|
For data management professionals, the most important aim of describing information needed to efficiently run a private or public sector organisation is to ensure that the definitions support the total sharing and distribution of information at lowest risk.
This conclusion was one of many perceptive and authoritative views arising from a comprehensive seminar on data modelling organised by DAMA UK entitled: ‘The Challenges for Data Modelling in the 21st Century’. DAMA UK is the British chapter of DAMA International, a respected and global professional institution representing the key interests of data resource management professionals.
Data models are used to understand and document the information requirements of a business; these data models are then used as the basis for the design of the databases at the heart of any information systems that support that business.
At the seminar, five highly qualified experts presented their views on the trends and issues facing public and private sector organisations and the individuals responsible for making decisions about the innovative and efficient use of data.
SUMMARY OF TOPICS, SPEAKERS, AND VIEWS
The first speaker, Harry Ellis, an independent consultant with over 45 years experience of the analysis and modelling of data, currently specialises in the problems associated with making information accessible to all using the semantic web.
In the seminar, he identified a number of enduring problems with the way that data modelling is practised. He suggested that our current approaches to data modelling sometimes stand in the way of achieving the most important issue facing us: the need to share information and data.
“In future, data management professionals should consider describing information needs as a series of assertions. This approach would then allow information to be distributed and shared without risk,” said Ellis.
He also suggested that the classification needs of an organisation be supported by a single global taxonomy which has the ‘class of everything’ at its apex and may be extended at will in both breadth and depth.
Taking this theme further, in the final presentation of the day, Dr Ken Allen, another independent consultant, addressed the issue from the perspective of whether the data in a data store is more or less important than the information to run the business.
Allen emphasised that modelling the information needs of the business is different from the modelling of data for a database design and that a holistic view needs to be taken of the management of all the information in an organisation.
“An organisation is interested in information, not data, and an information definition language is needed that is richer than the data modelling languages currently in use,” said Dr Allen.
To take a holistic view of information, he added, it is necessary to see the information within an organisation as belonging to the whole organisation rather than autonomous and independent sections of the organisation. At the same time users should be allowed to view information in their own terms and with their own perspectives. A single central perspective on information should not be imposed. Instead there should be an information structure that is able to meet all perspectives within the organisation.
Another aspect not often addressed in current data modelling techniques is that of temporal or historical data. Matthew West, from Shell International Limited, built on a point made by Harry Ellis and presented the concept of four-dimensionalism in data models, introducing a paradigm that sees objects extended in time.
“This concept is essential if the history of the enterprise and not just the current state is to be captured,” West said. He also talked through the theory behind four-dimensionalism and illustrated it with some real life examples.
Dagna Gaythorpe and Keith Gordon, both members of the DAMA UK committee, gave short presentations on the differences between corporate and project level data models and the impact that the object-relational features now available in database management software have on database development and data modelling.
Time invested in the development of a corporate data model can lead to faster, better and cheaper projects downstream. In this scenario, a lot of the basic work has already been done and may be reused in the knowledge that it has already been checked by other projects and is probably already in use somewhere within the organisation.
In addition, reusing data definitions reduces the amount of checking, reformatting and converting of data, leading to the development of simpler interfaces between systems when information is being shared.
The SQL standard now includes object-relational features such as user defined types and collection types. The user-defined types enable the storing of data of arbitrary complexity; data in a database is no longer restricted to simple text, numbers and dates. Furthermore, user defined types can have methods or functions associated with them that are contained within the database and available to all application programs.
“But perhaps the most significant impact of the arrival of user-defined types is that they display the object-oriented principle of inheritance and this may be used to provide direct support for supertype-subtype hierarchies within the database. The collection types can also be used to create more complex data structures,” said Dagna Gaythorpe.
Both speakers stressed that the challenge is to understand where the use of these facilities might be beneficial to a business and to make sure that they are then not overlooked through outdated data modelling practises.
www.damauk.org
•Date: 4th May 2007 • Region: UK/World •Type: Article •Topic: IT continuity
Rate this article or make a comment - click here
|
