In any modern search engine, there are a significant number of heterogeneous ranking systems (each with its own set of factors), performing their own, independent tasks.
Google has the following subsystems:
✅BERT
it is an artificial intelligence system used by Google that allows us to understand how combinations of words express different meanings and intentions.
✅ Deduplication systems
A Google search can find thousands or even millions of relevant web pages. Some of them may be very similar to each other. In such cases, our systems only show the most relevant results to avoid unnecessary duplication. Learn more about how deduplication works and how to optionally see omitted results when deduplication occurs.
Deduplication also happens with thematic snippets. If a web page listing rises to the level of a featured snippet, we don’t repeat that snippet further on the first page of results. This reduces results and helps people find the information they need more easily.
✅ Exact match domain system
Our ranking systems take into account the words in domain names as one of many factors that determine the relevance of content to search. However, our exact domain matching system works to ensure that we don’t give too much value to content hosted on domains designed to match specific queries directly. For example, someone might create a domain name containing the words “best-places-to-eat-lunch” in the hope that all those words in the domain name will push the content high in the rankings. Our system takes this into account.
✅ Freshness systems
We have various systems for identifying “queries that require freshness” designed to show more recent content for queries that you would expect it to. For example, if someone is looking for information about a movie that has just been released, they are most likely looking for recent reviews, not old articles about getting started on a movie. As another example, a search for the word “earthquake” can usually return materials about training and resources. However, if the earthquake happened recently, then news articles and more recent content may appear.
✅ Helpful content system
Our Useful Content System is designed to ensure that people see original, useful content written by people for people in search results, not content created primarily to get search engine traffic.
✅ Link analysis systems and PageRank
We have various systems in place to understand how pages relate to each other in order to determine what is being talked about and which pages might be most useful in response to a query. Among them is PageRank, one of the main ranking systems used in the creation of Google. Those who are interested can learn more by reading the original scientific article and the PageRank patent. PageRank has changed a lot since then, but it’s still part of our core ranking system.
✅ Local news systems
We have systems in place that are working to identify and showcase local news sources when appropriate, such as through our Top Stories and Local News features.
✅ MUM
The Multitasking Unified Model (MUM) is an AI system capable of understanding and generating language. It is not currently used for general search rankings, but for some specific applications, such as improving searches for information about the COVID-19 vaccine and improving display snippets.
✅ Neural matching
Neural matching is an artificial intelligence system that Google uses to understand the representations of concepts in queries and pages and match them to each other.
✅ Original content systems
We have systems in place to help ensure that we show original content in search results, including original posts, before those that simply cite it. This includes support for special canonical markup that creators can use to help us better understand which page is the master page if the page has been duplicated in multiple places.
✅Removal-based demotion systems
Google has a policy that allows certain types of content to be removed. If we process a high number of such takedowns on a particular site, we use this as a signal to improve our results.
✅ Page experience system
People prefer sites that provide a great page experience. That’s why we have a page quality scoring system that evaluates various criteria such as page loading speed, mobile friendliness, lack of intrusive ads, and page safety. In situations where there are many possible matches with relatively equal relevance, the system helps to prioritize content with the best page experience.
Passage ranking system
Passage ranking is an artificial intelligence system that we use to identify individual sections or “passages” of a web page in order to better understand how relevant a page is to search.
✅ Product reviews system
The Product Review System aims to better reward high-quality product reviews, content that contains in-depth analysis and original research, and is written by experts or enthusiasts who are well versed in the subject.
✅Rank Brain
RankBrain is an artificial intelligence system that helps us understand how words are related to concepts. This means that we can better return relevant content, even if it doesn’t contain the exact words used in the search, by understanding how the content relates to other words and concepts.
✅ Reliable information systems
Numerous systems work in different ways to show the most reliable information, such as helping to surface more authoritative pages and downgrading low-quality content, as well as improving the quality of journalism. In cases where reliable information may be lacking, our systems will automatically display content tips for rapidly changing topics, or where our systems are not very confident in the overall quality of the results available for search. They provide tips on how to search in a way that produces more useful results.
✅ Site diversity system
Our site diversity system works in such a way that we generally do not show more than two web pages from the same site in the top search results so that no single site dominates all of the top results. However, we may show more than two results in cases where our systems determine it is particularly important for a particular search. A variety of sites generally treat subdomains as part of the root domain.
✅ Spam detection systems
We use a range of spam detection systems, including SpamBrain, to combat content and behavior that violates our spam policy. These systems are constantly updated to keep up with the latest evolution of the spam threat.
✅ Crisis information systems
Google has developed systems to provide useful and timely information during crises, whether personal emergencies, natural disasters, or other widespread crises.