Raw

I had a couple of fairly random morning thoughts while reading this morning's Planet RDF, they turned into another long procrastinatory ramble.



First, Taowei on the Mindswap blog makes some good points in regard to SemWeb UIs, in particular about the similarity of the underlying data to what we already have in current DBs.

So, the real question is, what are the fundamental differences between Sem Web technologies in representing data and traditional relational databases that requires a new paradigm to analyze data.

I think the answer can be summed up as : The Web. The implications are that the amount of data potentially available is like nothing before, the quality will cover the whole range from noise to knowledge, the domains will cover the whole gamut of human knowledge. Applications need new skills like selectivity, provenance management, increased robustness and above how to deal with linkage.

Taowei is cautious about UI side, about what you get for free in terms of usefulness of UIs and their extension through ontologies/schemas. I don't really see this as an issue. Current applications can have sophisticated UIs for domain-specific material, there's no reason these shouldn't be used as views of a more diverse data set. The RDF/OWL model offers big potential for domain-independent browsing and data manipulation, and already tools like Tabulator and Longwell are demonstrating the utility of this approach. So although we won't get all the accessories for free, batteries are included.

Henry Story covers Nova Spivack's Meaning and future of the Semantic Web, noting that a journey of a thousand miles begins beneath one's feet. There's one bit of Nova's piece I want to pick up on:

Within a year or two you will see mass-consumer Semantic Web products and services hit the market, and within 5 years there will be at least a few “killer apps” of the Semantic Web. Ten years from now the Semantic Web will have spread into many of the most popular sites and applications on the Web. Within 20 years all content and applications on the Internet will be integrated with the Semantic Web.

I think the timescales here are realistic, and I think my own level of optimism is about the same as Nova's. But I'd frame the developments rather differently. I certainly believe there's enough momentum (or hype, if you prefer) that there will be mass-consumer apps using RDF in their core very soon. But I don't think this is a particularly significant indicator. Similarly, there's a good chance of a few killer apps coming along - Semantic Web tech does open a few interesting doors. But I think it may be slightly misleading to talk of complete integration of current content and applications.

I'm pretty sure it wasn't Nova's intention, but that paragraph does give the impression that the difference between the current Web and the Semantic Web can be viewed in black and white. I think it's far more like a continuum (a point I was trying to make in The Shortest Path).

Sure, there are the specific Semantic Web technologies (RDF, OWL, SPARQL etc). The core tech is now usably stable, and there is a good range of tools for different platforms (at a programming level at least, Nova mentions the paucity of non-expert development tools). Yet although anyone looking at the current state of affairs might notice a significant increase in the interest and use of the tech over the past couple of years, there's apparently no meteoric rise on the web at large. I've heard Dan Connolly wonder out loud regarding what constraints around the tech need to be loosened to make the stuff viral. I think he doth worry overmuch.

Ok, looking ahead say 10 years, I don't think we'll see the use of RDF and associated tech within a large proportion of applications on the Web. But I also don't believe this is cause for pessimism, far from it. Massive deployment of SemWeb tech isn't needed for there to be a huge growth in the utility of the web thanks to that tech. Even if only say 0.0001% of applications actually use Semantic Web technologies, we will still have a Semantic Web.

Nova has it the right way around when he talks of things being integrated into the Semantic Web. Systems both produce and consume data. It isn't necessary for all applications to e.g. produce RDF for it to be possible for RDF-based applications to consume their data. In the same way that in global terms it doesn't matter where data is located, it doesn't matter where integration takes place. We already have a vast amount of data on the web, it's just opaque to most software. In the short term I think we'll see this data become increasingly visible, it's already happening with things like RSS/Atom, Web 2.0 APIs, microformats... Semantic Web tech offers a way of getting extra value from this data through integration (and filtering, and reproduction).

I forget where, but I believe Tim Berners-Lee has talked about making little Semantic Web islands on the web, and their joining up into archipelagos then larger landmasses. But (to mangle the analogy), land is quite an expensive medium for connecting land. The Earth is mostly ocean, yet we can still travel between continents and islands. What's more (to get all meta/physical with the analogy), information can walk on water.

At this point in time, in terms of deployment, growth of semweb systems probably looks linear. But Semantic Web technologies are designed to benefit from the network effects that affect the web. They are connected to the web. So it seems to me likely that most of the growth in the near future will be built directly on the current, mostly passive, document-oriented web. The tools are already here for putting data (in the traditional RDBMS sense) on the Semantic Web. The existing individual Web 2.0 service APIs can be addressed through a consistent model, so the wiring is open to greater interconnection. There's plenty of work going on at the user interface end, right down to the Semantic Desktop. But this stuff is autocatalyctic, and I don't think it will be long before we're accelerating to the (yeah, pulled from the air) 0.0001% point, and beyond that it's anybody's guess. As Henry sagely put it, enjoy the ride.

@en