|semantics: can search phrases devalue other adjectives?|
<noun phrase> vs <adjective> <noun phrase>
| 5:16 pm on Oct 28, 2004 (gmt 0)|
Take three search phrases, generic example:
1) product type
2) class product type
3) subclass class product type
In the search results for #1, it seems like the class may be used as a silent discriminator. Listings not of that class tend to show first.
The effect can even be more pronounced between #2 and #3. For a real case, I couldn't find a search result mentioning the subclass until #362 of the SERPs. The subclass is not rare.
Is Google using such semantic distinctions, or or my observations just based on chance or other in-page factors?
[edited by: ciml at 10:03 am (utc) on Oct. 29, 2004]
| 10:05 am on Oct 29, 2004 (gmt 0)|
Perhaps in these cases widget has more competition than fuzzy widget, blue fuzzy widget has more competition than fuzzy widget.
| 5:11 pm on Oct 29, 2004 (gmt 0)|
In this case, I'm pretty sure the "competition" is of a semantic nature, not marketing forces. Most people searching for blue fuzzy widgets only have a hazy idea of the difference between circular blue fuzzy widgets and rectangular blue fuzzy widgets.
What I've noticed is that unless "circular" is actually specified in the search phrase, Google is deciding that they don't want to see the circular type, or even the rectangular type. So authorative sites which choose to be specific are semantically penalized, by hundreds of places in the competitive field I am actually talking about.
The penalty goes away for knowledgeable searchers. If they know to search for circular blue fuzzy widgets, Google does a great job. Unfortunately, most searchers don't know they need to be that specific to get what they want. Common parlance is "blue fuzzy widgets", and that's what they search for. Google is making a semantic distinction in this case which is contrary to what most searchers would desire.
I like the idea that Google is getting smarter and making semantic distinctions. In the long run, that's a good thing. What's bad is hiding the decisions it has made on behalf of the searcher. In this case, it could show the current SERPs as it does now, but offer three alternative, more specific searches.
<searcher has specified blue fuzzy widgets> circular blue fuzzy widgets
* Do you wish more specific results?
*** rectangular blue fuzzy widgets
*** circular OR rectangular blue fuzzy widgets
The third search might yield the best result for those hoping for authorative information on the product. I'm not saying Google should try to generally "bin" its results as other search engines have tried. My idea would take effect only where the SERPs have been tuned based on semantic factors not obvious to the searcher.
| 6:17 pm on Oct 29, 2004 (gmt 0)|
I'm not so sure on the "authorative sites which choose to be specific are semantically penalized" part, but think you are absolutely right on the offering of topical, contextual or clustering type of search term suggestions.
Nearly every other search engine or meta-search engine offers this.
(Altavista "related searches", Vivisimo "clustered results" and Alltheweb with "refine your results" etc..)
Even more so as the average search term usage of a searcher is still about two words, this should be a default IMO.
If I'm not mistaken Google said they sample tested this but found response unsatisfactory..
Often if I search in unknown territories I'm very happy with the terminology suggested by these clustering type of search engines. Who am I to know exactly what to search for, or which nomenclature to use?
Yet with all the "Google services" options and even the "Google labs" I fail to see a simple opt-in option for this type of service on the general Google search.
| 6:39 pm on Oct 29, 2004 (gmt 0)|
Does a tilde search on either product or type return class?
| 7:54 pm on Oct 29, 2004 (gmt 0)|
You're likely right -- offering search clusters might be generally useful. I was thinking only to do it if semantic decisions had been silently made, but I can't really justify my reasoning now. ;)
It's kind of hard to do without adding quotes into the equation. I tried ~"blue fuzzy widgets" and the discrimination against sites containing the terms circular and rectangular seemed to go away. E.g., a site specifying the rectangular subclasss showed up at position #73. Without the quotes and tilde, rectangular wasn't present until listing #362.
Ooops, I just repeated the search with the quotes and no tilde. The tilde didn't matter. The same site still appeared at position #73. Maybe tilde doesn't work with quoted phrases. (I had never used it before. ;) )
The problem is that "type" is extremely generic and "product type" isn't much better. Common-speach differentiation begins at "class product type".
If you want to know what I'm actually talking about, PM me. Webmasterworld rules prevent me from being specific in a public post. (The guys who really make widgets must love it here, though! :) )
| 9:50 pm on Oct 29, 2004 (gmt 0)|
Sorry I misread your post the first time around. I think what your seeing is the fact your subclass isn't sought after by the SEOs competing for the class product type search.
Compare "class product type" to "subclass class product type" in the Overture keyword to see why.
| 4:30 am on Oct 30, 2004 (gmt 0)|
What I see is lots of alternatives for "class product type", but none for "subclass class product type". In actual fact, though, "subclass class product type" is the traditional and common form. Most people wouldn't realize that, and just speak of "class product type".
My point is that an added word on the page seems to be causing Google to push such pages far down in the SERPs, even if the page is content heavy and has equivalent or higher PR to listings on the first and second SERPs. To me as a searcher, that is counter-intuitive.
| 9:40 pm on Nov 1, 2004 (gmt 0)|
|Most people searching for blue fuzzy widgets only have a hazy idea of the difference between circular blue fuzzy widgets and rectangular blue fuzzy widgets. |
In that case, will a site about "fuzzy widgets" in general not suit the user better? If they searched for "fuzzy widgets" and got results relating to "circular" and "rectangular" widgets then would they be confused? It seems to me that the current settings in Google mean that a user will always be shown general information if that is all they have asked for.
I would imagine that this would lean toward agreeing with the idea that Google DOES penalise more specific pages but this would require extensive research in semantics and worldwide product knowledge that I'm sure Google would just not be interested in attempting ....
| 7:13 pm on Nov 2, 2004 (gmt 0)|
The term in question is actually inherent in a large portion of the particular product class. Most of them are "rectangular", so people often don't bother saying it. (It's tedious to make a three word phrase into a four word phrase unless you have a reason, especially when you only have a tenuous understanding of what the fourth word means.)
your_store's theory is that the word doesn't show up in the first several hundred listings because SEO's avoid using it (or at least see no purpose in adding it). That's possible, I guess. Your theory seems to be semantic interpretation of search phrases would be enormously complex and hence unlikely. Maybe you are both right. I still tend to think there is some semantic interpretation going on, but I have no proof to back it up.
| 7:27 pm on Nov 2, 2004 (gmt 0)|
|the word doesn't show up in the first several hundred listings because SEO's avoid using it (or at least see no purpose in adding it). That's possible, I guess. |
I'd say the SEOs don't even know the 4th word exists. You have to realize most of the people competing for the phrase don't have an intimate knowedge of the product. They only know to call it "class product type"; because, that's all the OV tool will show them.
| 4:47 pm on Nov 4, 2004 (gmt 0)|
I hadn't thought of that. The situation is probably even harsher. Most of these sites exist to sell the cheapClass version of product type (that's not a judgement on the validity of the product, just its price). The class for which I'm interested isn't exactly cheap, but it shows up frequently in the OV tools. The internet marketers have a very low probability of selling the higher-level product. Many feel obligated to mention it, though, in glorious ignorance and cut&paste splendor. :(
The major players who desire to sell the high-end product maybe "dumb-down" the information they provide via the internet. The information is there, but it ends up buried in glossary entries or page two of 3-page product descriptions. Actual sales are mostly dependent upon face-to-face visits.
In summary, the answer to "can search phrases devalue other adjectives?" may be no, at least for the time being. Webmasters OTOH can offer sufficient ignorance in sufficient volume to bury informational sites beneath hundreds of pages of repetitive [Buy Now and Save!] buttons and superficial fluff.
| 8:52 am on Nov 22, 2004 (gmt 0)|
There are different ways of using some words, both as adjectives or nouns in some cases, but is it only "widgets" that's the noun with all the other descriptors being strictly adjectives, or could the others be used as either nouns or adjectives?
Example of what I mean:
fuzzy blue widgets = 2 adjectives, 1 noun
fuzzy dog widgets = 1 adjective with 2 nouns, one of which is being used as an adjective even though it's a noun - dog
labrador dog widgets = labrador and dog are modifying widgets, the noun, so are in usage serving as adjectives, but in actuality they are all nouns.
Not the best example, but it's too late to think too much. I do believe it isn't totally impossible that filtering could be done having something to do with parts of speech in phrases.