Geek magazine hacker daily blog

7 years, 2 months ago
Practically each CEO-optimizer (in a prostonarodiya a seoshnik, website promotion, internal optimization, drawing up a semantic kernel to order, to purchase windows of pvc … — as we got used to be represented, I apologize for attempt to yumornut) faced correlation problems of recognition of requests. And also with different grimaces, defects in search ranging.

This short publication will be devoted to this problem. Where I in a makeweight will try to give a row enough entertaining examples.

Synonyms and sheaves

We will begin, perhaps, from far away — with the fact that searchers have big base of statistics. Power and which influence you, probably, already felt. And the biggest problem was, is and it will be possible to be "a problem of understanding of the person" (what a sin to conceal, even we plainly do not understand each other). And some banal request, for example: "mittens" misleads a search engine as it is unclear whether the technique of production of mittens is interesting to the person, whether he wants to purchase them, whether it simply needs the picture, whether wants to learn about an origin of this word, al he in general meant a mouth (slang). And only the right technique of a solution is collecting of statistics and carrying out on the basis of collected information the analysis (or to connect to the equipment very bright machine for a thought-reading).
As the simplest mechanism of understanding of turnovers, slang expressions, prostorechiya (and so forth) in searchers begin to communicate words synonyms, and also words-transliteratsionnye and foreign-language, plus case and numerical variations (for example, "dolls", "dolls", "dolls", "kukla", "kukly").

Attention to selections of key phrases:
Поисковый запрос: куклы
However in different searchers PS "Yandex" at the request of "doll" have no different sheaves, for example, sheaves with the phrase of "dolls". And sometimes there is also, for example, as in request "Car":
Поисковый запрос: Автомобиль

And they communicate different, some in complete synonymous compliance, some are not present (that can be traced, actually, on issue) as occasionally it by words synonyms is also different.

Yandex and German

There is also other situation which to pain illogical also is small defect of the search engine "Yandex".
In particular we will consider recognition of a certain request — "German" and its interrelation with English. The difference between these phrases is available, so? However, "Yandex" watches at this situation a little from a different angle:

We enter request "German". Also we watch issue:

Поисковая выдача ПС Яндекс по запросу: Немецкий язык

Well, to whom does not happen, perhaps storms issue a little (though request high-frequency, what storm), perhaps accidentally somehow wormed way (though what randomnesses that 1-2 times, system). But is farther more. — We did not manage to pass to the second page of issue as on the 20th place we are met here by what client:

20 позиция запроса: Немецкий язык

Well, it is possible to think … All the same on the main page there is a phrase "German" what there. But on 44 positions we are met by other beautiful website, that phrase is not on which at all.

44 позиция запроса: Немецкий язык

And all attempts to find such coincidence inside only of this website ( crowned a failure (only rare single phrases and general "illumination" of request "English").

As a result, it is possible to draw a small conclusion that the search engine "Yandex" in our case in such request as "German" considers the word "language" as the predominating factor (as well as in requests of type: "a white penguin" on a noun — "penguin"), and determination "German" only an insignificant secondary factor. Moreover, on such keys the search engine even adds language variations (same "Deutsch" for example)

Subtleties in selection of phrases for a semantic kernel

Quite logical will be to organize all this information in something efficient now, than it is possible to use in practice. Namely — the analysis of these subjects will help us with creation of a semantic kernel, and also with promotion of a resource.

Semantic kernel

The webmaster or prodvizhenets are often faced by a problem of the choice of key phrases. However, the choice of this or that synonym for a key phrase can be fatal. As often happens here is how (for example, Google):
Bratz — some "doll" brand:
"Bratts => Bratz", but is not equal "Bratz";
"Bratz => Bratz", but is not equal to "Bratts".

or as "on-line" contacts "online" and "on line", but does not communicate about "online" and "online".;
At the same time "online" communicates about "online" and "on line", and also "on-line", but does not contact "online".

As a result it is always better to select core like "online" or as an output to go till "nch-way", there to whom as.
And you should not forget about complexity. All want to kill two hares ("Yandex", "Google"), at once in this case and it is more desirable to look for average value among both PS.


Similarly there are many ways of an unikalization of anchors. It both a fringing unique okolossylochny, and introduction to anchors of a stop word, and temporariness overlaying / padezhnosti, and dilution of a key on LF, and separation of keys in links. But as an alternative method to an unikalization reference, but transfer at the same time uniform usefulness on a key can just use phrases synonyms, it is desirable for a darkness in which issue has approximately similar character, and phrases vzaimoopredelyatsya in both keys (often happens to brands: Adidas, Mercedes, Opel, etc.), often noticed that at increase of reference weight, at least on one of two synonymous keys, another rises by a little similar coefficient.

About the publication

In a post only my confused opinion on the above nuances in SEO or "fortune-telling on a coffee thick" is displayed (as it is pleasant to whom more). With pleasure I will participate in discussion, and also in search of truth if it is provided possible.

And finally: "Everything SEO is reduced only to determination by search bots of "the care level" of the user". Therefore, webmasters/administrators/moderators, care for visitors! The whole world keeps only on care!

I thank for attention.

This article is a translation of the original post at
If you have any questions regarding the material covered in the article above, please, contact the original author of the post.
If you have any complaints about this article or you want this article to be deleted, please, drop an email here:

We believe that the knowledge, which is available at the most popular Russian IT blog, should be accessed by everyone, even though it is poorly translated.
Shared knowledge makes the world better.
Best wishes.