structure and vocabulary. Even other engineers must learn it in order to make the data usable. In most enterprises today, we have as many different ways to ask questions of our data as we have systems to store it. We have as many different vocabularies and data structures as we have systems. The problem is actually worse than it sounds. If we want to bring data together from many different systems or take data out of one system and put it into another one, we need to understand the vocabularies and structures of each and every system involved. That can be very difficult and time-consuming. The meaning of the data is supplied by the program(s) an engineer writes for the data. So, a new engineer looking at the data must generally understand the program in order to fully understand the data. Effectively, each system we build becomes the fiefdom of the engineers who build it. And each system becomes a silo. Combining data from multiple systems requires the time and cooperation of the engineers who maintain each system involved. This isn’t deliberate on the part of engineers. It is a consequence of the way we have designed systems over the past twenty to thirty years and the technologies available at the time. Semantic technology solves this problem by embedding the meaning of data in the data itself and by making it possible for different systems to use the same meaning and the same vocabularies. In traditional systems, sharing vocabulary and meaning is not practical. We must ask questions in terms of the structure of the data and the structure of the data comes directly from the problem the system has been built to solve. Each problem is different, so each structure is different. In contrast, semantic technology is not based upon data structure. In fact, semantic data has no pre-defined structure. What structure there is comes from the data itself and the relationships between the facts and things in the data. Today, very mature and established tools and methodologies exist for building systems entirely with semantic technology. If that sounds futuristic, consider that if you call the customer service to complain about your phone bill, chances are good that that representative is using a business system built upon semantics. Or, if you use Google to search for a new cell phone, chances are you will be given a list of web pages that have been encoded with semantics. Google, Best Buy, and other large companies have already adopted semantic technology and are pushing it into our daily lives.

The Business Value of Semantic Technology 2

Google has become a lot better at providing search results? That is semantics in action. Semantics enable these businesses to combine information much more quickly and economically.

What is Semantic Technology Semantic technology is based upon data stored in a graph, referred to as graph data, and a description of what that data means. As the name implies, semantic technology stores meaning with the data. It also removes the need to define structure, or data models. There is no predefined structure to which the data that is described semantically must be bound. This provides tremendous flexibility in what information can be stored, and it enables information to be combined and used both rapidly and in ways that are not possible with relational or traditional XML technology (structured data).

The two figures above illustrate the difference. The table on the left displays data as stored in a relational database—structured data. In order to turn this structured data into information, an engineer must write an application to interpret it, such as what a status code of “3” means. The graph on the right shows a piece of information that says Account 88234 has the domain mysite.org. The meaning of the data on the right is stored with the data itself, thereby making it information. And if we want to improve upon our understanding of it, we can simply add more information. It isn’t necessary to redesign a data model—there is no data model. To use this information, an application can simply consume it. An engineer does not need to write code to tell the application what it means or how to consume it. Because it contains meaning, it is information, not merely data.

The Business Value of Semantic Technology 3

Computer systems have traditionally been built upon data models. The models are built to satisfy the requirements for a specific system that have been developed in response to an identified need or problem. A data model serves to constrain data to a certain structure, and application logic is built upon this structure. Meaning comes from the application logic. There is no meaning stored with the data itself. There is, in fact, no information until the data is consumed by the application. In enterprises today, most of the data is siloed—because only the application(s) built specifically for it can turn the data into information. Although no engineer sets out to create a silo, silos are the unavoidable consequence of data models—even standard data models. Semantic technology avoids this consequence by putting meaning into the data. Rather than data models, the data is stored as a graph and the graph is self-describing. The only role an application has is to query and serve up the information to a user. The application is no longer a silo because the meaning of the data is not supplied by the application.

The Cost of Data One consequence of traditional data modeling is that in order to use the data to meet a variety of needs, it must be stored in a variety of different structures and must be constantly The Business Value of Semantic Technology 4

translated, copied, and kept in sync. One user’s need is almost always at least slightly different from another user’s need, and need and intended use affect structure. Data and data modeling incur very high engineering costs. Most of the cost of systems today and most of IT budgets come not from modernization or new development but from maintaining data and the applications that make the data usable. It comes from maintaining all the different copies of data that are created for each intended use. It comes from maintaining all the code that copies data from one system to another. It comes from maintaining a staff whose purpose is simply to understand what the data means and how it is shuttled between systems. And it comes from the need to change a myriad of structures to accommodate any new information needs. With semantic technology, how information is used does not affect how it is stored. The use does not affect its structure. There is no need to keep the information in a myriad of structures in order to satisfy a myriad of intended uses. It can be stored in one place to meet any number of intended uses. Semantic technology removes the need to move data in and out of authoritative sources. The data itself becomes authoritative information in the one place that it is stored. Because all applications can use the same piece of information, semantic technology removes the need to maintain a staff whose purpose is simply to “keep the silo operating.”

Information Information is one of the largely unsung inventions of science fiction. Flip phones, lasers, rockets—all accepted without a blink. But whenever we, as engineers, watch the hero effortlessly consume the information from a database, we wonder how that was possible. There was no design stage, no analysis, no development, no compiling, no testing, no debugging. It simply worked. This is not possible when an application is built to consume data. But it is possible when an application has been built to consume information. Once an application has been built around information, the mechanics of consuming it never change. It is the mechanics of consuming data that require so much design and development, so much analysis and testing and debugging. Information has tremendous value. The business value of semantics is the value of information itself and the reduction in the cost of using that information. The value of semantics is in all of the new uses to which the information can be put to when we no longer need to invest in operating the silos. It is a reduction in complexity, a reduction in operating cost, a reduction in the sheer amount of storage and computing capacity, a better use of talent, and a leap forward in our ability to further automate what we do.

The Business Value of Semantic Technology 5

Conclusion If you don’t understand what software engineers are saying, perhaps it’s because they are not creating information. They are engaged in the mechanics of designing, building, and integrating the data stored in their silos. Systems within most enterprises today contain an enormous amount of siloed data and relatively little usable information. This makes interoperability and maintenance both expensive and complex, and it makes modernization and the realization of new capability elusive. Semantic technology offers a solution to these problems. It makes information available and less expensive, it enables more and better use of information, and it enables new capability and real modernization.

The Business Value of Semantic Technology 6