January 21, 2013 at 13:24SeoPult Max - by leader patterns
“Spadeville, are you human or more than human?” Mascall asked.
- One who is no more than a man is nothing.
- Where are you from?
“From reflection, Maskall.” No other mother can give birth to truth. I thought and rejected, and thought again.
D. Lindsay, Journey to Arcturus
What is the main question of modern SEO? "How to bring the site to the TOP for certain requests"? Nothing like this. In fact, the most important thing is the cost of the target visitor, and the question is: “How to attract the maximum targeted traffic for the lowest possible money?” And although the share of content in search promotion costs is constantly growing, the main budget item remains links. You can buy them in the right volumes only on exchanges. But when it comes to specific actions, it is possible and necessary to fall into a stupor. How many links to take from TIC 10, 20, 500, etc. to advance a specific request - for example, "lithium batteries" - in a certain period of time? But links and donor pages have a lot of important characteristics, not only one TIC (for example, the system described below works with 184 criteria).
The purchase of links to the eye should remain in the long past first decade of the 21st century. Yes, by gaining links to decent donors, you can display a pre-optimized site in the TOP, it’s not a big deal. However, you can spend almost 5 times more money with this approach compared to the achievable minimum, and with sufficiently high competition this automatically knocks you out of the battle for the top ten issuance, if not from the business as a whole. Fortunately, nature does not tolerate emptiness - if among the legions of legendary school-year students there is a niche for a scientific approach and mathematics, it will be filled.
Read more
A New Approach to Empiricism in SEO
About two years ago, we decided to develop a methodology that takes the experimental approach to promotion efficiency to a new level. If the basic SeoPult algorithms work on the basis of cascades of filters, manually adjusted by our analysts according to the averaged results of experiments on a large, but still limited selection of “acceptor-donor” complexes, it was decided to make the new system: a) automatic and based on data collection and processing in real time; b) working with specific requests, and not with their groups; c) analyzing much more factors than is available to a living specialist; d) capable of highlighting success factors for current extradition leaders; e) capable of applying these factors for the purchase of reference mass,
To develop the mathematical principles of SeoPult Max technology, specialists from AlterTrader Research were involved (founded in 2004 as a research laboratory, whose priority was the development of scientific methods for analyzing financial markets and system trading technologies. Currently, the company has three departments: trading, financial mathematics and exploratory research). The result was two articles and a methodology to which we refer mathematics lovers: “An algorithm for selecting the most effective set of donors for website promotion in search engines” and “Implementation and analysis of the effectiveness of the method of constructing the optimal set of donors for website promotion in search engines”. Below we present a version of the popular article by Ilya Zyabrev, Oleg Pozharkov and Irina Pozharkova adapted for Habrahabr .
By the way, the algorithm is very demanding - both to incoming data received by diverse and rather non-trivial parsing, and to processing. Even for the initial launch, a mini-cluster of servers was required, and now, due to the high demand for SeoPult Max and the enormous amount of computation, the prefix “mini” is hardly appropriate.
Frequency distribution
Consider an example - an acceptor that has 100 links from donor pages. Let the TCI of donors be from 0 to 950. We divide the segment 0 ... 1000 into 10 equal parts (0-90, 100-190, etc.) and calculate how many values of the TCI of donors fell into each of them.
| TIC | Frequency |
|---|---|
| 0-90 | fifty |
| 100-190 | 28 |
| 200-290 | 7 |
| 300-390 | 4 |
| 400-490 | 3 |
| 500-590 | 2 |
| 600-690 | 2 |
| 700-790 | 2 |
| 800-890 | 1 |
| 900-990 | 1 |
| Total | 100 |
The acceptor from example 50 donors, in which the TCI falls in the range from 0 to 90, 28 - with the TCI falls from 100 to 190, etc. The frequencies shown are absolute, but it is often more convenient to use relative frequencies, which in this case show the proportion of donors whose TCI falls into the corresponding interval. To move to relative frequencies, it is necessary to divide the absolute frequencies by the total number of donors, i.e. in this case, at 100.
| TIC | Relative frequency |
|---|---|
| 0-90 | 0.50 |
| 100-190 | 0.28 |
| 200-290 | 0,07 |
| 300-390 | 0.04 |
| 400-490 | 0,03 |
| 500-590 | 0.02 |
| 600-690 | 0.02 |
| 700-790 | 0.02 |
| 800-890 | 0.01 |
| 900-990 | 0.01 |
| Total | 1 |
For each factor (we use 184 factors that have been identified on the basis of previous experiments that reveal the influence of donor qualities on changing positions of test acceptors), we can construct a distribution. To construct one distribution, it is necessary to isolate the factor, take one acceptor and all its donors. We get 184 distributions for the analyzed acceptor, which form the frequency pattern , a very specific and clear characteristic of the reference mass (i.e., the set of donors).
The algorithm implemented in SeoPult Max builds frequency patterns for the TOP-50 of Yandex output for a given request. In this case, 50 x 184 = 9200 factor distributions are compiled to obtain 50 frequency patterns. On the basis of these distributions, the “ideal” frequency pattern of the delivery leader for a specific search query is constructed, which also consists of 184 distributions. For each of the factors, the “ideal” frequency pattern has properties that are clearly expressed in the TOP-10 (with gain to the TOP-3) and, as a rule, are less pronounced outside the top.
An example of working with individual factors
As an example, we show the principle of calculation for three factors:
TRaslm - donor relevance to a given search query. On the donor page in the anchor, links to the acceptor may be the words of our request. Also, such words can be in the text of the donor page, or there are similar links to other sites on it. From a search engine perspective, the relevance of a donor to a query will be clearly greater than zero. The TRaslm factor is considered on the basis of the relevance formula aSLM:
" where t are the lemmas of donor words D,
Q is the set of lemmas of query words,
aSLM is the approximated spectral language model.
In other words, TRaslm is the sum of the values of aSLM donor lemmas that are found in the query. Lemma in this case - the canonical form of the word.
TRCross- the index of mutual relevance of the donor and acceptor, which can also be called "pseudo-thematic". It is considered to be a donor-acceptor bond. The texts of the donor and the acceptor may intersect in some set of words. For each word from the text, the relevance value of the document to the word query is calculated. The sum of such relevances at the intersection of words in the texts of the donor and the acceptor, divided by the sum of all the relevances of the words from the texts, will give the value TRCross. The formula for the factor looks like this:
where td are the lemmas of the words of the donor D,
ta are the lemmas of the words of the acceptor A.
HostDist is the “distance” between the hosts. It is considered for a donor-acceptor bond based on their IP addresses. This distance is used by Yandex to calculate the HostRank factor. Formula Factor:
where n is an integer from 0 to 31, the number of the most significant bit at which the IP addresses of the hosts of donor D and acceptor A are different. If the IP addresses match, then HostDist (A, D) = 0.
We calculate these factors for a specific request "website promotion". At the time of writing this article by the AlterTrader Research team, the TOP-10 issuance of Yandex looked like this (we took the first 10 sites on request, excluding the "spectral" impurity).
The distributions of the leading acceptors under consideration for each of the factors can be reduced to one graph.
1. Distributions by factor TRaslm
2. Distributions by factor TRCross
3. Distributions by factor HostDist
The graphs constructed for each site from the TOP-10 have a similar appearance, on the basis of which one can single out a regularity for each of the factors and build an “ideal” frequency pattern.
Example: “Ideal” frequency pattern for the TRaslm factor.
If we build similar averaged patterns for each factor for sites outside the TOP-10, we will get significant deviations from the ideal leader patterns, and the further from the TOP, the deviations will be stronger. The algorithm implemented in SeoPult Max is able to minimize deviation by changing the reference mass of the acceptor so that its pattern matches the patterns of the current Yandex search leaders for a specific search query.
Currently, all SeoPult clients can connect the SeoPult Max algorithm to their projects, which performs such an analysis on 184 factors, while constantly matching the distributions to the current pattern. The restructuring of many donors takes place within the framework of a given budget and the principle of minimal changes to an existing set.
Conclusion
“Thinking should be cool and at peace, and our poor hearts are pounding too much, our brains are too hot for this,” Melville wrote in Moby Dick. The artisanal, “too human” approach to promotion in 2013 will not give optimal results - only complex mathematical methods of processing a large amount of information (incidentally, obtaining it is also not so simple) allows us to develop a guaranteed optimal sequence of actions.
