User demand left search engine optimization right

Today, with SEO relevance or Web page ranking topic "Frequency or density" content is still heard, the past things let him past it.

First, we start with the frequency and density of the keywords to describe the first content of this article.

Keywords frequency and density

It seems we have always thought that when a user searches for a keyword, the frequency and density of the keyword appearing in the Web page is better, and of course there are some views that this value is controlled by 3%-5%, but do you know where this number comes from?

If I remember correctly, this is the number that appeared in a book written by an American SEO expert 06-07 years later, after the book was translated into a flood, and flooding was a very quivering word.

Even if this is the right and the past, if the book is written in 07, does it take 1 years to write a book? That means the concept was in 06, and practice proves that it doesn't take at least 1 years? It's been 7 years since the earliest possible start of this approach, and we're still skeptical about the IQ of search engine development engineers?

Even if the above content is forged, then we use an example to consider the phenomenon of keyword frequency and density, suppose we search for a keyword "Linyi eight or nine talent network", if according to the frequency and density to think about the problem, the emergence of "Linyi Talent Network" and "eight or nine points" of the Web page to get the most relevant pro-Lai.

Then, as the user of us, we search "Linyi eight or nine talent network", hope that the results of the feedback is a click to get the information, or a lot of articles filled with keywords? (rather than read novels)

If you do not understand, this article can be seen here in the upper right corner of the X button, or ALT+F4.

I always hope that by making everyone stand in the user's perspective to consider SEO or search engine, but a lot of friends send messages to me always ask some I do not know how to answer the deviation problem.

I hope we can understand a word, to meet user needs is the survival of the enterprise, Baidu it is just a business, only this, Baidu's audience on the user, is you, me and you my side of these people.

For SEO, become the industry leader needs to meet the needs of users, to obtain more traffic in the industry, this is the SEO on the one hand, but also I am good at, if Sir you have to say xxx marketing, XXX promotion concept, I am sorry, I do not understand. I can only from the user and technical point of view to describe my SEO.

OK, now let's talk about how the search engine calculates the relevance between the keyword and the page from a technical point of view.

As a user, we need to enter one or more phrases on the search engine to query what information we need to find. For example, the above mentioned "Linyi eight or nine talent network", at this time in less than 1 seconds of the search engine returned to us a lot of results, then it is how to think that the row in front of these results for search users useful, Or how does it calculate that the keyword entered by the user is closely related to the result of the return? Although the following is from a technical point of view to answer, but still is to meet the basic premise of user needs.

When we submit to the search engine a phrase, the search engine will be the word processing, which is also known to all, an early age of Baidu outsourcing service is Chinese participle, and now the vast amount of technology like.

The function of participle is to divide the phrase of the user into the core keyword of the independent expression meaning, why should this step be carried out? The answer is that the search engine needs to put the user input keyword segmentation after the Independent keyword needs matching finally integrated to calculate the user input keyword expressed meaning. (There will be a description later, just a preset here)

Here assume that we entered the keyword query, word after the Q1,Q2.......QN, for search engines, directly to the user input a phrase to match the needs of the type.

Another concept is that query or Q1 may express different needs, like the pronunciation in Chinese characters, or a word in an English word that represents a different meaning.

Search engines have a set of algorithms to calculate keyword requirements (familiar with search engine development friends should be able to understand what I said): "Word demand probability model."

The model uses a set of more complex formulas to express, need to have strong understanding or mathematical skills to understand the logic of the people, I here in the vernacular to introduce you.

The so-called probability model of word demand is to divide the word before or word into three categories.

1. Words (words themselves)

2, the corresponding requirements of the type of words (like an English word represents a different meaning, may have meaning 1, meaning 2, etc.)

3, words corresponding to the demand type probability (words match the meaning of 1 probability and accompany meaning 2 probability, etc.)

The following is a bit of a detour, I try to use the language I did not go to learn the skills to express clearly:

For the determination of the word before or after the word demand type, the search engine in the following algorithm is a more common:

1, through the existence of the search log to determine the requirements of the type of words

2, through the manual annotation to determine the type of word requirements

In both of these ways (the second is obsolete), you can map the q1-qn of a query word to the collection of requirements types that match it.

Maybe you have some questions about the last sentence, what is the q1-qn after query participle corresponds to the collection of requirements types that match with it.

For example, "Linyi eight or nine talent network" after the word "Linyi Talent Network" and "eight or nine points", which "Linyi talent network" may be matched to the types of requirements include:

1, video

2. Picture

3. Commodity

4, blog

5. Forum

6, the novel

The above 6 points and more types of requirements add up to the "Linyi Talent Network" a single word corresponding to the matching demand type collection.

As for the 3rd, it is clear that query corresponds to the probability of the demand type for each of the possible probabilities in the demand set.

So how does a search engine determine query requirements from the search log?

The search log records the user search time, user browser cookies (user ID), search keywords, search results URLs in the search results rankings, the user clicks the search results of the order, search results URL address.

That is, the historical user behavior can be matched to the requirements type of the most user query and the requirement type probability.

(Note 1: Here is a point, such as search "Linyi Talent Network", The video demand type probability is 0.5, the novel demand type probability is 0.3, cartoon demand type probability of 0.1, in 10 search results will appear in these three search results, according to the type of demand probability to sort, also explains why the keyword ranking will fluctuate one of the reasons, the timeliness of user demand type probability will also affect the ranking position. )

Understand the above content, also should be able to understand, the user query Word after Q1-QN needs matching and demand type probability after integration of the demand matching degree is query Word before the user needs.

This is just the type of requirement that matches query. So how does query's demand type probability be calculated?

Here also introduce a primer to calculate the query before the demand type probability, determine the demand type probability is also determined all meet the requirements of the type of page, according to the demand type probability matching user needs.

You usually get 1 or more child words after you search for a query word. May contain the requirement type in n, assuming that the value is 2, the search zombie will determine that the search results to be returned to the user should be: Video type search results and search results for the type of novel (see note 1).

The above content expresses the search engine how to determine the user query's demand, then how to calculate query and possibly participates in the rank page the correlation degree?

The search engine solves the core problem by solving the user's needs--understand the user search requirement.

For example, has learned that the search for "Linyi eight or nine talent network," the user is likely to find video class and novel information, then according to video class information to carry out the semantic analysis of the page, the general search engine semantic segmentation includes but not limited to the following:

1. Semantic analysis based on string

2. Semantic analysis based on computer understanding

3. Semantic analysis based on statistics

4, based on semantic participle

The above 4 content is not difficult to understand, there is also a point to note that the search engine will also filter some content, such as everyone knows, the search engine will filter some "," and so on, the search engine will generally filter:

1. Stop all words in the collection

2. Non-independent ideographic expressions

Illustration 2: Search engine will also use another technology to understand the needs of users, in general, users in the input of a query keyword, will be more important or can express the meaning of the core query placed in the head or tail. So the search engine will sometimes direct users to search the head of the keyword or the tail into the user needs of the demand type probability of the collection.

Now that the user needs are the video, how to match the video-related page is the last problem to be solved.

In general, there are two types of requirements:

1. Text class

2. Non-text class

Included in non-text classes but not limited to the following types

1, video

2. Picture

3. Commodity

4, blog

5. Forum

6, the novel

7, and so on

Simply so that you can determine the Web page and video category needs of the key words, the rest is the authority of the page to sort the value.