Monrai Blog

News about Cypher, Semantic Web, Natural Language Processing, and Computational Linguistics

Monday, July 28, 2008

Case for NLP in Semantic Web

Here is a great paper on the case for the use of NLP in Semantic Web applications.

CNL_Reportv7.pdf (application/pdf Object)

Tuesday, July 08, 2008

Response to Dan Grigorovici's articles (Part 2)

I finally got around to looking at MicroFo...s today (which I will henceforth refer to as MFs, if you know what they are, keep reading, if you don't know what they are, good!), and was amazed at how redundant it is. It's primarily RDFa watered down. But then something else struck me... it is actually being USED, by a growing number of software developers, web publishers, and end users. It seemed to catch fire rather quickly (relative to the growth and development of the W3C endorsed Semantic Web).

As I continue offer my response to Dan Grigorovici's blogs on the business and marketing aspect of the Semantic Web (or more accurately, the lack thereof from his prospective), I would like to turn the attention to the growing attraction to MFs over RDFa as offering one of the sources of the problem. According to the Wikipedia page, MFs is a grassroots effort, which gained popularity and then support from a corporate sponsor. It's not a standard, and is steered by the very loose-knit community, essentially, a mf begins it's life as a wiki entry. Question: What drives an end-user's decision to "roll his own" verses adopting an off-the-shelf solution? Answer: Familiarity. Neurologist know that the task of creating a new synaptic connection in the brain requires many times more resources and energy than reusing an existing neural pathway. So reuse, leveraging existing knowledge/practices is an aspect that is always sought after by the brain. It's often times infinitely easier for a developer who is proficient in HTML/CSS, to build a site from his own personal libraries of (familiar) templates, than to be given an existing site to remodel. And with MFs, and the Web community at large, it appears that there is less friction involved in starting from HTML --> HTML+semantics, and easily seeing the relation, than to one day have a RDF Primer and SPARQL specification dumped in your lap and told to go from QuadStore --> triples --> HTML+semantics. It's obvious that the Semantic Web vision is "right", because the world is obviously demanding it, but the world never wants to DO right. We must ensure that we arrive at the right destination, and get there the right way. If I set out to the store, and drop all my money on the way there, then what good is it if I make it to my destination?

I think that the momentum building around MFs indicates that the W3C's Semantic Web problem is largely one of PR and public image. If you are a good spin-doctor, I recommend the Semantic Web as a qualified client. I would go so far as to agree with Dan's arguments and Hank William's suggestion, that if you are a semantic web startup, drop the association with Semantic Web in any public or investor-facing collateral for your product or company. I really wish someone would have taken the RDFa spec, dropped the RDFa label, slapped the name MFs on it, stood up a wiki and discussion group, and proposed it from the grassroots level.

Keep in mind that by rebranding the Semantic Web, we mean only a cosmetic overhaul while leaving untouched the core the tenants we have, as a community, chosen as necessary to uphold: use dereferencable URIs, publish, use and reuse RDF vocabularies, reuse other's URIs for naming things when possible, among others. It's important that we hold true to these principles, and that we are not blown about by every wind of doctrine that surfaces regarding how the Semantic Web should manifest, or compromise the vision we've laid out. I, for one, have decided not to support MFs in any of my software, nor publish it on any of my sites. As a community, we have our view, and the rest of the world seems to have their view, let us stick to our view, and not give way to these factions. Many of us are involved in work to RDFize the existing world, as Kingsley Idehen puts it, to "build the house around the users", as opposed to dragging the user into the house. This is strategic and a form of compromise, but it can only go so far. The Semantic Web will need solidarity if it will be preserved, multiple standards split resources, focus and impede interoperability and reuse, and these splits will only weaken our foundations. Only by having one unified set of standards, emanating from one unified community, can we hope to build anything substantial on top of the Semantic Web.

Why does everything suck?: Killing Ontologies/OWL In The Semantic Web?