Thursday, November 02, 2006
More on the Semantic Web
I'd like to expand on my last post. Consider the following:
Or this:
And compare it to this:
The first two examples are pretty easy to interpret because they stick to conventional methods of representing data. This is done through a number of ways: proximity, position, shape, color, etc. In the first example, each dollar figure is associated with a unique year and quarter, simply because of it's position in the spreadsheet. The icons in the second diagram are associated with their labels through proximity - we can easily see that the computer in the upper right has an address of 10.0.0.2, and is not a router.
The third picture is tougher. All we can really say is that it is list of alpha-numeric strings with one highlighted. What does the highlight mean? What is the significance of the numbers? There is no way to tell because there is no context available to give relevance.
These examples illustrate the difference between data and information. Our brains look at the first two diagrams and immediately recognize a spreadsheet and a network diagram. It is our own familiarity with these methods of representing data that allows us to extract information from what is actually a grouping of letters, numbers and symbols on a page. Without that familiarity, the first two diagrams would be as incomprehensible as the third.
This is why I'm so exited about the Semantic Web. I see it as a way to give data true meaning, independent of human conventions. Nowadays, data within a system is represented according to conventions known only to that particular system itself. Any exceptions to this are far and few in between. One system's representation of a telephone number will not be recognized by another. In fact, a telephone number wouldn't even be recognized as such, just as a sequence of numbers.
Now - what if a system could truly understand the data that is was processing? What if all of the atomic bits of data floating around an organization - part, invoice, job and all kinds of other numbers, names, addresses - could actually be recognized for what they are? This is what the Semantic Web does - it identifies data as information. It's a foundational building block that could take us beyond the simple storage, transport and presentation of data to true understanding of what that data means.
Or this:
And compare it to this:
The first two examples are pretty easy to interpret because they stick to conventional methods of representing data. This is done through a number of ways: proximity, position, shape, color, etc. In the first example, each dollar figure is associated with a unique year and quarter, simply because of it's position in the spreadsheet. The icons in the second diagram are associated with their labels through proximity - we can easily see that the computer in the upper right has an address of 10.0.0.2, and is not a router.
The third picture is tougher. All we can really say is that it is list of alpha-numeric strings with one highlighted. What does the highlight mean? What is the significance of the numbers? There is no way to tell because there is no context available to give relevance.
These examples illustrate the difference between data and information. Our brains look at the first two diagrams and immediately recognize a spreadsheet and a network diagram. It is our own familiarity with these methods of representing data that allows us to extract information from what is actually a grouping of letters, numbers and symbols on a page. Without that familiarity, the first two diagrams would be as incomprehensible as the third.
This is why I'm so exited about the Semantic Web. I see it as a way to give data true meaning, independent of human conventions. Nowadays, data within a system is represented according to conventions known only to that particular system itself. Any exceptions to this are far and few in between. One system's representation of a telephone number will not be recognized by another. In fact, a telephone number wouldn't even be recognized as such, just as a sequence of numbers.
Now - what if a system could truly understand the data that is was processing? What if all of the atomic bits of data floating around an organization - part, invoice, job and all kinds of other numbers, names, addresses - could actually be recognized for what they are? This is what the Semantic Web does - it identifies data as information. It's a foundational building block that could take us beyond the simple storage, transport and presentation of data to true understanding of what that data means.
