A lot of SEOs believe that sentiment can affect the ranks of a webpage on Google. If all the webpages in the Search Engine Results Page have a positive sentiment, a page with negative sentiment will not have a good rank. At present, Google’s research is focused on sentiment analysis.
People are an ever-evolving source of signals that needs to be mined. Sentiment analysis is also known as opinion mining of people. It is the computational study of the opinion of people, their attitudes, and emotions towards a particular entity.
Now, to make things clear, a web page is not ranked because of the sentiment analysis. Sentiment analysis helps in understanding a webpage so that it can be ranked. Google needs to understand a webpage to rank it. Here is where the importance of sentiment analysis comes in.
Considering the usefulness of sentiment analysis in ranking a webpage, Bill Slawski, an expert in Google related patents states that Google would want some diversity if it wants to rank a webpage based on sentiment. There will not be all negative or all positive.
What he means is, sentiment doesn’t reflect the potential information gain from a content. The information gain is understood using neuro-linguistic programming to extract knowledge. Positive or negative sentiment is not the reflection of how much information/knowledge is present on a topic.
Bill Slawski thinks that Google is not going to favour one sentiment over another. That will be showing a bias towards a topic. This is why Google will look for some kind of diversity. Thus, if they want to rank a webpage based on sentiment, it will not show all positive or negative.
Google has been mostly silent on the use of sentiment analysis since 2018. On Twitter, someone asked, “…it seems like your search algorithm recognizes and takes into account sentiment. Is there a sentiment search operator?”
To that, Danny Sullivan, an American journalist replied, “It does not recognize the sentiment. So, there is no operator for that”. Earlier that year, there was an official Google announcement about the featured snippets, where sentiment was mentioned.
The context was, for some queries, there can be a diversity of opinions since Google might show two featured snippets, a positive and a negative. Google is exploring solutions for showing results of multiple responses.
For instance, people who search “are reptiles’ good pets” should receive the same featured snippet as “are reptiles bad pets”. Both of them are seeking the same information, which is how reptiles are pets. The page that states reptiles are good pets is the best for people looking whether they are good and vice versa. This is what Google is trying to resolve. Since the year 2018, Google has stopped showing any featured snippets for vague queries. It is encouraging users for a specific question.
Terminology crossover is an important aspect of sentiment analysis. It doesn’t account for a sentiment that is expressed by a domain-specific word. The words that are positive in one domain could be negative on another domain.
To deal with this, Google has set a domain-specific-sentiment lexicon. It can be used on documents of a specific nature. Sentiment classification also assists web searchers in seeking information regarding an entity by summarizing the sentiment of an entity.
The entities include companies, products, and people. The sentiment can be positive, negative, or neutral. The documents expressing these sentiments include webpages, texts, newspapers, magazines, emails, newsgroup posts, and any other electronic messages.
Now, a system that classifies a review will need to understand the positive or negative opinion of a sentence or a phrase. This is called opinion mining. It is imperative to classify sentiment on different levels because different applications have different needs.
For instance, a summarization system for the product reviews will require polarity classification at the sentence or phrase level. On the other hand, a question answering system will need a sentiment of paragraphs. A system that determines the articles that create an online news source will need a document level analysis.
A search engine cannot accurately answer a question without understanding the web pages that it wants to rank. It is not about using the data as a ranking factor. It is about using the data for understanding the pages so that they can be ranked using the ranking criteria.
There are multiple perspectives on sentiment analysis. One of them is, it is a way of obtaining candidate webpages for ranking. When a search engine can understand a webpage, it can apply the ranking criteria on the pages that will answer the question.
This is important for search queries with multiple meanings. Google neither rank nor answer a question that Google cannot understand. Google has stated multiple times that they don’t show pages that reflect the searcher’s sentiment or intent.
In fact, Google says the opposite. It tries to show a wide array of options. There is no sentiment analysis bias at Google. It is not led by a sentiment expressed within a search query. Even though Google claims it, the search results show something different.
If sentiment analysis is used by Google, a web page isn’t ranked because of the sentiment analysis. Sentiment analysis helps a web page be understood so that it can be ranked. Google can’t rank what it can’t understand. Google can’t answer a question that it can’t understand.
It is clear from the result that Google’s algorithm IS using sentiment analysis but it is important for Google to perceive the intent of the search. Thus, if Google receives a specific intent from the search query, it is capable of sharing the type of results a searcher is looking for. Sentiment analysis is a very valuable medium that digital marketing companies can use to understand the intent of their target audience.