<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	>
<channel>
	<title>Comments on: Bringing us the best of 1995 to you here today!</title>
	<atom:link href="http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/feed/" rel="self" type="application/rss+xml" />
	<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/</link>
	<description>Erik Selberg's Homepage &#038; Blog</description>
	<pubDate>Mon, 13 Oct 2008 05:00:15 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.5.1</generator>
		<item>
		<title>By: Erik Selberg</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-195</link>
		<dc:creator>Erik Selberg</dc:creator>
		<pubDate>Wed, 05 Oct 2005 07:51:40 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-195</guid>
		<description>So... yes and no as far as the UI goes. You can certainly instrument a rich (or AJAX, or whatever) client that reports every little movement back to the system and have the system attempt to learn from that behavior. Nobody has done that yet without making something totally unweidly, but perhaps some of the new AJAX stuff would enable that. And you know with all the coolness that we're seeing, somebody, if not everybody, will try dumping a ton of fun AJAX code onto search results to make them more interactive. Sort by Date! Collapse Snippets! Export to XML! Mark some as relevant, mark other results as not relevant, and requery!

The bigger issue is whether that actually help people.

These features invariably help people with answering hard or unanswerable queries... such as researching into whether or not Mier would make a good associate on SCOTUS. But it doesn't help for simple queries, or even slightly complex queries that are very answerable, such as "unlock audiovox 5600." So, I'm left wondering whether we do in fact want one interface that does everything, or two (or more)... one for simple, direct queries, and another that's more of the researching / answer hard questions route.</description>
		<content:encoded><![CDATA[<p>So&#8230; yes and no as far as the UI goes. You can certainly instrument a rich (or AJAX, or whatever) client that reports every little movement back to the system and have the system attempt to learn from that behavior. Nobody has done that yet without making something totally unweidly, but perhaps some of the new AJAX stuff would enable that. And you know with all the coolness that we&#8217;re seeing, somebody, if not everybody, will try dumping a ton of fun AJAX code onto search results to make them more interactive. Sort by Date! Collapse Snippets! Export to XML! Mark some as relevant, mark other results as not relevant, and requery!</p>
<p>The bigger issue is whether that actually help people.</p>
<p>These features invariably help people with answering hard or unanswerable queries&#8230; such as researching into whether or not Mier would make a good associate on SCOTUS. But it doesn&#8217;t help for simple queries, or even slightly complex queries that are very answerable, such as &#8220;unlock audiovox 5600.&#8221; So, I&#8217;m left wondering whether we do in fact want one interface that does everything, or two (or more)&#8230; one for simple, direct queries, and another that&#8217;s more of the researching / answer hard questions route.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Cybermagellan</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-191</link>
		<dc:creator>Cybermagellan</dc:creator>
		<pubDate>Wed, 05 Oct 2005 00:59:12 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-191</guid>
		<description>Is it ever really possible to ‘watch’ peoples behaviors when they search, and make meaningful ranking modifier changes, or do what I do in .Search - pin web pages/web sites?

yes it is...and the higher a certain link gets clicked the higher up the rung it should get till it is #1 or whatever. Eventually you'll have multiple #1s however the point being is that these will eventually filter down into taste of content. If you think about it right now it is possible however for a particlar way. You have to actually have metadata that says Q: FooQ A: FooA and then have someone search for FooQ. However what would happen if you were to attach just on the sly a What+FooQ, Who+FooQ, When+FooQ, etc. 

I would be interested in perhaps researching how this would work....perhaps making something to track items and see how they are ranked...Erik, I pimped out your blog post so sorry if you get a ton of hits :-D</description>
		<content:encoded><![CDATA[<p>Is it ever really possible to ‘watch’ peoples behaviors when they search, and make meaningful ranking modifier changes, or do what I do in .Search - pin web pages/web sites?</p>
<p>yes it is&#8230;and the higher a certain link gets clicked the higher up the rung it should get till it is #1 or whatever. Eventually you&#8217;ll have multiple #1s however the point being is that these will eventually filter down into taste of content. If you think about it right now it is possible however for a particlar way. You have to actually have metadata that says Q: FooQ A: FooA and then have someone search for FooQ. However what would happen if you were to attach just on the sly a What+FooQ, Who+FooQ, When+FooQ, etc. </p>
<p>I would be interested in perhaps researching how this would work&#8230;.perhaps making something to track items and see how they are ranked&#8230;Erik, I pimped out your blog post so sorry if you get a ton of hits <img src='http://selberg.org/wp-includes/images/smilies/icon_biggrin.gif' alt=':-D' class='wp-smiley' /></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Matthew</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-190</link>
		<dc:creator>Matthew</dc:creator>
		<pubDate>Wed, 05 Oct 2005 00:08:34 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-190</guid>
		<description>Erik, I think the key is to analyze how the user manipulates, uses and view search results.  Problem is, right now we don't have much to analyze because the UI is limited - there's not much the user can do with Google, MSN or Yahoo results besides scrolling through them and clicking on them.  Not a very efficient workflow to begin with, but also not a particularly good environment for "handling" data.  I'm working on a methodology which incorporates UI, and I think that's going to be the next big thing.</description>
		<content:encoded><![CDATA[<p>Erik, I think the key is to analyze how the user manipulates, uses and view search results.  Problem is, right now we don&#8217;t have much to analyze because the UI is limited - there&#8217;s not much the user can do with Google, MSN or Yahoo results besides scrolling through them and clicking on them.  Not a very efficient workflow to begin with, but also not a particularly good environment for &#8220;handling&#8221; data.  I&#8217;m working on a methodology which incorporates UI, and I think that&#8217;s going to be the next big thing.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Paul From Nata1</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-189</link>
		<dc:creator>Paul From Nata1</dc:creator>
		<pubDate>Tue, 04 Oct 2005 23:56:06 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-189</guid>
		<description>I agree.

On a limited corpus (i.e. for a search appliance), then human rating, pinning, adding extra meta data is extremely extremely useful (why I loved building .Search on top of Community Server)

For an unlimited corpus - no way!  That's what we're talking about right?  Users altering ranking modifiers as opposed to bots?  I agree 100% then that humans need to be creating/updating rules in the rule engine, but definitely not altering ranking modifiers a site at a time!!!

Is it ever really possible to 'watch' peoples behaviors when they search, and make meaningful ranking modifier changes, or do what I do in .Search - pin web pages/web sites?  

I love Scoble's search posts - its good to see them get attention.   I'll have to read this exchange over and over to truly 'get it'</description>
		<content:encoded><![CDATA[<p>I agree.</p>
<p>On a limited corpus (i.e. for a search appliance), then human rating, pinning, adding extra meta data is extremely extremely useful (why I loved building .Search on top of Community Server)</p>
<p>For an unlimited corpus - no way!  That&#8217;s what we&#8217;re talking about right?  Users altering ranking modifiers as opposed to bots?  I agree 100% then that humans need to be creating/updating rules in the rule engine, but definitely not altering ranking modifiers a site at a time!!!</p>
<p>Is it ever really possible to &#8216;watch&#8217; peoples behaviors when they search, and make meaningful ranking modifier changes, or do what I do in .Search - pin web pages/web sites?  </p>
<p>I love Scoble&#8217;s search posts - its good to see them get attention.   I&#8217;ll have to read this exchange over and over to truly &#8216;get it&#8217;</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Cybermagellan</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-188</link>
		<dc:creator>Cybermagellan</dc:creator>
		<pubDate>Tue, 04 Oct 2005 22:50:05 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-188</guid>
		<description>Right...so to help index it better if you knew a search engine used a more human approach would be to ask like Q: How do you unlock an AudioVox 5600. Then post A: And give your steps. That way when the spider hits your site it would index the part that says Q: "foo" or whatever and then when someone ask that question to the search engine the relevency would almost match at 100%.

I posted more about it at &lt;a href="http://spaces.msn.com/members/cybermagellan/Blog/cns!1pFqApB9d9HtlLrOmjssCQ4A!119.entry" rel="nofollow"&gt;Perfect Searching is done by humans&lt;/a&gt; (Sorry if that comes out plain text.</description>
		<content:encoded><![CDATA[<p>Right&#8230;so to help index it better if you knew a search engine used a more human approach would be to ask like Q: How do you unlock an AudioVox 5600. Then post A: And give your steps. That way when the spider hits your site it would index the part that says Q: &#8220;foo&#8221; or whatever and then when someone ask that question to the search engine the relevency would almost match at 100%.</p>
<p>I posted more about it at <a href="http://spaces.msn.com/members/cybermagellan/Blog/cns!1pFqApB9d9HtlLrOmjssCQ4A!119.entry">Perfect Searching is done by humans</a> (Sorry if that comes out plain text.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Erik Selberg</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-186</link>
		<dc:creator>Erik Selberg</dc:creator>
		<pubDate>Tue, 04 Oct 2005 21:00:08 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-186</guid>
		<description>So one thought is thinking about query intent... e.g. if the query looks like a "How do I..." then returning pages that look like they answer "How do I...." questions. It's a refinement step, certainly. 

Here's a more concrete example... I get a bunch of hits on this blog (not to mention links) to my post on &lt;a href="http://www.selberg.org/2005/07/22/unlock-an-audiovox-5600-for-free/" rel="nofollow"&gt;unlocking an AudioVox 5600 for free&lt;/a&gt;. The reason is after I figured out the instructions from a ton of different sources, I decided to put up The Answer and make things simple for people.

That being said, there's a lot of The Answers out there for questions that haven't been put up by someone... and I'm sure a lot of it is simple knowledge that once you go through it, you know and forget. Like setting up GPRS, for example.</description>
		<content:encoded><![CDATA[<p>So one thought is thinking about query intent&#8230; e.g. if the query looks like a &#8220;How do I&#8230;&#8221; then returning pages that look like they answer &#8220;How do I&#8230;.&#8221; questions. It&#8217;s a refinement step, certainly. </p>
<p>Here&#8217;s a more concrete example&#8230; I get a bunch of hits on this blog (not to mention links) to my post on <a href="http://www.selberg.org/2005/07/22/unlock-an-audiovox-5600-for-free/">unlocking an AudioVox 5600 for free</a>. The reason is after I figured out the instructions from a ton of different sources, I decided to put up The Answer and make things simple for people.</p>
<p>That being said, there&#8217;s a lot of The Answers out there for questions that haven&#8217;t been put up by someone&#8230; and I&#8217;m sure a lot of it is simple knowledge that once you go through it, you know and forget. Like setting up GPRS, for example.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Cybermagellan</title>
		<link>http://selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-183</link>
		<dc:creator>Cybermagellan</dc:creator>
		<pubDate>Tue, 04 Oct 2005 18:41:18 +0000</pubDate>
		<guid isPermaLink="false">http://www.selberg.org/2005/10/03/bringing-us-the-best-of-1995-to-you-here-today/#comment-183</guid>
		<description>Why not offer an algorithm based on the six key questions we all have, Who?, What?, When? Where?, Why? and How? Using this to perform your search it should have came back with results similar to "how do I set up GPRS?" with a link to the China Mobile website. I've been thinking about this for awhile now and find that most the time when we go to Google,MSN,AOL, or Yahoo! we find ourselves searching on a question that we have.

Offering the answer back as to that question this should be able to pinpoint better than a static search. Just my two cents.</description>
		<content:encoded><![CDATA[<p>Why not offer an algorithm based on the six key questions we all have, Who?, What?, When? Where?, Why? and How? Using this to perform your search it should have came back with results similar to &#8220;how do I set up GPRS?&#8221; with a link to the China Mobile website. I&#8217;ve been thinking about this for awhile now and find that most the time when we go to Google,MSN,AOL, or Yahoo! we find ourselves searching on a question that we have.</p>
<p>Offering the answer back as to that question this should be able to pinpoint better than a static search. Just my two cents.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
