One thing that I thought might be worth sharing separately is the "Linked Data Value Spiral" below. It tries to illustrate that semantic data don't have a single-loop life cycle, but that re-distributing utilized ( = newly meshed/combined) information will create a self-enforcing "Linked Data ecosystem". I tried to associate the individual value creation processes with SemWeb market sectors. (RDF stores, for example, are typical information organization products, paggr tries to remove the bottleneck between utilization and re-distribution, etc.)
It's just an abstraction, the boundaries are of course blurry (a SPARQL endpoint can help with both utilization and discovery), but I still find the simple spiral and its segments handy to classify current products and companies. It even helped me a little to identify market opportunities and gaps:
- The recent VoiD effort could have a significant impact on the whole Semantic Web progress.
- Entity extraction providers like Zemanta and OpenCalais could play a huge role to boost creation processes.
- What about "accelerator" products that offer a shortcut between utilization and creation (i.e. apps that create Linked Data while you are using them, with instant re-distribution)?
- Is it a problem when a service like Freebase exports RDF but doesn't provide links to external datasets?
- ...
Feel free to use and share.
Comments and Trackbacks
Their default web editor is powerful, but it's only a beginner-level tool for data entry. For heavy data entry and export, it's best to build custom tools.
It wouldn't be hard to build a custom RDF exporter that translated some concepts to other namespaces. For some trivial things, it could done with regexes (20 lines of code maybe?).
I'm still looking for useful apps to consume the data. So far, it looks like Yahoo will crawl and store microformats and RDFa/eRDF - so I'm going to try exporting to that first.
To answer my own question: I don't think non-linked export is a problem. At least not for the consuming parties (and not everyone has the required resources). It can still be a great contribution. The LOD folks created DBPedia mappings on their own and thus added value to the overall data cloud. Freebase could now use the enriched data to increase the value of their own dataset, too.
I think it's a good general model for talking about the self-reinforcing value proposition of Linked Data - though it does make me reminisce for Christmas a bit.
I think your point about Accelerator products is interesting. In general I believe that the majority of Linked data will - over time - essentially be the "exhaust" from other processes. As a simple example - when we're browsing and sharing links with the bit.ly URL shortener - in the background Calais is processing each linked page and populating the Calais Linked Data Cloud. For 95% of users this is immaterial - it's just exhaust from another process. For the other 5% it will provide an invaluable resource where the LD assets can be directly harvested. I'd challenge people to think about what useful exhaust they could expose without the risk of exposing proprietary data. Twine could tell us what's being shared, Freebase could tell us what's being viewed, Tripit could tell us about destinations.
Regards,
Heh, and you just triggered a mental link from the spiral to x-mas here, too. Hard to tame, this linkage stuff ;)