The EDM Council freely provides access to all published resources from this website for member firms and firms interested in future membership. Please register to get a web access ID and password.

3-5 November 2010 Radisson Blu Portman Hotel London, UK

EDM Council BLOG The Semantics Repository - What's it For, Anyway? With the Semantics Repository rapidly reaching beta status, more and more people are asking the obvious question: what can we do with it? How does it help us solve real data problems? There are several ways to make practical use of the Repository, but first let's look at why it was originally built. The original impetus for the Semantics Repository came out industry discussions on the need for business precision in the terms and definitions used to describe reference data. As often happens, technologists took ownership of the problem and produced things like the MDDL market data schema and the ISO 20022 Financial Instrument data model. Both of which were good technical structures, but did not result in a common set of terms and definitions at a purely business level – which was at the heart of the problem that the industry was seeking to address. The Semantics Repository was created to address that original problem and establish a standard set of securities terms with business definitions, decoupled from any technology design. So far, so good. We have delivered this, but now how should the industry make use of it? There are several ways to take advantage of the Semantics Repository. The first would be to use it as part of a managed development process. The goal is simple - get rid of those pesky ad hoc spreadsheets. You know the ones that are created when developers set out to design a data model and try to capture the business view of the world in a spreadsheet of terms and definitions. So instead of dozens of unconnected spreadsheets, use the Semantics Repository as the common source of terms and definitions. It can be downloaded as a single spreadsheet or individual section sheets and can be used to replace the less formal spreadsheets as well as being extended locally with terms and definitions that are not in the standard model. The Semantics Repository can also be used as a common point of reference for data mapping. At present, when someone maps between two or more data resources, the assumed meanings of the terms are defined only inside the brain of the person doing the mapping, so there is no oversight or maintainability of this information. In addition, if you are mapping to multiple data formats, it makes more sense to use one common set of terms as the "Hub" and save yourself lots of time and money. There are some more advanced ways of using the model content. How you approach this depends on what you want to do with data. Many are looking at the Repository for model driven development (to develop database or message schemes) while others are looking into semantic technology knowledge base applications. To illustrate … the data about real things (for example IBM Ordinary Shares) has to be formatted as "instance" data in a database, or as an "Individual" in a knowledge base – but there is no obvious reason to do both. To address existing data problems, we suggest you use the semantics repository as part of a model driven development approach. This is where development of databases and other technical resources is formally structured with physical models, logical models and, to drive it all from the business end, a business conceptual model. In this approach, the Semantics Repository provides a conceptual model for data elements. In the model driven approach, the first step would be to transform the Semantics Repository content into something that can be used to derive a logical data model. For example, there are some simple transformations that would result in what we call a Semantic Data Model. This is a technical model with the same content as the Semantics Repository but without all the extra information that determines the meanings. We are working with some of our early "Proof of Concept" partners to identify the precise rules for these transformations. The resulting model can be used as the basis for more detailed data model design. Alternatively, if performance requirements allow, it is possible to derive physical message formats from this directly and to create a common message architecture across the organization. The semantic technology approach is a lot more novel. It involves creating a platform where the terms and definitions and the data itself are maintained in a single repository. This requires a different platform and architecture to a conventional database, using something called a triple store. We are thinking about providing a triple store version of the Semantics Repository in the future based on one of our ongoing proof of concept activities. The benefit is that a semantic knowledge base gives you all the advantages of the latest semantics search capabilities, enabling users to search based on the meanings of terms. We encourage you to explore this new technology to do new and interesting things. For example you could develop a graphical interface so that a risk manager could dynamically browse between instruments, pools, counterparties, collateral and so on, by clicking from one "Thing" to the next on a diagram. Another approach we get asked about is the use of a metadata (data about data) repository. However, there are two kinds of data about data: data about the quality of data (provenance, confidence levels and so on), and data about the structure and nature of different data items along the lines of the model driven approach described above. Getting this right requires detailed understanding and should not be approached as a purely technical venture. The metadata structure (or "metamodel") of the Semantics Repository is freely available and can be used to define the semantic level in such a repository. In addition, there are now some metadata repositories that explicitly use semantics technology. At its simplest then, the choice is between using the Repository content to solve the data problems you already have in the firm, or exploring the new possibilities that are enabled by using the latest semantics technology platforms. The new technology opens up a lot of new opportunities, but it might be best to start by solving existing problems - learning to walk before you can fly. In future blogs we will look at some of these possibilities in more detail. Meanwhile do please let us know your successes and disasters so we can help you along the way, and improve our Semantics Repository and documentation to make this as easy as possible for everyone. Mike Bennett Head of Semantics and Standards, EDM Council March 5, 2010 By: Mike Bennett | 05 Mar 2010 | Comments: 0 |   ARCHIVES - 2010 - 2009 - 2008 - 2007 - 2006 - All SYNDICATION    COMMENTS

© 2010 EDM Council, All Rights Reserved