google index search

Semalt: Understanding The Google Knowledge Graph And How It Works



To facilitate Google's understanding of your web content, it is important that we understand how to influence topics in the knowledge graph. Semalt is a website dedicated to providing the best services for her clients, which means understanding everything SEO-related. As a company that focuses on both web design and web development, we incorporate SEO into every aspect of a website. This, of course, included an understanding of Google Knowledge graphs. 

What makes knowledge graphs special? 

Without a knowledge graph, it would be difficult for a search engine, particularly Google, to use structured data for entities. Markups and semantic data, in turn, help connect ideas and concepts, making it easier to transform them into structured data that we can use to fill or influence Google's knowledge graph. 

As SEO professionals, it is our duty to understand ways which we can influence these topics in the graph, so we can influence meaningful changes in Google's understanding of their content.

As SEO professionals, we understand how to influence topics in the graph to influence meaningful changes in Google's understanding of our client's content.

What is Google Knowledge Graph?

We consider knowledge graphs to be an intersection between a database and an encyclopedia. Developers refer to each article in a knowledge graph as an Entity by or a Topic in Google's customer-facing articles.

An entity can be about absolutely anything. Like with most databases, it has its unique identifier, which you sometimes see in Google's URLs. It looks like this: [kgmid=/g/11f0vfyswk&hl], the parameter name "kgmid" can, however, change depending on the entity type.

Usually, there are multiple statements about an entity: 
  • A title or name (like "George Bush").
  • Type or Types (like "person").
  • A Description (like "was once the President of the United States").
  • A list of image URLs (usually associated with usage rights).
  • A detailed description (some explanatory text with a URL of the source).
Google, however, states that while the information in the list above might be available directly in their search API, this data is considerably augmented internally. 

So in the example above, the data might also include the date Bush was born and died; it may include that he was married to Barbara Bush. It may mention some of his accomplishments in the office and so on. 

With all the information provided, it becomes easier to see how there isn't much difference when compared to an encyclopedia article. The difference, however, is that because all the facts are categorized into fields like "Number of years in office," machines find it easier to connect the dots between topics. Machines also find it easier to retrieve the right information almost instantly when a person makes a request. For example, if you search who was married to George Bush? You should see this:


Disambiguity

There are many knowledge graphs around the world, and Google's knowledge graph is just one of these many graphs. Other knowledge graphs include Wikidata.org, dbpedia.com, and more. Indeed, any form of data that is semi-structured can be described as a knowledge graph which is why databases like IMDB or encyclopedias are referred to as knowledge graphs. 

Initially, Google built out its knowledge graph from other data sets, including the CIA Factbook and Wikipedia.  We've also heard rumors that Google's knowledge panel can also function as its knowledge graph. This is however incorrect and should be disregarded. While the knowledge panel sometimes represents a subset of data in the graph, they aren't the same thing. 

The knowledge panel can also function as a visual representation of data items connected through Google's knowledge graph. Google's knowledge graph is, however, a less visual record about topics. 

The final point we seek to disambiguate is the terminology itself. Generally, a keyword knowledge graph is made of entities or topics. For the sake of clarity, we've noticed that Google prefers using the term "topics" considering that it is the term they use constantly in their public documentation. 

The topic tends to be more user-friendly, but it also made it difficult to understand when Google refers to topics or entities. 

Entity topic types

Generally, entities are given topic types. This could be a person; Event; Organization, place, or country. If the entity isn't any one of these, it is labeled as a "Thing". Hopefully, Google will continue to develop new entry types so that we will use "Things" less often. The Natural Language Processing API used by Google provides vital clues which prompts us to wonder if many topic types are actually being used. However, there are several other entity types found on Google's Knowledge Graph search API developer page. Currently Google seems incapable of categorizing as many entities as it has listed. We hope that this will change as time goes by. 

Besides, only about 20% of entities are recognized by the Natural Language Processing Algorithm used by Google when compared to their public offering. 

Benefits of Knowledge Graph

By organizing the world's information into topics, it becomes easier to crawl and index webpages and websites that search engines leverage. These include diversity, scale, integrity, and speed. 

Scaling benefits

The number of web pages on the Internet is a highly debatable subject. While some may argue that they are countless, the number of web pages most likely falls in the trillions, and this number keeps on expanding at vast rates daily. In contrast, the number of topics we as humans understand is significantly smaller, and the number is growing at a far lower rate. 

This indicates an inordinate duplication of the same ideas shared across several contents you find on the Internet. By sorting the details of a topic using a semi-structured system, the information contained on the Internet can take a smaller space and wouldn't be duplicated as much. 

Benefits of having different data source

By storing information about a topic, citing multiple data sources become way easier and you no longer have to send users to the same webpage over again. This tells us that Google can and does collect salient details about topics and have them displayed using a screen or a bunch of several other mediums but in an improved way for the user's query or the user. 

In theory, information about a topic may sometimes be derived from sources outside the Internet. 

Information Integrity Benefit

Whilst Google Knowledge Graph is known sometimes to contain errors and factual inaccuracies, it sometimes gets manipulated by bad actors or SEO professionals. This approach, however, has the benefit of providing a single point of truth to Google. However, this may not be the case for controversial topics. 

A new fact about any topic needs to pass through several verification processes before it can be added to Google's knowledge graph. But it is unlikely that Google will openly discuss these thresholds. 

On the negative side, one point of truth can be prone to bias or reduce the diversity of information if the data sources themselves are already adulterated. 

Information retrieval benefits

Thanks to sorting out information by their topics, retrieving information has become much faster both by Google and by users who may be searching for a needle of information in a haystack of web pages. 

Where is Google's Knowledge graph Used?

For many Android and smartphone users, the knowledge graph is mostly used for its "discover" feature on their home screens. Google uses information about a user's behavior to understand which topics are most important to you, and they can help users find those related topics based on their usage history. 

Google trends is also another place where users can find interesting topics. From that point, Google can give an indication of which topics are closely related to each other. However, current evidence suggests that this data is currently gathered by pulling other searches by the same user, which may lead to unexpected related topics popping up. 

As with the panel, when a user types in a search query that can be answered using a knowledge graph, you also get top-quality search results. Oftentimes, these results overshadow core search results, which end up pushing links to website fathers down the line. 

These answers are also displayed in a way that Google can use to provide answers for users who use voice search. 

Conclusion

Topics are highly visible in image searches, so you can find relevant information from searching using the picture of a relevant place or person. 

This is an excellent example of the ways in which Knowledge graphs are used as repositories for other data sources besides webpages.