v3.0.4/Information data pack example ?

Post Reply
jmk
Posts: 137
Joined: 31 May 2012, 12:08
Location: France

I have some difficulties to grasp the differences between data/information concepts and I did not find an example architecture for the information data pack in the download area.

Do you plan to release such an example ?
User avatar
jonathan.carter
Posts: 1087
Joined: 04 Feb 2009, 15:44

We do have plans to provide some examples and I appreciate that the distinction - whilst very important - can be quite subtle.

We will keep you posted on the example models but in the meantime, I will explain some more about the Information and Data classes.

Firstly, we use a common definition of Information and Data that describes Information as Data in context. The essence of this is that data has the same semantics and meaning regardless of where or when we use it, e.g. Product data is always about a specific product, with its attributes always meaning the same things no matter when we use it. In contrast, if we consider Stock information about our Product, the value of the Stock depends on the particular product, the location at which the stock is being held an so on.

Data elements are used as the building blocks to create/deliver the Information elements and it is typical to see a fairly small number of data subjects in the repository with there being more Information things than Data things.

The meta model provides you with the option to capture the attributes of the Information or Data elements that you need but if that level of detail is not of interest, you do not need to do so.

We have some powerful relationship classes for accurately modelling what is going on in the Information Architecture and these can be used at either the object or the attribute level. These are used to describe how applications and processes use information and from these relationships, the data that is being used to deliver that information. However, the forms in Protege should keep you pointing in the right direction.

Hope this helps you get started with Information and Data

Jonathan
Essential Project Team
jmk
Posts: 137
Joined: 31 May 2012, 12:08
Location: France

Jonathan,

thanks for your answer, the tutorial is much clearer now ( not because of bad writing but just because I'm not a native english speaker :).

Going back to your example (taking product/warehouse/product_location) : I understand that a good use of the data/information hierarchies would be to model "product" and "warehouse" as data and "product_location" as information.

Am I right ?
User avatar
jonathan.carter
Posts: 1087
Joined: 04 Feb 2009, 15:44

Product would be data, as would Warehouse and I think that 'product_location' would be an attribute of Product that might be reference to e.g. a Warehouse or a Bin (data things).

I sometimes test candidate data things by thinking about whether we would need to manage them in terms of master data. e.g. does it matter if I create two instances of Product data in two or more different places? Could this affect the quality if I have two different people creating data about the same Product instance in different systems at the same time? Probably, yes!

Typical examples would be Product data, Customer data, Account data (from the finance domain). Of course, things like transactional data (Purchase Order, Sale) are not covered by this test.

In my example, the Information would be "Stock in US Warehouse today". This information is using Product data, Location data, Time data and so on to give us the stock information for our products in the US warehouse. If we wanted to know the stock level of a particular product, that might be a different Information element, "Historical Stock of X in US Warehouse", again using Product data, Location data, Time data and so on.

As a general heuristic, the Information things are mostly what the business people need to know. That's why our Applications and Business Processes deal with Information things in the meta model, not Data, when they are interacting.
Your standard reports would be Information things, the BI instruments would be Information and in each of those, we can describe the Data elements that are used and any calculations that are performed to produce an Information Attribute from those Data elements. Modelling it this way helps us easily understand the impact of a change to any Data in terms of the Information things like the elements of our reports, and real time information that the business is using.

Think about the Data elements in terms of being the building-blocks that we use to assemble the Information elements. Information things use Data (either directly or indirectly through a computation) in order to produce what is required.

When trying to identify Information elements in your architecture, it is often good to start with questions like "what are the reporting requirements?" for the area we are modelling. That is, "what Information is required or being used?". Once you have those, if you need to understand the Data that is delivering that Information, you can then capture the Data things that are supporting each Information element.

Jonathan
Essential Project Team
jmk
Posts: 137
Joined: 31 May 2012, 12:08
Location: France

Jonathan,
thanks for your detailed answers.
Post Reply