Thread by @GiordMarco96

I've talked about Natural Processing Language (#NLP) before. What is the difference with NLG and NLU?

Behind these terms lies something more important for #SEO Specialists.

I will explain you what are these strange acronyms in this thread 🧵

For the NLP definition, check my other thread on the topic. It is a clear and concise explanation on the subject.

https://twitter.com/GiordMarco96/status/1499842120068177920

Natural Language Generation (NLG) can be defined as the use of Artificial Intelligence to create content.

This is what tools like Jasper.ai do. They can generate texts according to your instructions and depending on how they are trained.

Natural Language Understanding (NLU) focuses on how machines understand human language.

The goal is to make the most sense out of unstructured data for a machine and deliver critical insights.

An example of NLU is provided by content summarization. A machine "understands" the content and is able to reduce it.

This apparently easy task is super complex and took years of work to reach the level you see now.

The difference between NLP and NLG is quite clear.

Natural Language Generation is focused on creating text. How can I turn structured data into text?

NLP starts with unstructured to give you structured data.

NLU is clearly more focused on parsing and understanding a text.

Profanity filters are a clear example of a real-world application of NLU.

Source: nlp.stanford.edu/~wcmac/papers/…

The understanding of human languages goes through 3 key steps, also called levels of linguistic analysis:

- Pragmatics: language in use and context
- Semantics: meaning
- Syntax: order of words and phrases (the grammar)

We can say NLU is focused on reading comprehension from the machine point-of-view.

The 3 levels mentioned before play a crucial role in this process of understanding a text.

NLU and NLG are generally considered subsets of NLP.

Source: ibm.com/blogs/watson/2…

Simply put:

- NLP > reading
- NLU > understanding
- NLG > writing

This is the fastest way to think of these concepts. An #SEO Specialist can just focus on the most important tasks and that's it.

NLG is the most popular now due to text generation services.

The most common language to test different NLP tasks is #Python. The SEO community clearly prefers it due to its simplicity and relatively ease of use.

It's a great replacement for Excel but gives its best at scripting and NLP.

These distinctions are not practically relevant when you write code. They are important to understand the theory and the difference between tasks.

If you want to just practice you can jump to code!

The most common tasks that are useful for SEO involve:

- NER
- POS tagging
- Topic modeling
- Semantic clustering
- N-grams

This is only a selection of the most useful.

While some other tasks like Sentiment Analysis are great and cool, they are less useful than what I mentioned before for optimization.

Ok sentiment may play a vital role when writing some pieces of content but it's not crucial in general.

There are new and advanced NLP tools right now and some of them combine Knowledge Graphs.

If you are working on big projects I recommend you to use prebuilt tools and not even bother to reinvent the wheel.

On the contrary, smaller projects or even prototypes can be solved with custom solutions. This will help you in understanding what I usually mean that sometimes you have to get by.

Python is the most valuable companion in most (but not all) cases.

Modern NLP is focused on the so-called transformers, models with a specific type of architecture. I will cover them in detail in another thread as this would be too complex for the moment.

While an SEO Specialist is not interested in going too deep into technical details, it is highly beneficial to have general culture on such topics.

This is to improve communication with technical people and to have a more complete understanding of patents.

I will cover more details in the next threads. I am studying everyday to improve my knowledge on the topic and especially on the most important concepts.

For Pythonistas, the NLP libs I recommend are:

- transformers
- BERTopic
- spaCy
- nltk
- openai

The ones I use the most for the moment.

I admit that most sources are super confusing and sometimes contradicting each other.

An SEO Specialist doesn't need to understand the details, a high level overview is more than enough.

Learn it once and try to focus on practical applications instead.

Moving on to practical applications for industries, the goal is to create systems that reduce manual work and get you better insights.

Modern SEO analysis has better tools now and can deliver significantly improved outputs.

Even an open-source "tool" like Python can be good.

You don't need to go big at the start. Test the libraries I suggested above and see what works for you.

You can enchance your SEO analyses by including new details, such as entities or n-grams.

While they're cool, they need to be justified and communicated properly.

NLP technologies will be even more predominant in the future and for good reasons.

This umbrella term could include NLG and NLU as well, as explained before.

That's why studying some Linguistics is always a good idea.

One of the most underrated content on the web. This is pure gold and so are the other articles that often mention NLP.

Do yourself a favor and study from here.

wordlift.io/blog/en/serp-a…

I will publish more tips in the next few weeks, it takes quite some time to make them digestible.

If you are indecisive whether to start NLP or not, just do it.

If you don't feel like it's your path just pay for a tool or use prebuilt solutions.

No reason to overthink, get started ASAP so you will have less work to catch up later.

My threads and posts are finalized to give you the essential understanding of some topics.

They are starting to get more popular but we are still far from reading about them on other channels.

This is why awareness is one of the most important forms of support you can achieve.

If you liked this thread, please like and retweet to spread the message across the SEO community.