Data Abstraction for Data Citizens

Executive Summary

This webinar covers key concepts and practices in data modelling, and Howard Diesel makes the case for how central they are to data governance and to running a business effectively. He works through various modelling techniques and why documenting requirements matters, so every stakeholder ends up with a clear picture of the data structures involved and what they mean in practice. The webinar also covers building a comprehensive business glossary and the role visual communication plays in keeping teams collaborative and on the same page. Bring all of that together, and organisations get stronger data management and better-informed decisions.

Webinar Details

Elements of Abstraction in Data Modelling

Howard Diesel pointed out during the discussion on abstraction that there are varying degrees of it worth considering, and while abstraction can genuinely help, it’s worth distinguishing it clearly from generalisation. Remco illustrated the difference using party and role as abstract and generalised terms in a technical solution, but cautioned that this kind of abstraction doesn’t always translate well when communicating with business personnel. Both Remco and Howard agreed that abstraction is hard to understand, build, and maintain, and its value needs weighing carefully rather than assumed. Howard’s suggestion is to start the modelling process by describing the company’s requirements and characteristics first, matching the level of abstraction to what the customer actually needs.

Notes on Business Discussion

Finding the right level of detail and focus when analysing data matters a lot, not unlike adjusting a camera’s zoom. A broad overview gives you understanding, but too much detail can quickly become overwhelming.

Data privacy means treating everyone the same, whether they’re a customer or a vendor, and in the event of a breach, every data subject needs to be communicated with individually.

Communicating well with businesspeople at the start of a project comes down to agreeing on focus, terminology, and the level of information people actually want. This particular discussion centred on data breaches and reporting rather than other assessments or business events.

Data Modelling for Data Governance

Data models matter a great deal for data governance and can factor directly into a data governance assessment. The CDMP® exam covers subtypes, supertypes, and conceptual models, along with questions on the benefits, uses, and differences between conceptual models. Learning data governance concepts is genuinely valuable for personal and professional growth, something Patricia noted is an ongoing process.

Hesham asked about making data models more agile, so they can adapt to future changes, and was after best practices, methodologies, or frameworks that would help make models more adaptable.

Different Modelling Techniques

Howard brought up a presentation on Data Vault during the discussion, pointing to its agility and core concepts. Anchor Vault modelling is known for being especially agile and often gets called an additive model, though breaking apart and rebuilding under abstraction can get complicated depending on how the model was originally designed. Which technique makes sense really depends on the application environment and the goals at hand. Software vendors tend to prioritise abstraction and flexibility, but that’s not always necessary if it doesn’t actually add value to the business. Abstraction suits operational or module-based applications like ERP or CRM systems well, while dimensional or flat file modelling tends to be preferred for integration in data warehousing. That said, when frequent changes are expected, Anchor modelling is the better fit for data warehousing. And getting the “camera settings” adjusted right at the start of a project matters a great deal for how well it turns out.

Importance of Documenting Requirements in Data Modelling

Documenting requirements is what keeps a model’s direction clear, and an Excel spreadsheet can hold onto those “camera settings” as a reference for whoever picks up the modelling work next. Remco has been encouraging Hans to write up the different warehouse layers and their requirements, and Hesham sees real value in data governance and management here. Working a business glossary into a data model can help get everyone to agreement, and a scorecard showing the relationships between glossary terms is a useful addition. An image-based approach also tends to simplify communication and make signing off on business rules easier.

Building a Business Glossary and the Importance of Visual Communication

The Financial Markets Department was advised to start with a glossary, and Howard and his team spent a full weekend compiling around 3,500 business terms. The glossary presentation itself, though, didn’t generate much enthusiasm or discussion. To get past that, Howard suggested breaking the glossary into sections and offered to build a data model instead, which sparked some genuinely spirited debate. Hesham favours visual communication and is good at presenting information that way, and Remco leans on drawings and models during workshops to clarify ideas and get discussion going, stressing how much visual aids like PowerPoint and icons help convey complex ideas. The limitations data modelling tools have in representing concepts visually came up too, and were addressed head-on.