Processing and final ranking of feature weights (i)

Intermediary transaction http://www.aliyun.com/zixun/aggregation/6858.html ">seo diagnose Taobao guest cloud host technology Hall

Prepare to write an article about SEO optimization details, because we often said SEO to start from the details, but talk about talking about it, why? Because no one knows what details to pay attention to, understand all the details also mastered the SEO mechanism, this is the most difficult, In other words, the most basic is often people do bad. This article no longer narrated, first said the characteristic weight question.

Do SEO do not study the characteristics of weight I feel is a very absurd thing, but it seems that the big boys do not love to say, even the most professional forum in China, you see the chain, content, anchor text, server speed and other details, you can not even search for any new gadgets, this is not normal.

For the feature weight, its main role is to affect the SERP final results of a series of factors, and each different factors affect the stage and time is not the same, the scope of the impact is not the same. If you are a line of seoer, you have mastered the characteristics of what weight, you also need to know how the spider is handled, because only after processing the feature weight is intuitive data display.

There are only two types of data on the Internet, discrete and continuous. Continuity can be counted, such as the number of chain, keyword occurrences, and so on, these factors are processed directly using the count, but because some quantities may be very large, must take logarithm to reduce the impact, here is a problem http://blog.sina.com.cn/s/blog_ 5dac285301010jpj.html is a large number of times, once the logarithm will not reduce the impact, zero here said the mass comment network keyword density is too high, here is actually a lot of problems, in a certain measure, regardless of whether or not to take the right, the number is a positive effect, but the role of the reduction, which is such as an acceleration problem, force Gets smaller, the acceleration gets smaller, but the speed is still increasing ing. This is a mistake.

For discrete data, the general use of SVM processing mechanism, more is a number of latitude of an index, the most important here is how to add the degree of size to the discrete data, is the use of the transfer of power or to the different weight distribution is not the same proportion. Regardless of the calculation, the final results will be uncontrollable, but when the indicators become more often, uncontrollable factors will be greatly increased, but at the same time more to meet people's matching.

Baidu and Google the biggest difference is that the amount of indicators is not the same as many, Google, so the need to deal with strong, but the final ranking will be more objective, more broadly matched, Baidu's indicators are less, showing a prominent middle of the state, which is the difference between the two results ranked the most.

What's the use of having so much to say that someone must be impatient? I said this role is too big, for example, a lot of people do not understand why a certain page ranking abnormal good, how to think about why, in fact, this is the impact of discrete variables, discrete variables present one of the largest characteristics of the results of the randomness increased.

As for how to use feature weight to deal with problems, there is one of the most important is the combination of our SEO, many times she can guide us, such as a factor to do too well, because will go to logarithm, the impact will be reduced, we grasp this point can be the efficiency of the theoretical increase of 10 times times; She can also help us explain a lot of problems, such as according to the spirit of SE, when a website original article is no longer included, many people do not know why there is such a mechanism, kiss you know why?