Freedom of InformationBusiness metadata, though diffficult to provide efficiently, releases more of the potential of tabular data
By Shiraz Kassam Continued from Page 1 Observe the decision makers at work, shadow each of them for a few days, see the world through their eyes, and understand the kinds of challenges they face. Understand the kinds of decisions they make and how they make them. Ask them about their responsibilities in the decision-making process. Who makes the final decisions? Who are the primary contributors to the decision-making process? Note their roles, their "measures" (more on this later), and also their interests. This information will lead to customization at both individual and group levels. You also need to identify frequency and format. Get very familiar with how often each decision is made. Some may be made monthly while others may happen every six weeks or six months. As for format, know what level of detail the end users require and in what order they need the data, for example. Package It UpIn order to determine exactly how to bundle the data and metadata, consider each of the following points. Relevance. The data must be relevant to the decision maker. Keep in mind that the information of most interest to a decision maker is that which deals with measures. Measures are indicators by which an organization can judge its progress toward a business goal. Examples of measures are revenues, load factors, capacity usage, consumer debt ratios, and so on. The metadata must be relevant to the data it describes. Some of the data and metadata these decision makers need can be found inside the organization. But some of it has to be collected or bought from outside. The metadata you bundle with the source data should suit this kind of information. For example, specific industry reports that give an informed view would be most valuable if they allowed internal data to be viewed against a general, well-commented background. Aggregation. Both the character and the detail of the metadata you provide should be in tune with the level of data it complements. Too much is just as bad as too little. Also, remember the decision-making cycle. If the cycle time is six months, then both the data and metadata should span the same range of time. Security. Because some metadata may be sensitive for the organization (think of email, personal comments, informed opinions, business plans, and strategy), you must be careful about how it is distributed. Even if it is relevant to some set of data that a department bases decisions on, it may not be in the best interest of the enterprise to make certain metadata available to all who are interested in it. Sometimes a compromise may be struck, allowing certain information to be communicated only verbally, rather than in an easily reproduced format. Deciding how to bundle data and metadata is by no means easy. You will often need to return to previous steps to get the right information, confirm your decisions, and proceed back to the bundling step. Getting it right is an iterative process. Deliver ItI'm unaware of any electronic solution or software that supports the efficient collection or delivery of metadata. Vendors specializing in text mining, text warehousing, and search engine technologies are the likeliest source of an electronic solution. Thinking of metadata in its broader meaning — any information that places data in context to make decisions easier — opens you up to new opportunities in metadata collection and use. Presenting tabular data and context through relevant metadata will lead to a better understanding of the business climate and to timely decision making. Shiraz Kassam [Shiraz_Kassam@Merck.com] is a senior data architect at Merck & Co. Inc.
|
 |
Most Popular This Week
IE Weekly Newsletter
Subscribe to the newsletter
| |
| |||||||||||||||||||||||||||||||
