Thread by @GiordMarco96

What is Natural Language Processing (#NLP) and why it is so important for #SEO?

There are some clear benefits but first we need some clear explanations.

This thread shows you why this subject can improve your results and the role it plays in modern search engines. 🧵

NLP is a branch of Machine Learning that enables machines to understand and process natural language, i.e. what we humans say.

Our language is complex, ambiguous and sometimes misleading, it's a tough job for a machine!

More specifically, it's a subset of Computer Science, Linguistics and Artificial Intelligence.

Computational Linguistics is another subject closer to Linguistics than it is to Engineering. You can think of NLP as the other way around.

Source: direct.mit.edu/coli/article/4…

Modern search engines like Google are powered by such technologies and understanding some key NLP concepts can increase your chance of ranking.

In the last few years we have seen the rise of new methodologies and perspectives that totally differ from the past.

If you read some Google patents, you can quickly notice how some topics are frequent and why people emphasize the role of topical clusters.

On a higher level you would need to treat your website as a big Knowledge Graph. Cover topics extensively and achieve authority.

Google is able to process what you write and can understand its sentiment.

While the sentiment of a text is not a ranking factor, some may even argue that it can play a role in what you see on SERPs.

This is NLU, not NLP but still useful to know.

Keyword density is useless. Instead of counting keywords focus on syntax and entities in your text. Google can detect named entities in a text and understand their role in a sentences too.

That's why you should be specific and accurate while you write.

My article is longer, why is it not first?

There are many answers, one could be that your sentence structure is not so clear or you are diluting your content.

If you don't go straight to the point, don't expect people to read your blog posts. Google knows that very well too.

The list of NLP tasks is long, you should mainly care about:

- POS Tagging
- Stemming/Lemmatization
- Named Entity Recognition (NER)
- Word Sense Disambiguation
- Entity Linking
- Semantic role labelling
- Question answering

I will cover them in future threads. >>>

>>> Just recall for now that there are different tasks that you can execute with some NLP knowledge.

You can evaluate your own text and look at competitors with a more critical eye.

If you are just starting you'll also encounter stop-word removal and tokenization.

Machines don't actually understand what you say. They eat numbers and perform some operations to process human language.

State-of-the-art models are pretty good at it and the best is yet to come.

Google applies BERT models to snippets to understand if they're good enough for the user. To win them, you should understand what HTML/structured data components your competitors are using.

Bring more facts and have a crystal clear language. Zero fluff.

Be careful tho, we are not optimizing for BERT, it doesn't make any sense to say such a thing.

We are just improving our syntax to provide better and clearer answers to the user's needs.

It's not a game to trick search engines, it's adding more value.

One of the most powerful concepts you may have heard is "semantic triple".

A triple can be described as the minimum unity to encode data into the semantic web.

Triples are made up by a "subject–predicate–object" structure.

👉 John was born in 1945.

de.marklogic.com/blog/making-ne…

Entities are then connected via relationships. You can express them via semantic triples.

You are starting to connect the dots. Information in the semantic web is ruled by such triples.

Knowing the main entities you have to cover in a page is a method of page optimization.

The entire goal of learning more #NLP is to create pages that are easily understandable by machines.

Build a "content network" in order to be recognized by Google as an authority on a given topic.

This doesn't substitute your value proposition or push you to fake content.

Several triples connected via relationships can form a Knowledge Graph.

It is a structure to represent knowledge that understands machine language in the form of standards.

Therefore, order is not enough, there should be some ratio behind a graph.

Speaking of which, Knowledge Graphs are organized according to ontologies.

They define what exists in your domain and their properties. This will be the topic for a future thread.

Google introduced the Knowledge Graph in 2012. Nowadays it's one of the hottest #SEO topics and for good reasons.

Having a website with proper connections between entities is the key.

Read more: holisticseo.digital/theoretical-se…

LSI keywords don't work at all. It's the usual #SEO myth created by people with no clear understanding of search engines.

More here: seobythesea.com/2021/12/what-a…

Did you know that Google doesn't scrape questions? They generate them!

You can do the same to add value to your content strategy.

Source: patents.google.com/patent/US92137…

RankBrain is an algorithm used to understand search intent. It is one of the most important factors affecting ranking.

That said, you should have understood that focusing on single words or keywords is super useless.

Intent, entities and order are your new friends.

Why did I add "order"? Because having clear sentences is mandatory and you need some degree of methodology to make it work.

Being methodical is a huge help for such technical topics.

All of this fits into the Semantic Web.

You can think of it as the proposal of structuring and tagging web pages for machine understanding. This is done via a set of standards.

And that's why we talked about Knowledge Graphs before.

And that's one more reason to start creating topical maps with tools like Xmind/MindMeister/Miro.

You can think of your content strategy as a way to build an imaginary Knowledge Graph and fit it into your website.

Clustering keywords is an enormous help for this.

You can be more proactive and combine what you know with #Python libraries.

Some examples include:
- NER on competitors page
- SERP analysis
- N-grams analysis
- Semantic clustering
- Smarter internal linking
- Knowledge Graphs

...and much more

Some indirect potential benefits of NLP are:

- More understanding of Google patents
- Booming market
- New SEO perspectives
- Better topical maps

I will try to cover more and more topics in my threads.

If you want to support me, be sure to follow me and like/retweet to spread awareness about these topics.