|
Allow me to frame my perspective: I am a system administrator who also programs applications for the hospital department for which I work. I work in an environment that is thoroughly mixed, with Mac, Windows and *nix machines all over the place.
I have a fair bit of experience with relational databases. I've read Chris Date. I have some, though not extensive, experience with XML. I've used XML as a data interchange format, and of course have worked with XHTML.
I understand the RDF data model, and appreciate greatly the fact that it *has* an underlying data model. But I don't yet see a killer app. I don't see what value RDF brings to the table. I suspect my own ignorance is the major fault, but still, I have yet to see a compelling use of RDF in the wild. Lack of marketing, maybe? I don't know.
Okay, my bottom line is pragmatic: I use tools, and for me, in certain situations, XML has great value as a tool. That value comes primarily from the ubiquity of XML-oriented tools, and my ability to get data into and out of an XML format easily in any of the platforms in which I commonly work. Platforms, in that last sentence, should be taken in a broad sense -- I work in Java, Python and VB, on Windows and Mac clients, with Windows and Linux servers, using lots of different database engines.
Having a good model for one's data is generally very important in my work. In my case, that model generally is a database schema and an associated object model. To my eyes, modeling something in an ERD is not much different from modeling something in RDF. Different in the details, certainly, but not in the fact that you are modeling the relationships and constraints of sets of data. (Or did I miss something really large in my cursory overview of RDF?)
Having a model is important. Getting all the interested parties to sign off on that model is also important, and often is more work than the model. What comes after that is the pragmatic choice of tools. I can express my data model in the form of a relational schema, or I can choose to model in RDF and OWL.
If I choose to use a relational database, I know that I can easily manipulate my data, searching, inserting, changing and deleting. There are many-many tools available to me that make my work easier, in terms of helping the end-user create, record, analyze and retrieve their data. Furthermore, general knowledge of databases is widespread in IT, and support is easy to come by. Connectivity options abound. These are all big value-adds for me choosing to use a relational database as part of my problem solution. These are parts of the wheel I don't have to reinvent. (When it comes time to program, the object-relational impedance factor causes me heartburn sometimes, but... :)
Now, RDF and OWL provides a useful data model. Since most of my problem domains don't inherently involve document mark-up (and thus dictate an inherently XML-centric approach), what value do RDF and OWL bring to the table?
From the looks of it, using RDF and OWL would mean reinventing a very large number of wheels. In my case, my typical use of XML involves adding XML translators to an existing application, in order to allow easier access to the services provided by the application, or to enable my application to use other apps' services. I can see that there may, somewhere down the road, be a need for something similar for the semantic world. TimBL's semantic web vision is compelling. But it isn't really here yet. And no, the weak form doesn't count.
So, where's the value to my work? What do RDF and OWL give me that I can't get elsewhere, with less work? Why is this important to me, the scientific/corporate developer?
|
Previous Message
