Comments on: Vertical Search - Why? http://www.dronamraju.com/blog/2005/03/vertical-search-why.html Opinions on Business, Technology & Life from Ravi Dronamraju Fri, 21 Nov 2008 02:08:44 +0000 http://wordpress.org/?v=2.6.2 By: Runescape Money Blog http://www.dronamraju.com/blog/2005/03/vertical-search-why.html#comment-108375 Runescape Money Blog Sun, 02 Sep 2007 16:02:55 +0000 http://www.dronamraju.com/blog/?p=37#comment-108375 <strong>Runescape Money Blog</strong> I have written a followup to this on my blog. Feel free to check it out if interested. -Louis--- This post was created by Comment Poster. Want to post thousands of comments to blogs instantly and increase your visitors exponentially? Check it out he... Runescape Money Blog

I have written a followup to this on my blog. Feel free to check it out if interested. -Louis— This post was created by Comment Poster. Want to post thousands of comments to blogs instantly and increase your visitors exponentially? Check it out he…

]]> By: Ravi Dronamraju http://www.dronamraju.com/blog/2005/03/vertical-search-why.html#comment-71 Ravi Dronamraju Fri, 11 Nov 2005 03:14:02 +0000 http://www.dronamraju.com/blog/?p=37#comment-71 kruttik, google, yahoo and other major search engines use context. However, the current context gleaning algorithms are very susceptible to spam. Thus, the refinement of pagerank. PageRank was innovative because it made it very difficult to spam. However, these days SEOs figured out how to spam PageRank. There are variations called TrustRank etc, that are used to associate authenticness to document content. So, Every search engine starts with context. Unless you have a major break through in distinguishing spam from relevant content, just based on document info, I'd not be optimistic on relying document content alone. kruttik,

google, yahoo and other major search engines use context. However, the current context gleaning algorithms are very susceptible to spam. Thus, the refinement of pagerank. PageRank was innovative because it made it very difficult to spam. However, these days SEOs figured out how to spam PageRank. There are variations called TrustRank etc, that are used to associate authenticness to document content.

So, Every search engine starts with context. Unless you have a major break through in distinguishing spam from relevant content, just based on document info, I’d not be optimistic on relying document content alone.

]]> By: Kruttik Aggrawal http://www.dronamraju.com/blog/2005/03/vertical-search-why.html#comment-69 Kruttik Aggrawal Thu, 10 Nov 2005 12:12:14 +0000 http://www.dronamraju.com/blog/?p=37#comment-69 Hi Ravi, I dont know if I am qualified enough to speak to ur mails but being a true passionate Computer science student, its my duty to let the knowledge flow properly and especially in the feild of search engines since i have given 4 years of my life to it. I am surprised to see this euphoria over the invent of vertical search engines and the kind of facilities it provide to specific searches but one thing little thing that is missed by all of u is the bottle neck it brings to the inquisitiveness that researchers have in the feild of AI implementation. An ideal search engine is neither a vertical-centric search engine(Indeed.com , simplyhired.com, Workzoo, NimbleCat, Fatlens) nor a pattern-centric search engine(google,yahoo,altavista etc) but a context-centric search engine. Let talk technical now. What google does by its page-ranking is that it evaluates the confidence value that the site holds for the data it contains and displays it on the screen, which sometimes gives very relevent answers, but the success is not credited to the fact that the site had relevent data but to the fact that its linked more than others. CRAZY!!. Image yourself developing a website that has the most relevent data for a topic ever written but it would not be visible to the ranks until it been linked to enough sites. The context is lost. Basically the flaw in all the search engines including the vertical ones is that its pattern centric. Relevent is drawn by every logic other than the most basic one, CONTEXT. I have been working on context-search engines and hope would give u better insight asap. Thanks anyways, Hi Ravi,
I dont know if I am qualified enough to speak to ur mails but being a true passionate Computer science student, its my duty to let the knowledge flow properly and especially in the feild of search engines since i have given 4 years of my life to it.
I am surprised to see this euphoria over the invent of vertical search engines and the kind of facilities it provide to specific searches but one thing little thing that is missed by all of u is the bottle neck it brings to the inquisitiveness that researchers have in the feild of AI implementation.
An ideal search engine is neither a vertical-centric search engine(Indeed.com , simplyhired.com, Workzoo, NimbleCat, Fatlens) nor a pattern-centric search engine(google,yahoo,altavista etc) but a context-centric search engine.
Let talk technical now. What google does by its page-ranking is that it evaluates the confidence value that the site holds for the data it contains and displays it on the screen, which sometimes gives very relevent answers, but the success is not credited to the fact that the site had relevent data but to the fact that its linked more than others. CRAZY!!. Image yourself developing a website that has the most relevent data for a topic ever written but it would not be visible to the ranks until it been linked to enough sites. The context is lost.
Basically the flaw in all the search engines including the vertical ones is that its pattern centric. Relevent is drawn by every logic other than the most basic one, CONTEXT.
I have been working on context-search engines and hope would give u better insight asap.
Thanks anyways,

]]> By: Adam Gedde http://www.dronamraju.com/blog/2005/03/vertical-search-why.html#comment-37 Adam Gedde Thu, 26 May 2005 04:52:00 +0000 http://www.dronamraju.com/blog/?p=37#comment-37 Ravi:<BR/><BR/>I agree with Dave's comment - there is a lot of content out there that isn't worthy of vertical search. Heck - it isn't even worthy of inclusion in the search engines (<A HREF="http://www.seroundtable.com/archives/001963.html" REL="nofollow">Google's sandbox effect is trying to weed the spammers out</A>)<BR/><BR/>I also agree with what you said - it's going to be nearly impossible for people to get a complete view of the inventory of content online if they were only to turn to the big search engines. This is the main focus of the Vertical Search market - providing a complete view.<BR/><BR/>The Vertical Job Search Engines out there are beginning to understanding how to gather, organize, and disseminate content from all their different providers. The end goal - to give job seekers a one stop shop for job searching.<BR/><BR/>But, I would argue that the crawling model which Google and other traditional search engines <B>is</B> applicable to the Vertical Job Search Market.<BR/><BR/>Our company, Fetchster.com, is a <A HREF="http://www.fetchster.com/" REL="nofollow">Minnesota Job Search Engine</A> which uses crawling technology to spider the employment portals of Minnesota companies and provides those results in a web-searchable database.<BR/><BR/>In fact, you'll find many of the same Minnesota listings on Fetchster as you do on SimplyHired and Indeed.<BR/><BR/>Despite showing that crawling technology works in this space, we still have a long way to go. We think our method provides the most accurate, interesting, and unique job results. But that's not to say that we're right, or that any other player in the industry is right or wrong. The VJSE market is very young, but one thing's for sure - there are a lot of great innovators out there, and that's going to benefit the job seeker in the long run.<BR/><BR/>Adam Gedde<BR/><A HREF="http://www.fetchster.com/" REL="nofollow">Fetchster.com</A> Ravi:

I agree with Dave’s comment - there is a lot of content out there that isn’t worthy of vertical search. Heck - it isn’t even worthy of inclusion in the search engines (Google’s sandbox effect is trying to weed the spammers out)

I also agree with what you said - it’s going to be nearly impossible for people to get a complete view of the inventory of content online if they were only to turn to the big search engines. This is the main focus of the Vertical Search market - providing a complete view.

The Vertical Job Search Engines out there are beginning to understanding how to gather, organize, and disseminate content from all their different providers. The end goal - to give job seekers a one stop shop for job searching.

But, I would argue that the crawling model which Google and other traditional search engines is applicable to the Vertical Job Search Market.

Our company, Fetchster.com, is a Minnesota Job Search Engine which uses crawling technology to spider the employment portals of Minnesota companies and provides those results in a web-searchable database.

In fact, you’ll find many of the same Minnesota listings on Fetchster as you do on SimplyHired and Indeed.

Despite showing that crawling technology works in this space, we still have a long way to go. We think our method provides the most accurate, interesting, and unique job results. But that’s not to say that we’re right, or that any other player in the industry is right or wrong. The VJSE market is very young, but one thing’s for sure - there are a lot of great innovators out there, and that’s going to benefit the job seeker in the long run.

Adam Gedde
Fetchster.com

]]> By: Dave http://www.dronamraju.com/blog/2005/03/vertical-search-why.html#comment-36 Dave Sun, 27 Mar 2005 01:35:00 +0000 http://www.dronamraju.com/blog/?p=37#comment-36 Nice post Ravi. Your comment about 'supportive applications' is right on target, and your reasoning is very sound for why some content/data (tho not all) is better served by vertical search.<BR/><BR/>I work for <A HREF="http://www.simplyhired.com" REL="nofollow">SimplyHired.com</A>, one of the companies you mentioned, and you've picked up on a lot of the issues we've been thinking about.<BR/><BR/>Indeed, Workzoo, & Nimblecat have all been doing pretty well on job search to date. We've probably got a little bit of catching up with them to do on a few features (though not on the breadth of data), but we hope to offer some unique stuff of our own in the next month or two.<BR/><BR/>One thing we do think is important and perhaps a little different than those guys is that we think the UI should be pretty simple, and advanced features should be layered on carefully. Also, we enjoy having a little fun with our site here & there, and we hope the <A HREF="http://www.simplyhired.com/aboutus.html" REL="nofollow">voice</A> we've chosen is an enjoyable one to read.<BR/><BR/>anyway, i agree with you that supportive apps are important... <BR/><BR/>(btw, interested in a job? ;)<BR/><BR/>- dave mcclure Nice post Ravi. Your comment about ’supportive applications’ is right on target, and your reasoning is very sound for why some content/data (tho not all) is better served by vertical search.

I work for SimplyHired.com, one of the companies you mentioned, and you’ve picked up on a lot of the issues we’ve been thinking about.

Indeed, Workzoo, & Nimblecat have all been doing pretty well on job search to date. We’ve probably got a little bit of catching up with them to do on a few features (though not on the breadth of data), but we hope to offer some unique stuff of our own in the next month or two.

One thing we do think is important and perhaps a little different than those guys is that we think the UI should be pretty simple, and advanced features should be layered on carefully. Also, we enjoy having a little fun with our site here & there, and we hope the voice we’ve chosen is an enjoyable one to read.

anyway, i agree with you that supportive apps are important…

(btw, interested in a job? ;)

- dave mcclure

]]>