Question

So I've avoided NLP SEO for some time because I thought it was related to "Neuro-linguistic programming" but just recently found out it's something completely different (lol) But after reading about it a little, isn't it. just doing proper technical SEO with content that has relevance? Proper silos with relevant content fulfill NLP SEO, right? I have only just skimmed over how it works, but yeah, is there anything more to it?  

Answer 1

More accurately "NLP SEO" is just another buzz expression with no standard definition. Natural language processing isn't even what most people seem to think it is. It's a broad set of predictive methodologies or heuristics. "Heuristic" is a fancy word computer scientists use for "short cut" or "we're too impatient to wait for our primitive technology to crunch through all this data so we're just going to take a wild-a** guess here and hope it works out in the end". The goal behind natural language processing isn't to understand content. We don't have computers that understand anything. The goal is to improve an algorithm's ability to pick one or more groups of words most likely to be associated with another set of words in a correct grammatical and idiomatic sequence. It's the computer science equivalent of a parlor trick done on a massive scale. What you need to know is that NLP makes it easier for publishers to just write about what they want to write about rather than use a formula to plug keywords into specific points of a page. Using the keywords still helps - will probably always help - but the NLP algorithms are better able to match queries to documents by predictive models rather than exact-match models.  

Comments

@narcotize I am curious to know how predictive model works? Do they predict the result by understanding the user behavior from set of data?  

---

@inflationism In the simplest terms, they are looking for/at weighted averages that represent the relationships between words - in natural language. Broadly speaking, the way you refine a prediction is to look for more signals that provide context. So you could say they are looking for super-relationships between groups of (words and phrases they have determined are related to each other). It would appear from all the work they do with queries that Bing and Google (and probably also Baidu and Yandex) are using query data to identify patterns that can be used to improve their predictions. Thus, yes, as far as that can take them, they are "understanding the user behavior from set data". But they also build click models that attempt to simulate user behavior and they create prediction models based on what they observe in those click model experiments. And they do other tedious, boring but cool stuff, too (of which I know only a fraction).  

---

@narcotize Got it. Thanks.