BERT, Google and natural language processing
The root of these changes can be found in the latest major update to Google’s search algorithm: BERT, a solution that integrates new parameters based on artificial intelligence. According to its creators, this is the biggest change in SEO in the past five years.
BERT is able to parse search expressions in both directions, that is, it takes into account the words located to the right and left of each keyword and links all the terms of the query to each other. It also has a system of meaning assignment It helps you understand words like link, pronouns, and prepositions. The result of all this is that BERT can understand what is the intention of a natural language query, that is, he can understand the way humans express themselves. This is particularly relevant in the case of voice searches.
BERT is not intended to completely replace Google’s current algorithm (RankBrain), but is a kind of “add-on” to better understand user queries. This means that many of the SEO techniques that we’ve been using in recent years, such as focusing content on a main keyword, its associated terms, and long-tail variants, will continue to hold true for years to come. But things won’t be as simple as they used to be.
Key Concepts of Google’s Natural Language Processing
A key concept that must be understood natural language processing or NLP is that of entities.
An entity is a singular, unique, well-defined and distinguishable thing or concept, for example people, places, objects, artistic elements, events, numbers or abstract concepts.
Entities affect search results primarily based on the relationship between them. This relationship is judged primarily on the basis of what is called co-occurrence. The more two entities are mentioned together, and the more authoritative the result that mentions them, the stronger the relationship between them.
These recommendations will help you get Google to see your brand or business as an entity:
Have your own website and a verified social media presence.
Have an account on Google My Business.
Have an entry on Wikipedia.
Have organization schema data markup with SameAs attribute referenced to your social networks.
Have content that talks about your products and services and that reinforces your website’s association with the semantic category Google assigns to it.
Have inbound links from authoritative pages that are thematically relevant.
Here is access to Google’s natural language tool.
Another very important concept is salience, a metric used to determine how well an entity stands out in text. Essentially, high salience means that for human reading this entity is more relevant than others that are represented on the same page.
The number of references to an entity is a factor taken into account in determining its salience. Google takes into account the frequency with which two entities are mentioned together and the importance of one entity in its category.
Finally, the importance of structured data in this new SEO model. Structured data is a kind of “tag” that tells Google what type of content we are talking about and allows it to better organize information. There are many types of structured data, from recipes to job postings. Using this data not only helps Google to understand the content of a page, but also to show more complex results on search results pages.
Other important data that Google offers with NLP are:
- The level of feeling.
- The agreement between the words of the text, that is to say the syntax.
- the categories possible.
7 Ways To Optimize SEO To Improve Your Content With Natural Language Processing
1) don’t get obsessed with keywords
The goal of Google’s new algorithm is to be increasingly able to interpret the natural language people use. What is appreciated above all is the user experience, and users like to find naturally written texts.
As a result, the practice of filling in texts with exact keywords is more obsolete than ever. Instead, try to use plain language and be as helpful as possible in answering user questions.
2) Increase the salience of entities
With automatic natural language processing, you have to stop thinking about keywords to ask yourself entities we refer in a text and how we are interested in positioning them.
Therefore, we have to think about highlighting the entities most relevant to us, connecting them with other relevant entities and increasing their importance in the text and the links directed there. Instead, we should give less importance to entities which are less relevant or which are not recognized by Google as such.
3) Connect the questions and the answers
The key to optimizing content for PLN is to have sentences with simple structures, especially when it comes to answering questions. The format must be [Entidad] this is [Respuesta].
In addition, giving direct answers to questions increases the chances of appearing in the google zero position (result highlighted before the first position).
4) Identify the key facts
Google hopes to meet some data in a certain format to answer a question. For example, if we are talking about the height of a mountain, Google will expect to find a certain number of meters. Taking this into account when preparing our content will help Google interpret our page correctly.
5) be clear and direct
Some content creators are tempted to “beat around the bush” to increase the number of words in a text, mistakenly believing that Google favors extension first.
In fact, what interests us is give a clear and concise answer to the user’s main question, as this increases both understanding on the part of Google and the likelihood that it will place us in position zero.
If you are dealing with a complex subject, it is always a good idea to provide details and explanation, but always focusing on providing useful information to the user at all times.
6) Take advantage of the structure to convey meaning
The way we structure a text at the global level is very important for natural language processing. To better position yourself in Google, you need to create very clear and well-organized texts. We should also keep in mind that the place where an entity or data appears helps us to convey its relative importance. Here are some examples :
Inverted pyramid: Google assumes that the texts are organized in an inverted pyramid structure, that is, the most important information is placed at the beginning. Therefore, the higher a piece of content is on the page, the greater the relative importance it will be given.
Header hierarchy: Organizing the text using headings (H1, H2, H3 …) is a good SEO practice which is still very valid. For example, if an entity appears in an H2, Google will understand that it is more important than the one that appears in an H4, and in turn it will be more important than the one that appears in the body of the text.
Proximity: the closer two entities are to each other, the more they will be linked in the eyes of Google.
Lists: HTML lists establish a relationship between the various elements that compose them, which are understood as part of a whole.
7) Provide correct and complete information
If a response to a the question is not correct or is not up to date, Google is less likely to select it. Therefore, it is important periodically review the content to ensure that all data remains correct and that it is optimized according to PLN criteria.
We must also ask ourselves if our articles answer all the topics and questions that a user may ask themselves during a research. While it is important to be clear and to the point, it is also important to cover the full extent necessary for Google to understand the content and users to be well informed. To do this, we can expand the information with related entities in the background.