Google can identify and evaluate authors using E-E.A.T.

Google gives more weight to the author of the content when ranking search results. This is made clear by the introduction, perspectives, About this result, and about this author in SERPs.

This article explores the ways in which Google may evaluate content pieces based on their author’s experience, expertise and authority (E-E.A.T).

Google’s Quality Offensive

Google has highlighted that the E.E.A.T concept is important for improving search results quality and user experience on the SERP.

The quality of content on the page, the link signals (e.g. PageRank, anchor texts) and the entity-level signals are all important.

The E-E-A.T. system does not evaluate individual content, as it does with document scoring.

The concept is thematically related to the originator and domain. The concept is independent of both the search intent and individual content.

E-E.A.T. is a factor that influences search results independent of the query.

E-E.A.T. is a term that mainly refers ot thematic areas. It is also understood as a layer of evaluation which assesses a collection of content as well as off-page signals, in relation to entities like companies, organizations, individuals and their domains.

The author is a source of content.

Google began to incorporate the ranking of content sources into search rankings long before (E)E-A.T. The Vince update of 2009, for example, gave brands’ content a ranking boost.

Google tried to collect author ratings through projects such as Knol and Google+. These have been discontinued for a long time.

In the past 20 years, Google has issued several patents that directly or indirectly refer to social networks like Google+ and content platforms like Knol.

To improve the quality of the search results, it is important to evaluate the author or origin of the content.

Google has no business including inferior content on its search engine, given the amount of AI-generated spam and traditional spam.

The more information it has to index and process when retrieving information, the more computer power it requires.

E-E-A.T. can help Google rank content based on entity level, domain level and author levels. This is done on a larger scale without crawling every piece of material.

This macro-level allows content to be classified by the entity that created it and crawl budgets can be adjusted accordingly. Google can use this method to exclude whole content groups from indexing.

How can Google attribute content and identify the authors?

The person entity type includes authors. It is important to distinguish between entities that are already recorded in the Knowledge Graph, and entities that were previously unknown or not validated in a knowledge repository like the Knowledge Vault.

Google’s machine learning and language model technology can extract entities even if they aren’t yet in the Knowledge Graph. The solution is called entity recognition (NER), which is a subtask in natural language processing.

NER recognizes entity types based on linguistic patterns. Nouns are generally (named) names of entities.

Word embedding (Word2Vec), a modern information retrieval system, is used for this.

A vector of numbers represents each word of a text or paragraph of text, and entities can be represented as node vectors or entity embeddings (Node2Vec/Entity2Vec).

The words are classified into a specific grammatical category (noun, verbs, prepositions etc.). via part-of-speech (POS) tagging.

Nouns are entities. The main entities are subjects, while the secondary entities are objects. The entities can be related to each other by using verbs and prepositions.

In the example below “olaf Kopp”, “head seo”, “co-founder” and “aufgesang”, are named entities. (NN = noun).

Natural language processing can identify entities and determine their relationship.

This creates an semantic space which better captures and comprehends the concept of a entity.

More information can be found in ” Google’s NLP and search queries.”

Document embeddings are the counterpart of author embeddings. Vector space analysis is used to compare document embeddings with author vectors. You can read more about this in the Google Patent ” Generated vector representations.”

Vectors can represent all types of content, allowing:

• In vector spaces, compare the content vectors with the author vectors.
• Documents are grouped according to their similarity.
• Assigning authors

The distance between document vectors, and their corresponding author vectors, describes the likelihood that the author has created the documents.

If the distance between the document and the vector is less than the other vectors, and a threshold is met, the document will be attributed to its author.

It can also be used to prevent documents from being created with a false flag. As described above, the author vector can be assigned to a specific author entity using the name of the author specified in the document.

Other important sources of information on authors include:

• Wikipedia articles about the person.
• Author profiles
• Speaker profiles
• Social media profiles.

In the first 20 results of a Google search for the name of a person or entity, you’ll find Wikipedia entries and profiles of the author, as well as URLs of domains directly related to the author.

You can check the mobile SERPs to see what sources Google has a direct relationship.

Google recognized that all the results displayed above the icons of the social media profiles were sources, referring directly to the entity.

This screenshot shows the entities and sources linked in the search for “olaf Kopp”.

Also, it displays a brand new version of a Knowledge Panel. It appears I am now part of a beta-test.

This screenshot shows that Google has linked my domain, my social media profile and my entity directly to the Knowledge Panel.

The About description comes from my author profile on Search Engine Land, USA and the agency website in Germany.

Google can identify authors by using personal profiles on the internet.

Google can assign content to an author by using author boxes or collections. Author names are not sufficient as identifiers because ambiguities may arise.

To ensure consistency, you should be paying attention to the author descriptions of everyone. Google can check their validity by comparing them.

Google Patents that are Interesting for E-E.A.T Rating of Authors

Patents that Google has filed provide a look at possible methods of identifying authors, assigning content and evaluating it by E-E.A.T.

Content Author Badges

This patent describes the process of assigning content to authors by way of a badge.

Author badges are assigned using IDs such as email addresses or the author’s names. Verification is performed by an add-on installed in the browser of the author.

Generating author vectors

Google has signed this patent with a duration of up to 2036. There are only patent applications in the USA which indicates that this is not used yet for Google searches globally.

Patent describes how vectors of authors can be represented based on training data.

The author’s writing style and word choice determines the uniqueness of a vector.

This allows content that was not previously attributed by the author to be assigned, or authors who are similar to each other can be grouped together.

The content ranking for an author or authors can be altered based on past user behavior (on Discover for example).

Content from authors that have been discovered as well as similar content would be ranked higher.

This patent is based upon so-called embeddings such as authors or word embeddings.

Embeddings are today the standard technology for deep learning and natural-language processing.

It is therefore obvious that Google will use such methods for author recognition and crediting.

Author reputation scoring

The patent was signed for the first time by Google in 2008. It has a term minimum of 2029. This patent was originally referring to the now-closed Google Knol Project.

It’s even more thrilling that Google has drawn it again under the new name Monetization online content in 2017. Knol was shut by Google in 2012.

Patents are used to determine a score of reputation. This can be done by taking into consideration the following factors:

• The author’s frame level.
• Publications in well-known media
• The number of publications.
• Age of recent releases
• The author’s official tenure as an author.
• The number of links generated from the content.

Authors can have different reputation scores and aliases for each topic.

The patent makes many points that are relevant to a closed system like Knol. This patent is sufficient at this time.

Agent rank

The Google patent was signed for the first time in 2005, and it has a term of at least 2026.

It was registered not only in the USA but also in Spain, Canada, and around the world, which makes it more likely that it will be used for Google searches.

The patent explains how digital content can be assigned to a publisher or author. The content is ranked according to a number of factors, including the agent rank.

The Agent Rank does not depend on the search intention of the query. It is based on the documents that are assigned to an agent, and their backlinks.

The Agent Rank is exclusive to a single search query or cluster of search queries, or an entire subject area.

The agent ranks may also be calculated based on search terms, or categories of terms. Search terms (or structured collection of search terms i.e. queries) can, for example, be classified into different topics. For instance, sports or medical specialty, an agent could have a rank that is different with respect to each of these topics.

Author Credibility

The Google patent, which was signed in 2008, has a term minimum of 2029 and is only registered in the USA.

Justin Lawyer created it the same as the Patent Reputation Score for an author, and it is directly related to its use in searches.

The patent contains similar elements to the patent mentioned above.

It is for me the most interesting patent to evaluate authors in terms trust and authority.

This patent identifies various factors which can be used algorithmically to determine the credibility of an author.

This article describes how a search-engine can rank documents based on the credibility factor of an author and their reputation score.

The reputation of an author is based on the number of topics that they have published on.

The publisher’s influence on the author’s score is not relevant.

In this patent there is another reference to links being a factor that could be used in an E.E.A.T rating. The number of published links can affect an author’s score.

These are some possible signals that can be used to determine a score of reputation:

• How long has the author been writing in a particular subject?
• Authors are aware of their work.
• Users’ ratings of content published.
• If the content of an author is published by another publisher with ratings above average.
• The author’s content.
• When was the last time an author published?
• The author’s ratings of similar publications.

The patent also provides other interesting information regarding the reputation score.

• The reputation of an author is based on the number of topics that they have published on.
• The publisher’s influence on the author’s score is not relevant.
• If the same content or excerpts have been published more than once, it may affect your reputation score.
• The reputation score can be affected by the number of links that lead to the content.

The patent also addresses the credibility of authors. The following factors have been mentioned as influencing ones:

• Verified information on the author’s profession or role in the company. The credibility of the business is also taken into consideration.
• Relevance of occupation with the topic of the published content
• The author’s level of education and experience.
• Experience of the author based on time. The more an author has published on a subject, the more credible they are. Google can determine the experience of an author/publisher based on the date of their first publication.
• The number of articles published on a particular topic. It is assumed that an author who publishes a lot of articles on a particular topic has some credibility.
• Time since last publication. The more time has passed since an author published on a particular topic, the lower a reputation score may be for that topic. The higher the score, the more current the content.
• Mention of the author/publisher on award lists and best-of lists.

Methods and systems for re-ranking search results

The Google patent, which was signed for the first time in 2013, has a term of at least 2033. The patent has been registered both in the USA as well as worldwide, so it is likely that Google will make use of it.

Chung Tin Kwok is one of the inventors, and was also involved in other Google patents that were relevant to E-E.A.T.

In the patent, it is described how, in addition, to the references of the author’s contents, search engines can also take into account the proportion he can add to a corpus of thematic documents in an author score.

In some embodiments, determining the original content score for a respective entity involves: identifying a number of portions in the content index that have been identified as associated with the entity in question, each portion representing a certain amount of data within the content index; and calculating the percentage of those portions in the content index that are the first instances.

This document describes the re-ranking search results according to author scores, which includes citation scores. Citation scores are based on how many references an author has to his documents.

The proportion of the content an author contributed to a corpus containing documents on a particular topic is another criterion.

“[W]herein determining the Author Score for a respective Entity includes: determining a Citation Score for the respective Entity, wherein a citation is the frequency of citing content associated with that respective entity; determining an Original Author score for that respective Entity, wherein an original author is the percentage of content related to the respective entities which is the first instance in an index known content; and finally combining both the citation and original author scores using a predetermined formula to produce the Author score

It can be used to evaluate authors in general, as well as identify “copycats”.

The rating system for authors

Here are some more factors to consider. I’ve already discussed a few of them in my article, “14 ways Google might evaluate E-A.T“.

• Content quality: This is the overall quality of a topic, regardless of format or domain. This can be done by using user signals, links or other content-level quality signals.
• PageRank or links to the author’s work.
• The author’s co-occurrence in the content (podcasts or videos, websites, PDFs or books), with terms or topics relevant.
• Search for the author’s name in conjunction with topics or keywords relevant to your search.

Author entities can be tagged with E-E.A.T.

On a large-scale, machine learning techniques make it possible for unstructured content to be recognized and mapped in terms of semantic structure.

The Knowledge Graph now allows Google to understand and recognize many more entities.

The source of the content is becoming increasingly important. E-E-A.T. can be applied algorithmically beyond documents, domains and content.

The concept can also include the author entities (i.e. the authors and organizations that are responsible for the content).

In the coming years, I believe that E-E-A.T. will have an even greater impact on Google’s search results. This factor could be just as important as optimizing individual content for relevance.

The Search Engine Land article How Google can identify and evaluate authors using E-E.A.T first appeared on Search Engine land .