₦airaland Forum

Machine learning beginning to changing the way we live. And even if you think you have not encountered this technology yet, you should know that it is already everywhere. In fact, you have found it in the recommendations of Netflix and also in Google.

The concept of machine learning is simple. But what if we investigate further its functions, characteristics and advantages?

One of the news related to this topic came from Apple. And most likely, you've heard of her.

The brand bought the machine learning firm Perceptio Inc , in an attempt to incorporate advanced artificial intelligence. He perceived the development of facial recognition algorithms that were running on users' devices without resorting to the cloud. The idea is that it could classify images on smartphones and reduce data overload that, in general, requires conventional methods.

BERT: Google Algorithm Update, Everything you should know about it

Pay attention to this data. According to a Statista report 'What part of your IT budget for 2017 is earmarked for machine learning?', 26% of companies recognize that more than 15% of their IT budget has been invested in the development of machine learning solutions.

Machine learning is beginning to change the way we live, and it is time for us to understand what it is and why it is important.

Delving into machine learning


Before entering machine learning, it is convenient to clarify a frequent doubt. Machine learning and artificial intelligence are not the same . In fact, machine learning is presented as an approach to the broader field of artificial intelligence .

Artificial intelligence is a simulation of intelligence in computers. As such, it is something that specialists have been working on for decades and encompasses a wide range of approaches.

Machine learning, meanwhile, is a form of artificial intelligence in which the computer learns itself to complete a task.

There are different ways to make the machines learn by themselves. At the core of deep learning are " artificial neural networks ." And the thought behind these neural networks forms what you will see next:

As you know, the best learning system is the human brain. That is why machine learning emerges as an attempt to imitate the way in which the brain learns. Neural networks become that attempt.

That is, they are mathematical representations of the way in which the human brain operates.

Types of machine learning
1.- Supervised learning
This type of machine learning works on previously tagged databases. For example, a computer has the ability to differentiate a photograph of a motorcycle from another and a bicycle.

For this, it is recommended that the labels be placed by a person. This guarantees the effectiveness and quality of the exercise.

The idea of ​​this type of machine learning is that computers train their "intelligence " from past examples. In this way, they can then perform all the calculations that are proposed so that a human being does not intervene and work can be expedited.

As most recognized examples you can see:

Voice recognition.
Spam detection
Also the recognition of writing.
2.- Unsupervised learning
This second category works as follows. The algorithm dispenses with any label. That is to say that it does not have the help of any previous indication.

However, it is awarded a large amount of data with the specifications that each object has. Thus, it is easier to detect what it is based on the information that has been previously stored.

3.- Reinforcement learning
The last type of machine learning has as a reinforcement engine. That is, the machine has the ability to learn based on different tests and errors given in different contexts.

Regardless of whether you know the results from the start, the machine still does not know what are the best decisions to achieve the correct results. In this case, the algorithm, by itself, associates the success patterns. And after repeating them constantly get the exact and infallible results.

The clearest example may be the navigation of an automatic vehicle.

Companies that already incorporate machine learning
1.- Facebook


One of the places where you will also have encountered machine learning is in the social network par excellence. Facebook uses its facial recognition algorithm to take your job off when tagging your friends in photographs.

It is an algorithm that memorizes certain patterns that define the faces of those people you have previously tagged to, thus, identify them automatically later.

2.- Pinterest
However, Facebook is not the only platform that has incorporated these technological advances. Even if you don't expect it, Pinterest also uses machine learning to boost engagement with users. But how?

The platform also employs an algorithm that learns from user activity. Thus, it manages to capture each one's favorite content based on their interests, tastes and habits.

3.- Paypal
Surely you already know Paypal, the payment platform that becomes an electronic wallet. And as such, you need to have a high level of security and trust from your users.

To remain one of the safest platforms, PayPal uses machine learning to identify and address fraud . Analyze large amounts of customer data and then segment them according to their activity.

4.- Cortana and Siri


There is no doubt that it has been seen that iPhone has caused several changes in the last decade. But he has not been the only one.

You see, the voice recognition systems of virtual assistants like Cortana and Siri also make use of machine learning. Although they use other technologies such as neural networks and natural language processing. In this way they manage to simulate that almost human conversation.

But how are these advances generated? How do you get machine learning to train?

In this case you need to work to recognize human language. And for this, several techniques with different levels of complexity are used. In the case of Siri, templates that resemble human behavior are used. That's why he has the ability to understand jokes.

5.- Google
There is no doubt that Google is one of the largest portals capable of collecting all types of user data. And as such, do you think I could do without machine learning?

The answer is not only a resounding no. But also, you should know that the Internet giant uses machine learning algorithms in all its services, starting with web search optimization. In fact, it studies some neural network that manages to detect the intention of each user.

But beyond Google Search , services like Gmail also make use of machine learning. Especially for your intelligent response services.

On the other hand, Google Maps has the ability to detect street numbers and names from Street View images.

6.- Netflix


Netflix, among other things, has managed to become a solid brand for its system of recommendation and customization of series and movies.

Thanks to machine learning, the platform applies the technology to:

The promotion of the content.
Your pricing model.
To programmatic marketing.
The delivery of the content.
7.- Uber


A few years ago nobody imagined that when ordering a taxi (using a mobile application), it would indicate the arrival time and the estimated price of your entire trip. Not much less, that would have a personalized food service to liven up the tour.

Uber does it thanks to machine learning, since it carries out a very specific study of the travel and order data of each user.

8.- Spotify


The weekly Spotify discovery is one of the examples of machine learning. The platform uses several recommendation models depending on the activity of each user. But it also does it thanks to a more exhaustive analysis of texts and audios that manage to group the likes, always similar, of others, to recommend them.

9.- Amazon
Amazon is another company that uses machine learning to improve its services.

In its search service, the product catalog is structured according to customer preferences.
In its recommendation system, the product catalog is organized according to interests.
And finally, it also faces the payment fraud prevention system.
However, the value of machine learning depends on the volume of communication of each organization. In addition, it also varies based on users and the possible channels through which these services can be implemented.

how to make a website optimized for seo
Now that you know what SEO is and what are the main factors that Google takes into account when positioning a website, you need to learn what you have to do so that your page has opportunities to position up in the SERPs.

In this chapter, we are going to talk about how to optimize the main positioning factors as well as the main SEO problems that arise when optimizing a website and its possible solutions.

We will divide the themes of this chapter into 4 large blocks:


1. Accessibility
The first step in optimizing the SEO of a website is to allow access to search engines to our content. That is, you have to check if the website is visible in the eyes of search engines and, above all, how they are viewing the page.

For various reasons that we will explain later, it may be the case that search engines cannot correctly read a website, an indispensable requirement for positioning.

Aspects to consider for good accessibility

File robots txt
Meta tag robots
HTTP status codes
Sitemap
Web structure
JavaScript and CSS
Web speed
File robots txt
The robots.txt file is used to prevent search engines from accessing and indexing certain parts of a website. It is very useful to prevent Google from displaying in the search results the pages we don't want. For example in WordPress, so that they do not access the administrator files, the robots.txt file would be like this:

Example

User-agent: *

Disallow: / wp-admin

EYE: You must be very careful not to block the access of search engines to your entire website without realizing it as in this example:

Example

User agent: *

Disallow: /

We must verify that the robots.txt file is not blocking any important part of our website. We can do this by visiting the URL www.example.com/robots.txt, or through Google Webmaster Tools in “Tracking”> “robots.txt tester”

The robots.txt file can also be used to indicate where our sitemap is located by adding on the last line of the document.

Therefore, an example of a full robots.txt for WordPress would look like this:

Example

User-agent: *

Disallow: / wp-admin

Sitemap: http: //www.example.com/sitemap.xml

If you want to go deeper into this file, we recommend you visit the web with the information about the standard

Meta tag Robot
The meta tag "robots" is used to tell search engine robots whether or not they can index the page and whether they should follow the links it contains.

When analyzing a page you should check if there is a meta tag that by mistake is blocking access to these robots. This is an example of how these tags would look in HTML code:

Example

<meta name = ”robots” content = ”noindex, nofollow”>

On the other hand, meta tags are very useful to prevent Google from indexing pages that do not interest you, such as pages or filters, but follow the links to continue tracking our website. In this case, the label would look like this:

Example

<meta name = ”robots” content = ”noindex, follow”>

We can check the meta tags by right-clicking on the page and selecting "see page source code".

Or if we want to go a little further, with the Screaming Frog tool we can see at a glance which pages of the entire web have said tag implemented. You can see it in the “Directives” tab and in the “Meta Robots 1” field. Once you have located all the pages with these tags you just have to delete them.

HTTP status codes
In the event that any URL returns a status code (404, 502, etc.), users and search engines will not be able to access that page. To identify these URLs we recommend you also use Screaming Frog, because it quickly shows the status of all the URLs on your page.

IDEA: Every time you do a new search in Screaming Frog, export the result in a CSV. So you can gather them all in the same Excel later.

Sitemap
The sitemap is an XML file that contains a list of the pages of the site along with some additional information, such as how often the page changes its contents, when was its last update, etc.

A small excerpt from a sitemap would be:

Example

<url>

<loc> http://www.example.com </loc>

<changefreq> daily </changefreq>

<priority> 1.0 </priority>

</url>

Important points to check regarding the Sitemap, which:

Follow the protocols, otherwise, Google will not process it properly
Get uploaded to Google Webmaster Tools
Be updated When you update your website, make sure you have all the new pages on your sitemap
All pages on the sitemap are being indexed by Google
In case the website does not have any sitemap we must create one, following three steps:

It generates an Excel with all the pages that we want to be indexed, for this we will use the same Excel that we created when doing the search of the HTTP response codes to Create the sitemap. For this, we recommend the Sitemap Generators tool (simple and very complete)
Compare the pages that are in your excel and those in the sitemap and remove from the excel those that we do not want to be indexed.
Upload the sitemap through Google Webmaster Tools
Web structure
If the structure of a website is too deep Google will find it more difficult to reach all pages. So it is recommended that the structure is not more than 3 levels deep (not counting the home) since the Google robot has a limited time to track a web, and the more levels you have to go through the less time you will have left to access to the deepest pages

That is why it is always better to create a web structure horizontally and not vertically

How To Write A Blog: 13 fundamental tips for complete content creation

Our advice is to make an outline of the whole web in which you can easily see the levels you have, from home to the deepest page and be able to calculate how many clicks it takes to reach it.

Locate at what level each page is and if you have links pointing to it using Screaming Frog again.

JavaScript and CSS
Although in recent years Google has become smarter when reading this type of technology we must be careful because JavaScript can hide part of our content and CSS can mess it up by showing it in another order that Google sees it.

There are two methods to know how Google reads a page:

Plugins
"Cache:" command
Plugins

Plugins such as Web Developer or Disable-HTML help us see how the web “crawls” a search engine. To do this you have to open one of these tools and disable JavaScript. We do this because all the drop-down menus, links, and texts must be able to be read by Google.

Then we deactivate the CSS since we want to see the actual order of the content and the CSS can change this completely.

"Cache:" command

Another way to know how Google sees a website is through the "cache:"

Enter "cache: www.myexample.com" in the search engine and click on "Text-only version". Google will show you a photo where you can know how to read a website and when was the last time you accessed it.

Of course, for the "cache:" command to work correctly our pages must be previously indexed in the Google indexes.

Once Google indexes a page for the first time, it determines how often you will visit it again for updates. This will depend on the authority and relevance of the domain to which that page belongs and the frequency with which it is updated.

Whether through a plugin or the "cache:" command, make sure you meet the following points:

You can see all the menu links.
All web links are clickable.
There is no text that is not visible with CSS and Javascript enabled.
The most important links are at the top.
Loading speed
The Google robot has a limited time when browsing our page, the less it takes each page to load more pages will reach.

You should also keep in mind that a very slow page load can cause your bounce rate to skyrocket, so it becomes a vital factor not only for positioning but also for a good user experience.

To see the speed of loading of your website we recommend Google Page Speed, there you can check what are the problems that slow down your site in addition to finding the tips that Google offers to tackle them. Focus on those with high and medium priority.

2. Indexability
Once the Google robot has accessed a page the next step is to index it, these pages will be included in an index where they are sorted according to their content, their authority, and their relevance to make it easier and faster for Google to access them.

How to check if Google has indexed my website correctly?
The first thing you have to do to know if Google has indexed your website correctly is to perform a search with the “site:” command, so Google will give you the approximate number of pages on our website that it has indexed:

If you have linked Google Webmaster Tools on your website you can also check the actual number of indexed pages by going to Google Index> Indexing status

Knowing (more or less) the exact number of pages your website has, this data will help you compare the number of pages that Google has indexed with the number of real pages on your website. Three scenarios can happen:

The number in both cases is very similar. It means that everything is in order.
The number that appears in Google search is smaller, which means that Google is not indexing many of the pages. This happens because you cannot access all the pages on the web. To solve this, review the accessibility part of this chapter.
The number that appears in the Google search is greater, which means that your website has a duplicate content problem. Surely the reason why there are more indexed pages than there really are on your website is that you have duplicate content or that Google is indexing pages that you don't want to be indexed.
Duplicate content
Having duplicate content means that for several URLs we have the same content. This is a very common problem, which is often involuntary and can also have negative effects on Google positioning.

These are the main reasons for duplicate content:

“Canonicalization” of the page
URL parameters
Pagination
It is the most common reason for duplicate content and occurs when your homepage has more than one URL:

Example

example.com

www.example.com

example.com/index.html

www.example.com/index.html

Each of the above direct to the same page with the same content, if Google is not told which one is correct, it will not know which one it has to position and it may just position the unwanted version.

Solution. There are 3 options:

Redirect to the server to make sure there is only one page that is displayed to users.
Define which subdomain we want to be the main one (“www” or “no-www”) in Google Webmaster Tools.
Add a “rel = canonical” label in each version that points to the ones considered correct.
URL parameters
There are many types of parameters, especially in e-commerce: product filters (color, size, score, etc.), ordering (lower price, relevance, higher price, grid, etc.) and user sessions. The problem is that many of these parameters do not change the content of the page and that generates many URLs for the same content.

www.example.com/boligraphs?color=negro&pre-desde=5&precionahasta=10

In this example, we find three parameters: color, minimum price, and maximum price.

15 Tips To Generate Content Ideas For Your Blog

Solution

Add a “rel = canonical” tag to the original page, so you will avoid any confusion on the part of Google with the original page.

Another possible solution is to indicate through Google Webmaster Tools> Tracking> URL parameters which parameters Google should ignore when indexing the pages of a web.

Pagination
When an article, product list or pages of labels and categories have more than one page, duplicate content problems may appear even if the pages have different content, because they are all focused on the same topic. This is a huge problem in the e-commerce pages where there are hundreds of articles in the same category.

Solution

Currently the rel = next and rel = prev tags allow search engines to know which pages belong to the same category/publication and thus it is possible to focus all the positioning potential on the first page.

How to use the NEXT and PREV parameters

1. Add the rel = next tag in the part of the code to the first page:

link rel = ”next” href = ”http://www.example.com/page-2.html” />
2. Add on all pages except the first and last tags rel = next and rel = prev

link rel = ”prev” href = ”http://www.example.com/page-1.html” />
link rel = ”next” href = ”http://www.example.com/page-3.html” />
3. Add the rel = prev tag to the last page

link rel = ”prev” href = ”http://www.example.com/page-4.html” />
Another solution is to look for the pagination parameter in the URL and enter it in Google Webmaster Tools so that it is not indexed.

Cannibalization
Keyword cannibalization occurs when there are several pages on a website that compete for the same keywords. This confuses the search engine by not knowing which one is the most relevant for that keyword.

This problem is very common in e-commerce, because having several versions of the same product "attack" with all of them the same keywords. For example, if a book is sold in a softcover, hardcover and digital version, there will be 3 pages with virtually the same content.

Solution

Create a main page of the product, from where you access the pages of the different formats, in which we will include a canonical label that points to said main page. The optimum will always be to center each keyword on a single page to avoid any cannibalization problem.

3. Content
Since in recent years it has become quite clear that content is king for Google. Let's offer him a good throne then.

The content is the most important part of a website and as much as it is well optimized at SEO level, if it is not relevant with respect to the searches that users perform it will never appear in the top positions.

To make a good analysis of the content of our website you have a few tools at your disposal, but in the end, the most useful thing is to use the page with Javascript and CSS deactivated as explained above. This way you will see what content Google is actually reading and in what order it is arranged.

When analyzing the content of the pages you must ask yourself several questions that will guide you through the process:

Does the page have enough content? There is no standard measure of how much is "sufficient", but it should at least contain 300 words.
Is the content relevant? It must be useful for the reader, just ask yourself if you would read that. Be sincere.
Do you have important keywords in the first paragraphs? In addition to these, we must use related terms because Google is very effective in relating terms.
A page will never position for something that it does not contain

Do you have keyword stuffing? If the content of the page "sins" of excess keywords to Google will not do any grace. There is no exact number that defines a perfect keyword density, but Google advises to be as natural as possible.
Do you have misspellings?
Is it easy to read? If we don't find the reading tedious, it will be fine. The paragraphs should not be too long, the letter should not be too small and it is recommended that there be images or videos that reinforce the text. Remember to always think about what audience you write for.
Can Google read the text of the page? We have to prevent the text from being inside Flash, images or Javascript. We will verify this by looking at the text-only version of our page, using the Google command cache: www. example.com and selecting this version.
Is the content well distributed? It has its corresponding H1, H2 and so on labels, the images are well mockup, etc.
Is it linkable? If we do not provide the user how to share it, it is very likely that he will not do so. It includes social media sharing buttons in visible places on the page that do not hinder the display of content, be it a video, a photo or text.
Is actual? The more up-to-date your content is, the higher the frequency of Google tracking on your website and the better the user experience.
Advice

You can create an excel with all the pages, their texts and the keywords you want to appear on them, this way it will be easier for you to see where you should reduce or increase the number of keywords on each page.

4. Meta tags
The meta tags or meta tags are used to convey information to search engines what the page is about when they have to sort and show your results. These are the most important labels that we must take into account:

Title
The title tag is the most important element in meta tags. It is the first thing that appears in the results on Google.

When optimizing the title, keep in mind that:

The tag must be in the <head> </head> section of the code.
Each page must have a unique title.
It should not exceed 70 characters, if it will not appear cut.
It must be descriptive regarding the content of the page.
It must contain the keyword for which we are optimizing the page.
We must never abuse the keywords in the title, this will make users distrust and Google thinks that we are trying to deceive you.

Another aspect to take into account is where to put the “brand”, that is: the name of the web, usually it is usually put at the end to give more importance to the keywords, separating these from the name of the web with a script or a vertical bar

Meta-description
Although it is not a critical factor in the positioning of a website, it significantly affects the click-through rate in search results.

For the meta-description, we will follow the same principles as with the title, only that its length should not exceed 155 characters. For both titles and meta-descriptions, we must avoid duplication, this can be verified in Google Webmaster Tools> Search aspect> HTML improvements.

Meta Keywords
At the time the meta keywords were a very important positioning factor, but Google discovered how easy it is to manipulate the search results so it eliminated it as a positioning factor.

Tags H1, H2, H3 ...
The labels H1, H2, etc. They are very important to have a good information structure and a good user experience, since they define the hierarchy of content, something that will improve SEO. We must give importance to H1 because it is normally in the highest part of the content and the higher a keyword is, the more importance Google will give it.

“Alt” tag in the image
The "alt" tag in the images is added directly to the image code itself.

Example

<img src = ”http://www.example.com/example.jpg” alt = ”keyword seo” />

This tag has to be descriptive with respect to the image and content of that image, since it is what Google reads when tracking it and one of the factors that it uses to position it in Google Images.

conclusion
You already know how to make a page optimized for SEO and that there are many factors to optimize if you want to appear in the best positions of the search results. Now you will surely ask yourself, what are the keywords that best position my website?

We don't know exactly what those keywords are, but we can help you find them , find out in the next lesson!