How does the Google search engine understand the texts?

For some years, Google has developed an algorithm capable of understanding the texts. For this reason, a fundamental aspect of the specialization of an SEO specialist or a copywriter is writing and readability. The text must satisfy the needs of the users, also increasing the position in the SERP.

Are we really sure that Google understands the text?

We know that Google understands the text, but within certain limits. The most important thing is that Google is able to correctly match what the user types in the search bar, with the best search result. To do this, Google cannot trust only the information that the user makes available, namely the meta data.

Furthermore, we also know that it is possible to classify a sentence that is not used in the text (although it is still good practice to identify and use one or more specific key phrases). So, Google does something to read and evaluate the text contained on a page of your website.

You might also like: SEO strategy voice search and the success of Personal Assistant

What is the current status?

The method used by Google to understand the texts is unknown. That is, information is not available in a simple and free way. We also know, judging by the results of the research, that there is still a lot of work to be done to achieve an optimal result. But there are some clues here and there from which we can draw interesting conclusions.

For example, we know that Google has made great strides in understanding the context. We also know that Google tries to determine how words and concepts are related to each other.

Word Embeddings

An interesting technique that Google has filed patents and worked on is called Word Embedding, "Meetings of words" or "Related Words". Flying over the details, the goal is basically to find out which words are closely related to other words. Practically: a software takes a certain amount of text, analyzes them and determines which words tend to be together more frequently, and turns each word into a series of numbers. In this way it is possible to represent words as a point in space in a diagram, like a scatter plot.

The diagram thus obtained shows which words are related and how. More precisely, it shows the distance between words, representing a kind of galaxy made up of words.

So, for example, a word like "keywords" would be much closer to "copywriting" instead of "kitchen utensils".

This procedure can be applied to both words and sentences and / or paragraphs. The larger the data set that feeds the program, the better the algorithm will be able to categorize and understand words, understand how they are used and what they mean.

Practically, Google has a database that includes the entire network. Thus, with a set of information of this size, it is possible to create reliable models that can evaluate the value of the text and the context.

Related entities

From the correlation of words, we take a small step towards the concept of related entities. If we try to do a search, we can see what the related entities are. By typing "types of pasta", at the top of the SERP you should see "I Formati della Pasta". These varieties of pasta should also be sub-categorized. There are many similar SERPs that reflect the way words and concepts relate to each other.

The patent relating to the entities that Google has filed actually mentions the database of indexes relating to the entities. This is a database in which concepts or entities, such as pasta, are stored. These entities also have characteristics. Lasagna, for example, is a pasta. It is also made of pasta. And it's a food. Now, analyzing the characteristics of the entities, they can be grouped and classified in all kinds of different ways. This allows Google to better understand how words are related and, therefore, to better understand the context.

Practical conclusions

If Google understands the context of the page, it will certainly evaluate it and judge its content. The better the correspondence with the notion of Google context, the better will be its chances of being in evidence. It will be necessary to express the concepts exhaustively. In a broader way, expressing also the related concepts.
Simple texts, clearly expressing the relationships between the various concepts, help your readers to understand better, and also help Google.

Difficult, inconsistent and poorly structured writing is more difficult to understand for both humans and Google. You must help the search engine understand your texts by focusing on:

• Good readability, that is to make your text easier to read as possible without compromising your message;
• a good structure, that is adding subtitles and clear transitions;
• Good context, that is, adding clear explanations that show how what you are saying refers to what is already known about a topic

A good result will help your readers and Google understand your text, and therefore all the goals you set for yourself.

Especially because Google seems to be trying to create a model that mimics the way we humans process language and information.

And this makes us think that Google still uses keywords, to match your page to a query.