Linked Data, SemWeb, Meaning Maps etc.

Well Bex seems to be coming around to the SemWeb ideas. That's good. Still some way to go but we will arrive at our own pace.

As an aside, SemWeb = Linked Data Web.

"The challenge of the Semantic Web, therefore, is to provide a language that expresses both data and rules for reasoning about the data and that allows rules from any existing knowledge-representation system to be exported onto the Web." from TBL & Friends' original seminal article

What is interesting to me as an observer, fan and modest practitioner of SemWeb technology (blending it with ECM systems primarily)merge-700542.jpg is reading and watching the evolution of ideas from the SemWeb community. It's more idea convergence. This is great. It means that there are a host of smart people all working towards a common vision. For instance, last year I suggested,

"A BI engine can create a concept map based on term frequencies, term proximities and term usage. When combined with metadata classifications and transactional data, an organic classification structure begins to surface. Mapped against topics, groups, or classification phyla, specific semantic ontologies begin to emerge. These organic ontologies can then be used as the basis for BI inferences that produce predictions for users. "
Then today I read a great and complementary article by fellow Minnesotan Dan McCreary about how to leverage text extraction and transform capabilities to enrich and tag unstructured information (with RDFa for example) in order to unlock the data and concepts from the content item. Then there is this article on the specifics of Semantic BI from just 2 days ago. Author Brooke Aker concludes,
"Including unstructured data provides a more complete corporate picture. Semantic intelligence extracts knowledge from these sources and when combined with BI, businesses receive knowledge from structured and unstructured modalities, which will eventually be a requirement as we move into the next phase of the Web."

Then I spot another confluence between my folksonomy paper (PDF warning) from 2007 and this great article from Stephanie Lemieux on tagging at work. She has a great list of "Do's & Don'ts" that enterprises should take to heart.

Ideas ebb and flow. We are in a period of very rapid thought and exploration and innovation when it comes to social web technologies and bringing meaning to the data of the web. Despite the economic hardships, agile businesses will take these ideas to heart and maybe weather the storm ahead of their competition.


don't be so sure... I have planned a 3-part takedown on how LinkedData will likely fail. Stay tuned...

Posted by bex on March 29, 2009 at 04:42 AM CDT #

I can't wait! The clash is good for spotting and fixing weakness

Posted by billy on March 29, 2009 at 05:18 AM CDT #

the basic problems I see are: 1) URLs are pretty awful unique IDs 2) there's not much incentive for me to turn my unstructured info into structured info, AND maintain it 3) we've tried similar things in the past that all failed, and I'm not seeing much difference here. 4) I've never seen a semantic web search engine that produced useful information for a broad range of queries... even for subjects they claimed to spider. 5) HTTP is the worst possible protocol for maintaining a system of linked data.

Posted by bex on March 29, 2009 at 03:35 PM CDT #

Hmmm: 1) URI's are good unique IDs. Why are they bad? Wide adoption and infrastructure can't be ignored. 2) I agree there is not much incentive for you to microformat your info and maintain it. That is why a precondition of SemWeb success is organic or automatic enriching. Folksonomy and crowdsourcing are the first baby-steps into this arena. They're not playing the same game yet, but the stadium is the same. 3) Meh. Edison tried 10,000 times to invent a lightbulb, Legislators try to pass the same bills over and over again. Eventually they get the mix "right"... 4) Agreed. Things are starting small and narrow in well defined (vertical) concept spaces. Check out this story about the combination of TCM and MWM (traditional chinese medicine and modern western medicine) concept maps being merged for a hospital and EMR application: Or this one touting the a capability strikingly similar to Oracle UCM's content categorization capability but applying it to Bio-Med. This is where things will start and then grow methinks. 5) Calling HTTP the "worst possible protocol for maintaining linked data" is a very aggressive claim. You have to overcome a mountain of empirical evidence and adoption and apps to the contrary. If there is even one protocol that is worse, your claim is false. I won't argue that it is the best, but I will submit that IPoAC (Internet Protocol over Avian Carriers) is worse than HTTP for maintaining linked data.

Posted by billy on March 30, 2009 at 04:20 AM CDT #

Post a Comment:
Comments are closed for this entry.

Enterprise 2.0 and Content Management


« April 2014