Fog Creek Software
Discussion Board




What will win and why.

Ok,  inspired by Crimsons "How Do You Keep Up" thread.

We are the experts, the hype machines driven by vested interests and Journalists are not. 
Lets hear your opinions on where we are headed.

Pick a new technology and give your view.

RDF

I was _Very_ excited by RDF.  It makes instinctive sense to me that more structured data means more sophisticated searching and manipulated. 

If the web was RDF not HTML i should be able to seach not only for the terms "chess tutor washington" but say "get me all the chess teachers in the washington area who charge under $50 an hour and are rated over 2200". 

Obviously currently it would be very hard to create a web search just to do that particular search (given all the different ways the required info is presented) never mind a generic mechanism to make searches "of that form".

Interest Rating : **** (4 of 5 stars)
Future : Been disapointing so far.  Will probably be superseeded before it has any mainstream effect.

There is no doubt the next big leap forward will be AI. 
Or to put it another way 'much more sophisticated mechanisms to manipulate information'. 


Braid_ged.

braid_ged
Wednesday, July 02, 2003

RDF will not win. Why? Because it is a HUGE effort to implement it.

In order to be useful, a lot of people and companies have to implement it - and most of them won't bother.

I am a lot more excited about things like ASP .NET - a powerful, RAD environment for the web.

This could make complex web sites a lot more common.

Avenger
Thursday, July 03, 2003


As a side question... why are "complex web sites"... or complex anything deemed such a virtue?

Joe AA
Thursday, July 03, 2003

Well, why is a well-cooked food considered such a virtue?

Why not read the raw fish and potatoes?

JJ
Thursday, July 03, 2003

Braid,

the kind of searches you talk about relate to the "Semantic Web" envisoned bt Tim Berner's Lee. Read his paper on the Scientific American regd. this.

Prakash S
Thursday, July 03, 2003

As others have said far better, there is no Semantic Web, its just a Syntactic Web. You do not get anything for free, you just get a markup syntax that you have to apply to make it easy on the SM people.
"Hear ye, hear ye: from now on you will add markup to your webpages with links to the Grand Unified Onthology. You millions of people will do this massive work so that We, the gods of SM, can implement a simple graph traversal algoritm and claim to have solved AI. Now if you will excuse us, I hear the next truckload of government research grants dumping money pulling into our courtyard."

Know this: People do not do markup.
Let's try that again: People  do  not  do  markup!
It has been tried in news agencies: "Catagorize my story? Keywords? I am a journalist, Sir. I >make< the news. Want someone to do household chores, call my cleaners!"
On the Web, you say? People can't even do proper HTML, and only search engine spammers pay attention to meta-tags. "Are you saying www.bighooters.sex is not rdf:about rdf:collection evry damn concept in the universe?".
Now, when was the last time you filled out the catagory and keywords fields in a Word document. Yes, it is rigth there under "Summary" in the document properties. You knew that, rigth? For those of you with in the nice big companies with the KM-enableled Intranet, did you bother to select the right catagory and keywords even when the system forced you to, or did you just hit OK?

I rest my case.

Just me (Sir to you)
Thursday, July 03, 2003

Prakash,

My buddy, (read a lot of your posts, love your work),

What does Tim suggest we build the semantic web with ?

You guessed it, RDF.

Braid_ged

Braid_ged
Thursday, July 03, 2003

Just me (sir)--
You are so **on the money** with that comment. I'm involved in a major KM snake-oil project at megacorp inc, and guess what? The whole fabric is tied together by user-supplied metadata, which is a fairly preposterous and unworkable approach.

Dr. Van Nostrand
Thursday, July 03, 2003

"Know this: People do not do markup."

Marked up text is the single most valuable commodity in natural language processing research.  With well marked up corpi, a researcher can make break throughs and work wonders.  Without it, you can't get started.

Jim Rankin
Thursday, July 03, 2003

<pedant>Corpora. Third declension. :-)</pedant>

Gareth McCaughan
Thursday, July 03, 2003

Well not that I have thought that much about the "semantic web", but it seems to me a solution to the markup problem is to make it worthwhile for people to markup.

i.e. if people got more hits to their websites because of markup, then they would markup.  People go to great lengths to try to improve their search engine ranking (there are whole websites dedicated to this, apparently).  If there were some search engine that was better than Google because it could analyze markup, then a lot of people would use it.  And if more complete markup resulted in more hits, then people would do it in a second.

But of course I can see lots of problems with people creating misleading markup just to increase their rating.  Oh well.

Andy
Friday, July 04, 2003

Realize that the major reason why Google was better than the others was that they >stopped< relying on markup.
I reffered to the meta-tag spamming problem before. Maybe I should put it more clear: "People do not do honest markup, but they will try to abuse whatever you give them for selfish gains".

Just me (Sir to you)
Friday, July 04, 2003

Maybe I am not knowledgeable enough, but what is the point of the semantic web unless every (or many) web sites implement it?

It seems to me that the value would be greatly lessened, if it's just a few who do it... and therefore I predict there is a massive obstacle to its adoption.

S. Tanna
Friday, July 04, 2003

*  Recent Topics

*  Fog Creek Home