|Alltheweb: "how to make pizza" changed|
Algo change or hand tweak?
(see [webmasterworld.com...] #52)
I was sure it will happen so took a shot:
How to make pizza 06-14 [howtomakepizza.batcave.net]
How to make pizza 06-21 [alltheweb.com]
Hand tweak or algo change? I tend to think hand tweak, cuz they did not bother with SERPs past top ten: mostly irrelevant/spammy as it was in top 10 before tweak.
p.s. I am too concerned about Google domination, like many others. But in my eyes Alltheweb search quality was and is worst of all majors.
Maybe you are getting something different - but they look the same sites listed at my end.
I don't see it either, now. All top 10 was more less about making pizza, further mostly unrelated.
Saw the new serp too, rzfree. Looked improved.
Also + 50.000 pages found. ATW has updated at least parts of the db.
FAST is known to permanently do little tweaks to the algo.
So, the new serp looks far more relevant, given that it's a nearly natural language, and thus rather "uneducated" query.
An "educated" query would have been to put the phrase, or better yet the parts of the phrase into quotation marks.
"how to" "make pizza" [alltheweb.com]
But of course only very few users will apply such "advanced techniques", so any major SE must aim at handling queries as they come.
FAST monitors queries, and I'm pretty sure they have a system in the works, which will try and learn from those queries.
On the right hand of ATW serps underneath the Topics, often an additional navigational feature pops up, called "narrow your search".
I've been speculating a while back on the source of the suggestions made there. I'm pretty sure now, those are derived from monitoring popularity of queries.
Hey rzfree - looks like you helped FAST out ;)
"nearly natural language"
Norwegian or English?
I'm just wondering how many of the staff at FAST are fluent in English ( the predominant search language) and if they have any native english speakers on board.
I ask because I believe that it would be hard to create a natural language search facility without an ultra strong grasp of the language involved.
They DID tweak their algo, because the unique name of my web design studio now comes up at number 1.
Formerly it came up (I think number 19), behind a radio station and many other unrelated serps, that had the keywords sprinkled in its body text.
Just to clarify: I did not say FAST claims being a natural language SE. I said the query from rzfree's example is nearly a natural language query.
John, FAST has staff from and operations in the US and the UK. They serve queries in nearly 50 languages.
You wouldn't assume Google or AV are not capable of dealing with results from French or Spanish, because they are US based?
> English predominant search language
Probably. Won't last long though. Less than 50% of the worldwide inline population is natively English speaking, with falling tendency.
This is important to understand for any SE with global ambitions.
There's definitely been an algo tweak... I'd posted a while back that on searches where a common phrase like "San Francisco" is a modifier, they were grouping "San Francisco" as a common phrase... which makes sense... but that grouping was giving far different results from San Francisco not grouped, as if they were giving a lot of extra weight to the common modifying phrase.
They seem to have adjusted this. The grouped and ungrouped results are still different, but the modifying phrase has been dialed down. Google is still having problems with this, I think because of link relevancy.
I'm also seeing some changes on rankings of some interior pages that I can't explain... which I guess is the way they want it. ;)
>The grouped and ungrouped results are still different, but the modifying phrase has been dialed down
Sounds convincing. Good explanation, which I'd love to test out. Unfortunately I do get the old serps again, even via anonymizer.
Robert, you are seeing the new serps?
>>Robert, you are seeing the new serps?<<
I think I am, but I've got to confess I didn't record what I saw on June 22. Looking now, on one search, the sites I'm monitoring are still in the top ten, which they hadn't been when the grouping came in. They were way up there before the grouping.
AND, when I just punched the search button again, they moved up a few places... from 9 and 10 to 6 and 7... probably closer to what I saw on June 22 that made me happy.
BUT, running the same search again, they've dropped back down to 9 and 10, and they seem to be staying there.
PS to the above...
Just checked Lycos too. The #9/6 result shows as #2, which is roughly where it had been before the grouping came in. The #10/7 result, though, (which once upon a time had been 4 or 5) isn't even in the top 30. So yes... it's a hard theory to test right now. ;) I'd definitely hold the pepperoni on that pizza.