Semantic Web may get Pass Marks

One of my friend sunny forwarded me this link Why the Semantic Web Will Fail, which enlisted perhaps a lot of frustrations researchers have faced and much more artifacts. But I had an immediate reaction.

IMHO, certain ways semantic web may work if we have a look at following thoughts :-

1) When we talk of semantic web, the first picture comes into mind is to "integrate ALL information across web", right from US to India, from Russia to Japan, Europe, Africa everywhere. The image is very similar in case of enterprises also, as we assume to do things what author has said like expose APIs, agree over standards etc etc. The vastness and complexity of the WHOLE system makes people doubt about its success, which I think is wrong perspective.

2) The right picture of semantic web, actually lies in small small parts of world. Many semantic web projects are on (there are companies whose sole business is this!). Even from nature of information perspective, disparity on meanings (atleast on words, may not be on definitions) does not vary smoothly across one end of world to another. Disparities are minimal in localized areas, and they disruptively increase as certain boundaries are crossed. Hence, this nature of information disparity directly suggests multi-tier integration model in place of one-tier model which point (1) suggests. This means, information has to be grouped locally (which is based on nature of information, somewhat coupled with geography of its storage) in 1st stage and then integrated globally in 2nd stage to provide larger picture (there can be more than 2 stages).

3) Semantic Web can not work like web services, which succeeded on standards. This is mainly because information across web can not be controlled over schema and all, its too small for that. On the other hand, things like metadata in current form does not have richness to allow semantic searches. Methods like semantic thumbnails, controlled folksonomy etc can definitely augment semantic searches (atleast based on words, may not be on definitions) across web. The alternative of using RDF may not be viable as its expensive.

We have been quite successful in machine-to-machine integration (i.e. web services) and also people-to-people integration (i.e. web2.0). Both are loosely correlated in terms of domain, technologies, standards, market etc. It is people-to-machine or machine-to-people integration that merges two extremes of web interactions today, which is tough. Semantic Web plays exactly a role there.

2 comments:

  1. Ab to tum itne aage badh gaye ho ki baatein bhi samajh mein nahin aati hai... anyway keep going dude:)

    ReplyDelete
  2. hey man.. nth like that.. yeah this blog actually came out after a conversatoin we had with folks.. so stuffs r little short.. im not a semantic guy though :P

    ReplyDelete