How can I increase sentiment score of my content?

[Ababilstar]

As a newbie, I was unaware of NLP. Though I heard about it, I didn't care about it. But as I am now focused to increase the quality of content, I decided to take look. I found a youtube video where the Youtuber showed how google analyzes the content with the BERT algorithm.

He used Google Cloud's Natural Langauge API to measure the sentiment score manually. In the next video, he used Surfer SEO to keep the content in positive sentiment besides other features. However, I checked one of my content with google's NL API. It showed a neutral score. That also showed that the negative sentences, the lines with the words or phrases like: recommend, prevent, only, have to, need, etc are red marked with a negative score. Which means they have negative sentiments.

So, here is my question.
We are humans. we often use these types of words that are seen as negative in google's eye. Then what should I do? Is it necessary to avoid all these sentences?
I have got another understanding. That is lowering the number of these negative sentiment containing lines considering the size of my content. Is this correct?

I have another question. Is there any list of positive and negative sentimental words for NLP? Or any suggestion like: "Don't use this _ _ _ _ _ , use this _ _ _ _ _ instead." I know that there are many guides like this. But I am looking for one which is specialized for NLP.

Love and respect for BuSo.

 

[Philip J. Fry]

You assume that sentiment is a ranking factor. This is not true. Sentiment is use to analyze things like tweets but it’s not use like that for search engine rankings. NLP is used to determine relevancy in search. When someone searches, they want relevant, authoritative results. It doesn’t matter if the sentiment is negative or positive.

In fact, let’s go with your argument that sentiment does matter. You’re assuming that Google only wants to show positive sentiment articles. That’s absurd. That means that all review articles anyone googles would be positive. Negative reviews or neutral reviews are just as helpful. Google doesn’t use sentiment as a ranking factor, even if we use your assumption.

 

[Ababilstar]

You assume that sentiment is a ranking factor. This is not true.

I am not sure. But the video I watched showed something different. In that video, he searched for a keyword and neither of top 10 results had a negative sentiment score.

I'm not saying that they ranked only for sentiment score. But is it completely ignorable?

Would you please clarify?

 

[Philip J. Fry]

That was just one example of a correlation. It doesn’t mean that all Google search engine results pages are expected to be like that.

No, I’m not going to explain to you how search engines are built. You can just read a search engine engineering textbook.

 

[Ryuzaki]

I'm not saying that they ranked only for sentiment score. But is it completely ignorable?

I never, ever bother with positive or negative sentiments when it comes to SEO on Google and I do just fine. Google wants to be an Answer Engine more than a Search Engine, and sentiment has nothing to do with that.

However, other site's algorithms, like Facebook, most definitely have sentiment built in, and I don't mean just rewarding negative behavior like Twitter does with the form of likes. Facebook understands sentiment before the users tell it with the 5 or 6 emoticon reactions and change the reach of the posts accordingly. Because they want to be an Emotion Engine. They know what sentiment each individual likes to engage in and pushes that to them.

It's something you'd want to be aware of in Facebook PPC ads even. But for Google... if they're attempting to decipher and understand sentiment better, it's likely only to better understand and fulfill the intent of the searcher's query.

 

[Philip J. Fry]

But for Google... if they're attempting to decipher and understand sentiment better, it's likely only to better understand and fulfill the intent of the searcher's query.

Eh, I doubt it. For example, a query that has sentiment intent would be "BrandName scam." The pages that rank for this query shouldn't be towards any sentiment in particular but for whatever pages are most relevant to "BrandName scam" and with the most authority (PageRank). There might be more negative sentiment showing up, given the nature of the query, but that doesn't mean that negative intent pages rank better on that query. Correlation does not imply causation in this example.

The only time I know of where Google mentions anything about sentiment is when quality raters research the a website or the main content. However, this is only with quality raters and hasn't been added to the algorithm yet, as far as I know.

"2.6.1 Research on the Reputation of the Website or Creator of the Main Content

Use reputation research to find out what real users, as well as experts, think about a website. Look for reviews, references, recommendations by experts, news articles, and other credible information created/written by individuals about the website. Stores frequently have user ratings, which can help you understand a store’s reputation based on the reports of people who actually shop there. We consider a large number of positive user reviews as evidence of positive reputation. Many other kinds of websites have reputations as well. For example, you might find that a newspaper (with an associated website) has won journalistic awards. Prestigious awards, such as the Pulitzer Prize award, or a history of high quality original reporting are strong evidence of positive reputation."

https://static.googleusercontent.co....com/en//searchqualityevaluatorguidelines.pdf

 

[Ryuzaki]

@Philip J. Fry, yeah, reading that quote from my post makes me realize I worded that weird. I think if they're working on sentiment, it's only to understand the query better, and to fulfill the search intent better, but not in relation to delivering results based on sentiment. Simply to better understand the queries. I'd see it as another onion layer on top of natural language processing.

Syntax, short form typing, sentence fragments, sentiment... all of that helps decipher the query, but isn't applicable to the results themselves. I hope that's more clear.

However, this is only with quality raters and hasn't been added to the algorithm yet, as far as I know.

This is tricky. My understanding is that they're using the human raters to validate what the algorithm is already doing, and not the other way around.

I'd argue that the quote you pasted isn't really about sentiment as defined in the use of language parsing.

Brand reputation definitely requires sentiment if they want to parse text and all, but they're clearly in that quote talking about things they can objectively and quantitatively define, like star ratings, thumbs up versus down, and even the number of awards a brand has won.

Sentiment, in this regard, seems to be more about positive and negative adjectives, verbs, and nouns, which are qualitative in nature.

Regardless, we're on the same page. I do NOT think sentiment is involved in the re-ranking of the SERP results. I think it would be detrimental to do so, as you've stated. It'd be among the most easily manipulatable pieces of the algorithm to ever exist.

 

[Philip J. Fry]

@Ryuzaki Thanks for making it crystal clear. We are on the same conclusion, that SERP rankings have nothing to do with the sentiment of the page; however, arriving at a correct conclusion from faulty reasoning is not knowledge. Knowledge is a justified, true belief. The belief, the conclusion, must be true, as in factually correct, and the justification needs to be correct too. if you have a correct conclusion but, if the justification was incorrect, that's not knowledge. That was just luck. If you're going to make an investment of time and money, which entrepreneurs and investors do, you want to make decisions with knowledge. Not doing so is gambling and speculating.

As for quality raters, yes, you're right. I confused quality ratings up with the testing process for an algorithm change. Per https://www.google.com/search/howsearchworks/mission/users/ :

We work with external Search Quality Raters to measure the quality of search results on an ongoing basis. Raters assess how well a website gives people who click on it what they are looking for, and evaluate the quality of results based on the expertise, authoritativeness and trustworthiness of the content. These ratings do not directly impact ranking, but they do help us benchmark the quality of our results and make sure these meet a high bar all around the world.

Thank you. So, now we know that current SERPs are being reviewed by EAT on the main content and domain level. That explains why (good) big brands are ranking well and sites with no reputation or a bad reputation are having a hard time ranking: Google has set qualitative guidelines to reward sites with good qualitative properties. If you're not building a brand, you're going against the tide.