Almost every conference or talk I have attended lately about semantics, ends up with the following question when RDF is the technological solution used: “Well, you could do that with a traditional Relational Database, right? So why do you use RDF?”.
I do not consider myself at all a Semantic Web expert, but I have been around for a while, so I will try to give my two cents to that:
1- Relational Databases have been for 60 years out there. They made our life easier and much better. Fair enough. However they have not some of the fundamental issues that remind unsolved in Computer Science. Can they be hanged on the Web on a scalable basis? Would it make sense to have dbs describing Web Services or resources in general? Why do we all have to stick to SQL, is this the summum? I am afraid not.2- It is quite true that inference is a Semantic Web cornerstone and there is no serious inference in RDF. But is also true that RDF provides faceted search and a number of interesting possibilities to pool out data from heterogeneous sources. Remember the AI Winter? A number of research projects promised something they were never able to provide. Sometimes is better to go for the low-hanging fruit and then scale it up.
3- RDF is somehow a database. Better, it is a simple fully-fledged model (subject-predicate-object) to represent ANY relational database (since forty years based on the ERD model) so we face complementary perspectives, not opposed.
The question is more: why is people reluctant to cope with new things? Perhaps because they think it is just the same king with a different crown. However, incremental innovation is precisely made of this finite jumps and in research, clearly, we stand on the shoulder of giants.
Post a Comment