How Google rewards our presence on Twitter and Facebook?

The algorithm used by Google web pages to positioning is not usually something that controls a company. Still, almost all companies have clear the importance of being present in Twitter and Facebook to better position themselves, not just the need to be there if that is where the buyers are, but the importance of how to be present. Víctor Pérez Acosta speaks of this in this post: "Why social networks are important in SEO".

We talked several times about the importance of social networks in building online reputation. But now we speak of the importance of the links on Facebook and Twitter have in achieving a good SEO. Recently Google, the blog of Matt Cutts has confirmed that the measurements on these social networks are considered in the algorithm that calculates the final position of a site in search results pages. Expand on the concept.

It is true that mentions on Twitter and Facebook have already been considered in real-time search. The novelty is that the indications now are considered the natural position, with perhaps more important than a link elsewhere. The first thing you need to consider is that not all the links created on Twitter and Facebook have the same value. Google always tries to assess the popularity of the website and in this case taken as greater or lesser extent the popularity of the followers of each user. That is to afford a link on Twitter by a user who has few followers represent a minor benefit to do so by a user who has many followers. Consequently there would be a good idea to start a campaign of door openers in hopes of influencing the ranking of a site. 

How then capitalize on this new feature. Ideally, you face a social marketing campaign and achieve effective and stable presence in social networks through the creation of online communities. So if most of the efforts of SEO and SEO so far were only focused on "traditional SEO", it is time to seriously consider a campaign medium and long term include social networks mentioned. This is not to create profiles and keep people, but to achieve true interaction, failed to produce a response from the community and be considered a site of value for users.

Consider social networking as a construction technique of informal networks is not get to see the true value that they can have as power and mobilizing in the ranking of a website. Social networks have become a reference point for potential buyers and merit for themselves the effort that can be applied in developing their growth.

Read more »

Posted in: Blogger Essential,Facebook,Google,Invention,learning,SEO tool,Twitter,Web Informations,Wordpress

How to protect your copyright content of website from theft?

How to protect content from copyright theft?

Approaching it is time to big summer school holidays, and thousands of students flock to attracting and endless ocean of possibilities - the Internet, in order to cut the bubble out there, and by next Sept. 1 to drive up to the school on the new line of Lamborghinis.

Of course, the desire of young people is commendable, but our young innovators are always trying to follow the path of least resistance: "I want to do nothing, and in order to loot the stream flowed!" In this endeavor our "future Zuckerberg buys parsilku and begins to methodically steal content from sites and blogs , rivet doorways yes sattelity and pollute the search engines.

Search engines somehow trying to deal with this phenomenon, creating all sorts of filters, but
What remains to do to the victims of theft - for those who do not know firsthand that earnings on the Internet is a big job: the creators of blogs, for which every post no more, no less your own child, the owners of information projects, invest in content, to hire professional copywriters and Journalists?

Until recently they had to just sit and watch as the young innovators "free download" their content megabytes and prevraschiyut it in doorways and linkopomoyki for selling links. But not now!

How did it start?

On the preservation of the content on their own website especially do not think about it, yet, according to a famous Russian proverb about thunder and man, I do not find yourself in the role of victims of kidnappers content. So it was with us. One day I discovered that the site of my wife, which she so diligently and lovingly filled content of more than 5 years, razderbanil on Doric one (and maybe more than one) beginner black optimizer. I will not describe in detail the reaction of my wife on the news. Let me just say that the tragic situation lies in the fact that Yandex first indexed the stolen articles, not the originals from our website. How can this happen?

It is known that search engines love updated site. Search engine spiders visiting your site more often if it is constantly new articles. But the problem is that an ordinary blogger physically can not produce as much content. But the average thief - you can! Still - armed with a parser, it configures its own pseudo-site on avtopublikatsiyu stolen content so at least once in five minutes! In the eyes of search engines such site into a dynamic news - not less. Especially if you make it properly marinate Domain ...

How did I do?

I went to Google a solution. That's what I found:

• Scripts that disable selection / copying text in the browser - a kindergarten, and only. From the parser does not save, but ordinary users pobesit - for sure!
• Encryption of content and output it to the visitor through JavaScript - not suitable if the main source of traffic to your website - search engines.
• Plugins for WordPress, delaying renovation of RSS-feed - save only if the content is stolen only RSS-grabber, and even then not always.
• Some unbelievable database of malicious robots, mostly from China and Russia, it is unclear as there falling enjoyed by the U.S. government and the pair zabugornye university - apparently, they have there, too, sawing ...

Numerous discussions on blogs on this issue unanimously end conclusion: "You do not want to steal - not published on the Internet!" Maybe I'm bad looking, but it would seem, the problem has existed for a long time - should not exist and the solution. But no.

What you need to protect?

And need the following:

• Do not give access to content by any automated systems, except for a specified list of useful bots (search engines, RSS-aggregators, etc.), and all people freely admit to the site. In other words, we need even before you load the page to determine who is trying to download it: the boat or person.

So, not finding a suitable solution, I began to develop his own. I needed to make the system of content protection, which would be independent of the site engine, because my wife and a few sites, all on different engines. Two weeks picking documentation Web standards gave me the way to solve the problem. And the solution is lying at the intersection of several technologies. In the end, loomed a software package, which would simply be criminal not to make it public service!

What happened?

It took three months to develop ready-to-use system consisting of a "brain", analyzing incoming traffic to the site and the Personal Account with all the necessary settings, and billing.

In a small digression to point out that in assessing the labor should not rely on I specify a period of 3 months. Man-hours actually typed on the strength of the month. I just worked in between boiling the bottles for my son's four now:)

So, let me submit to your court my creations:



The system allows you to define BOTFILTER bots before loading the first page of the site.
This implies the main features of the system:

• prevent copying of content malign robots
• Spam Protection

Yes! Captcha is no longer needed! Robots which are able to recognize the captcha go forest:)

Of the extra buns are currently working:

• Speed-limiting access to the site. If some kind person will very quickly update your site any heavy page, you can make a bad site. BOTFILTER will not permit, and a couple of minutes will limit access to this rascal.
• DNS-editor to be placed on the service domain with the support of wildcards-records. Just a useful thing, if your hosting does not allow such.

Connect your website or blog to BOTFILTER very simple: just change the NS-server for your domain, and system settings include the real IP-address of your site. After this, all traffic to your site will go through the "brain" of the system: spiders, which you yourself have chosen, and people would normally see your site, and malicious bots instead of the site will be error "403 Access Denied".

How to check whether the system? Very easy too. After connecting and updating the DNS-records (your site should start pinging on the new IP-address), try very quickly poobnovlyat your site F5. Will give you a message:

You can also check job protection through the curl , or wget , if you are on Linux: just type
# curl ваш-сайт.ru
instead of the site and get the message "You are a bot, dear!", and going through a browser, make sure that the site opens as usual.

The system is quite reliable. The architecture is such that if BOTFILTER deny the "brain", the sites of customers will continue to boot normally. A more detailed description of the system you can find on the site botfilter.ru .

The plans for the future

Firstly, if I was not cast aside tomatoes in the near future I plan to tell habrasoobschestvu more detail about the internal structure of their system. Moreover, the happy owners of their own servers can build for themselves a similar system to protect content from theft, cross haproxy , nginx , Node.JS and Memcached . But those who do not have your own server, of course, is easier and cheaper to use my service.

Secondly, I plan to improve the resiliency of the system by increasing the server capacity and the purchase of 1Gbit-channel. At present the system in testing processes up to 5000 requests per second, but this is clearly not enough when suddenly one of the clients will pull hard real DDoS-attack.

Third, despite the fact that I did not plan to protect the sites from DDoS - for this is the special services , but in the near future BOTFILTER system will protect the site from overloading. We all remember the story of the site Skolkovo . So, if they used BOTFILTER, their site would not fall.

Well, in the fourth, I want to try to stick to the foreign market. The site's interface has been translated into English. Work is underway to translate the documentation.

Instead of a conclusion

If someone has already guessed how this all works, much to ask - do not write about it in the comments. Will not simplify the thief lives. Let them head broken.

Run the project with absolute zero, so that without support habrasoobschestva I can not do. If you have friends, bloggers or site owners do not consider for work to send them a link on this topic. I am confident that my service will still be useful to people.

Read more »

Posted in: Blogger Essential,Google,learning,SEO tool,Web Informations

Making Post title come before the Blog title in Blogger Blogs

How to Make Post title come before the Blog title in Blogger Blogs?

All the people are having the desire that, their blog should appear initially during the websearch in google, yahoo, bing, etc...,
But they will be dissatisfied.
It is because of below issue,
In blogger templates it's default format that

Blog title: Page title

This would cause difficult for SEO's like google, bing, yahoo to crawl your blog during Web Search's.
To overcome this issue you need to change the title format as

Page title: Blog title

Page title as infix and Blog title as prefix.

To do this follow below simple procedure.

Go to - Dashboard > Design >Edit HTML

Then you need to find the below code

<title><data:blog.pageTitle/></title>

Use Ctrl+F command

Finally replace the above code with the below code

<b:if cond='data:blog.pageType == &quot;index&quot;'> 
<title><data:blog.pageTitle/></title> 
<b:else/> 
<title><data:blog.pageName/> » <data:blog.title/></title> 
</b:if>

Now your blog is in superior state.

Read more »

Posted in: Blogger Essential,HTML,SEO tool,Web Informations

How Google ranks a web page?

Since a decade Google dominates the market for Internet search engines. Its strength is that it intelligently sorts its results by relevance. How is this possible? Since its inception in 1998, Google continues to evolve and most improvements remain closely guarded secrets. The main idea, by cons, has been published. the backbone of its success is a sound mathematical modeling that we track here.

What does a search engine?

A database has a predefined structure which allows to extract information, such as "name, street, postal code, phone, ... . The internet, cons, is unstructured: it is a huge collection of texts of various kinds. (I am excluding here of different formats and media, as we return to the search for good keywords.) A tentative classification is doomed to failure, especially since the web is rapidly changing: a multitude of authors Independent constantly adding new pages and modify existing pages.

 

To find information in this amorphous pile, the user can search by keywords. This requires some preparation to be effective: the search engine first copy the web pages one by one in local memory and sort the words alphabetically. The result is a directory of keywords associated with their web pages.

 

For a given keyword there are typically thousands of relevant pages (over a million for "tangent", for example). But some pages are more relevant than others. How to help users identify potentially interesting results? It is here that Google has made its great innovation.

The web is a graph!

Enjoy the little structure that is available. The internet is not a collection of independent texts, but a huge hypertext pages they refer to each other. To analyze this structure we will neglect the content pages and consider only links between them. What we get is the structure of a graph. The following figure shows an example in miniature.

 

 

Here the vertices represent web pages and the arrows represent the links, that is to say the citations between web pages. Each arrow points to the top issuer to the page cited.

 

In the following web pages I note with           and write          . In our graph we have a link   5, for example, but no link   1. However, in this first example, all pages communicate via paths to one or several steps.

How to use this graph?

The links on the Internet are not random but have been carefully edited. What information can we give this graph? The basic idea, yet to be formalized, is a link   is a recommendation of page P_j   to go and read the page    . Thus a vote of P_j   for the authority of the page    .

 

Analyze our example in this aspect. The following presentation of our graph suggests a possible hierarchy - even to justify.

 

 

Among the pages           P_1 $ is a common reference and seems a good starting point to search for information. It is the same in the group           where   P_9 serves as a common reference. The structure of the group           is similar, where   P_7 is the most cited. Note however that the pages     and   P_9, already recognized as important, refer to page D_5  . One might well suspect that the page D_5   contains essential information for all, it is most relevant. In the following we will try to formalize this classification.

First model: naive counting

It is plausible that a page receives a lot of important links. With a little naive, we also believe the statement converse: if a page gets a lot of links, then it is important. Thus we could define the measure of importance     of     page number links as   received    . In this formula is written as follows:

 

  M_i: = \ sum_ {j \ to i} 1.  

Here the sign     denotes the sum over all links pointing to the page    , and the words to sum all worth  . That is,     is equal to the number of "votes" for the page    , where every vote contributes the same value  . It's easy to define and calculate, but often does not meet the importance felt by the user: in our example there are       4 to       3. What is worse, this naive counting is too easily manipulated by adding pages without interest recommending any page.

Second model: weighted metering

Some pages emit a lot of links: they appear to be less specific and their weight is lower. We therefore share the vote of the page P_j   in   ell_j equally, where     denotes the number of bonds issued. Thus we could define a finer measurement:

  M_i: = \ sum_ {j \ to i} \ frac {1} {\} ell_j.  

That is,     count the number of "weighted votes" for the page    . It's easy to define and calculate, but still does not match well with the perceived importance: in our example there are       2 to     and    . And as before the count is too easy to fake.

Third model: Recursive counting

Heuristically, one page     seems important if many important pages of the quote. This leads us to define the extent of     recursively as follows:

  M_i = \ sum_ {j \ to i} \ frac {1} {m_j} ell_j.  

Here the voting weight   is proportional to the weight of     page issuer. It's easy to make but less obvious to calculate ... An efficient method will be explained later [ 3 ]. To reassure you can already see that our example does admit the solution   \ begin {matrix} & & & P_1 & P_2 & P_3 P_4 & & & D_5 P_6 P_7 & & & P_8 P_9 & P_ {10} & P_ {11 } & P_ {12} & \ \ m & = & (& 2 & 1, & 1, & 1, & 3, & 1 & 2 & 1 & 2 & 1, & 1, & 1 & ). \ End {matrix}   Unlike previous models, the page D_5   is identified as the most important. It's a good sign we're on the right track.

 

Note that  is a system of n linear equations with   unknowns. In our example, where   12, it is already difficult to solve by hand, but still easy on the computer. For much larger graphs we need specialized methods.

Random walk on the canvas

Before attempting to solve the equation, trying to develop an intuition. For this let us imagine a random surfer who wanders on the Internet by clicking on links at random. How is his position?

 

For example, suppose that our rider starts at time   on page   P_7. The only link to D_5  , so at time   1 surfer found there with probability  . By leaving three links, then at time   2 it is one of the pages P_6         P_8 with probability  . Here are the following probabilities (rounded to     near)   \ 

begin {matrix} & P_1 & P_2 & P_3 P_4 & & & D_5 P_6 P_7 & & & P_8 P_9 & P_ {10} & P_ { 11} & P_ {12} \ \ t = 0 & .000 & .000 & .000 & .000 & .000 & .000 & 1.00 & .000 & .000 & .000 & .000 & .000 \ \ t = 1 & .000 & .000 & .000 & .000 & 1.00 & .000 & .000 & .000 & .000 & .000 & .000 & .000 \ \ t = 2 & & .000 & .000. 000 & .000 & .000 & .333 & .333 & .333 & .000 & .000 & .000 & .000 \ \ t = 3 & .167 & .000 & .000 & .000 & .333 &. 000 & .333 & .000 & .167 & .000 & .000 & .000 \ \ t = 4 & .000 & .042 & .042 & .042 & .417 & .111 & .111 & .111 &. 000 & .042 & .042 & .042 \ \ t = 5 & .118 & .021 & .021 & .021 & .111 & .139 & .250 & .139 & .118 & .021 & .021 &. 021 \ \ \ dots \ \ t = 29 & & .059 & .117 .059 & .059 & .177 & .059 & .117 & .059 & .117 & .059 & .059 & .059 \ \ t = 30 & .117 & .059 & .059 & .059 & .177 & .059 & .117 & .059 & .117 & .059 & .059 & .059 \ end {matrix}  

There is a distribution which converges quickly to a stationary distribution (at     near the end of a thirty iterations). Test this observation by a second example, this time from the page    :   \ begin {matrix} & P_1 & P_2 & P_3 P_4 & & & D_5 P_6 P_7 & & & P_8 P_9 & P_ {10} & P_ {11} & P_ {12} \ \ t = 0 & 1.00 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 \ \ t = 1 & .000 & .250 & .250 & .250 & .250 & .000 & .000 & .000 & .000 & .000 & .000 & .000 \ \ & t = 2 & .375 .125 & .125 & .125 & .000 & .083 & .083 & .083 & .000 & .000 & .000 & .000 \ \ t = 3 & .229 & .156 & .156 & .156 & .177 & .000 & .083 & .000 & .042 & .000 & .000 & .000 \ \ t = 4 & .234 & .135 & .135 & .135 & .151 & .059 & .059 & .059 & .000 & .010 & .010 & .010 \ \ t = 5 & .233 & .126 & .126 & .126 & .118 & .050 & .109 & .050 & .045 & .005 & .005 & .005 \ \ \ dots \ \ t = 69 & .117 & .059 & .059 & .059 & .177 & .059 & .117 & .059 & .117 & .059 & .059 & .059 \ \ t = 70 & .117 & .059 & .059 & .059 & .177 & .059 & .117 & .059 & .117 & .059 & .059 & .059 \ end {matrix} 

 

 

Although dissemination take more time to stabilize, the stationary measure is the same! It also coincides with our solution  , here divided by   1 USD . The pages where     is large are the most "popular" or most "popular". In the quest to rank web pages in order of importance is even an argument for using the measure   as an indicator.

 

The model of the random surfer may seem surprising, but in the absence of more precise information, the use of probabilistic considerations is often very helpful!

The Transition Act

How to formalize the distribution shown above? Suppose that at time   our random surfer is on page P_j   with probability    . The probability of P_j from   and follow the link   then      . The probability of arriving at a time   on     page is

 

P'_i  : = \ sum_ {j \ to i} \ frac {1} {\} p_j ell_j. 

 

Given the initial distribution  , the transition law defines the following distribution    . Thus we get a line   from   line in our examples. (In probability theory this is called a Markov chain.) The stationary measure is characterized by the equilibrium equation  , which is precisely our equation departure.

Watch out for black holes

What happens there when our graph contains a page (or group of pages) without end? For illustration, here is our graph plus a new page     without end:

 

The interpretation as a random walk solves the equation without any calculation: page     absorbs all likelihood because our random surfer will fall sooner or later on this page, where it remains for rest of his life. So the solution is   \ begin {matrix} & & & P_1 & P_2 & P_3 P_4 & & & D_5 P_6 P_7 & & & P_8 P_9 & P_ {10} & P_ {11} & P_ {12} & P_ {13 } & \ \ m & = & (& 0, & 0, & 0, & 0, & 0, & 0, & 0, & 0, & 0, & 0, & 0, & 0, & 1 &). \ End {matrix}   Our model is not yet satisfactory.

The model used by Google PageRank

To escape the black hole, Google uses a more refined model:

• with a fixed probability   surfer abandons its current page P_j   and again on the   web pages, chosen so equiprobable;
• Otherwise, with probability  , the surfer follows a link from page P_j  , chosen so as equiprobable. (This is the usual random walk).

This trick of "teleportation" avoids being trapped by a page without issue, and guaranteed to arrive anywhere in the graph regardless of connectivity issues.

 

In this model the transition is given by

 

P'_i  : = \ frac {c} {n} + \ sum_ {j \ to i} \ frac {1-c} {\} p_j ell_j.   The first term   comes from the teleportation, the second term is the random walk before. The balance measure thus satisfies the equation

  M_i = \ frac {c} {n} + \ sum_ {j \ to i} \ frac {1-c} {m_j} ell_j.  

The parameter   is still to be calibrated. For   we get the previous mode.

 

In general, choose the constant   non-zero but close to zero. For example, the choice   0.15 is about $ 6 U.S. dollars follow links on average, which seems a realistic description.

 

To conclude the analysis of our example, here is the random walk starting from the page    :   \ begin {matrix} & P_1 & P_2 & P_3 & P_4 D_5 & & & P_6 P_7 P_8 & & & P_ {10 P_9 } & P_ {11} & P_ {12} \ \ t = 0 & 1.00 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 & .000 &. 000 \ \ t = 1 & .013 & .225 & .225 & .225 & .225 & .013 & .013 & .013 & .013 & .013 & .013 & .013 \ \ .305 & t = 2 & .111 & .111 & .111 & .028 & .076 & .087 & .076 & .034 & .020 & .020 & .020 \ \ t = 3 & .186 & .124 & .124 & .124 & .158 & .021 & .085 & .021 & .071 & .028 & .028 & .028 \ \ t = 4 & .180 & .105 & .105 & .105 & .140 & .057 & .075 & .057 & .057 & .040 & .040 & .040 \ \ t = 5 & .171 & .095 & .095 & .095 & .126 & .052 & .101 & .052 & .087 & .042 & .042 & .042 \ \ \ dots \ \ t = 29 & .120 & .066 & .066 & .066 & .150 & .055 & .102 & .055 & .120 & .066 & .066 &. 066 \ \ t = 30 & .120 & .066 & .066 & .066 & .150 & .055 & .102 & .055 & .120 & .066 & .066 & .066 \ end {matrix}   The stationary measure is quickly reached, and the page       before = 0.15 m_5 pages     and   with         0.12.

The fixed point theorem

To develop a promising model we used heuristic arguments and experimental designs. Now fix this model and ask it on a solid theoretical foundation. Our calculations result in fact in our example miniature, but is this always the case? The following beautiful result answers in full generality:

 

Fixed point theorem. - Consider a finite graph and fix any parameter   such that  . Then:

• Equation admits a unique solution that       1. In this solution       are strictly positive.
• For any initial probability distribution on the graph, the diffusion process converges to the unique stationary measure  .
• Convergence is at least as fast as that of the geometric     to $ 0.

Emphasize the importance of each of these three points. The first simply ensures the existence and uniqueness of a solution to our problem. Better yet, a solution not only exists but the second point tells us how to calculate: an iterative algorithm. Here the independence of the concept ensures a certain numerical stability: the calculations with point numbers, rounding errors are often unavoidable, but fortunately by such disturbances do not affect the final result. Finally, the third point guarantees that the convergence rate is high enough, which is crucial for any application size. For its Google ranking processes several billion web pages. This Herculean task is only possible with the iterative algorithm, and the theorem guarantees its effectiveness regardless of the graph.

 

This theorem is as elegant and useful. The idea of proof is surprisingly simple: we show that the transition law defines a map     is contracting over  . The result thus follows from the fixed point theorem of Banach.

Conclusion

To be useful, a search engine should not only list the results of a query, but rank them in order of importance. However, estimating the relevance of web pages is a profound challenge to modeling.

 

As a first approximation Google scans the graph formed by links between web pages. Interpreting a link   as "voting" on page P_j   for     page, the PageRank model defines a measure of "popularity."

 

The fixed point theorem ensures that this equation admits a unique solution, and justifies the iterative algorithm to approach. It is easy to implement on a computer and quite effective for graphs of size.

 

Armed with these mathematical tools and a clever business strategy, Google makes billions of dollars. He had to think!

Read more »

Posted in: Blogger Essential,Google,learning,Ranking,SEO tool,Web Informations