Guide to Knowledge Graph Semantics 101

A Knowledge Graph is Structured. A Knowledge Graph is Connected. A Knowledge Graph is Collaborative.

The goal in any form of marketing is one of communication: you want to communicate your information to the world. In order to connect with the right people at the right time and through the right channel, your content needs to be properly understood.

This is hard enough to do successfully between people, but it becomes even more difficult when machines become conduits for this information. Not only must we then anticipate the needs and interests of our audience, but this information needs to be translated into a machine-readable, processable, and searchable format.

As Google themselves stated: “At its core, Search is about understanding language.” Following this statement, it only makes sense that the biggest challenge facing SEO today is one of semantics. At its most basic, semantics is a term for studying language, how it expresses meaning, and the rules it must abide by to convey this meaning. It comes as no surprise that semantics have had a significant impact on search function and SEO strategy.

Semantics is a term for the study of language, how it expresses meaning, and the rules it must abide by in order to convey this meaning.

The problems and promises of semantics in search were already being explored by Tim Berners-Lee as early as 1999. But his 2001 article The Semantic Web, written in collaboration with James Hendler and Ora Lassila, brought the capabilities of semantics in computing to the fore.

They imagined a not-so-distant future where humans could speak with agents in order to access information, set up appointments, and “assist the evolution of human knowledge as a whole.”

This would first require the creation and application of ontologies — formal vocabularies to define concepts, data objects and the relationships between them. Ontologies improve communication between humans and computers by mobilizing semantics through the addition of meaning and context. 

Ontologies…formally define data objects and the relationships between them…by mobilizing semantics through the addition of meaning and context.

Semantic Technology

The potential of the Semantic Web inspired the creation of many domain-specific ontologies, but none provided the breadth, depth and flexibility necessary for the vast array of data that search engines would be dealing with. So, in 2011, Google collaborated with Microsoft, Yahoo and Yandex to develop the Schema.org vocabulary.

This informal ontology, of over 840 types and 20+ properties per type, can be applied to web content in the form of Microdata, RDFa, or JSON-LD. Once applied, schema markup allows machines to understand the difference between things, like local businesses and products. Moreover, properties can be added to provide more information and context to this data.

Does the local business serve a specific area? Does the product come in different sizes or colours?

This is information that users were querying about mostly by way of keywords, or strings. Schema.org’s ability to define and connect information would turn data into a graph of things and transform the capabilities of search in the process.

The initial aims of the Semantic Web progressed further in 2012 when Google first introduced their Knowledge Graph. Google described this endeavor as “a critical first step towards building the next generation of search results” which would benefit from the collective and collaborative intelligence of the web to understand the world “a bit more like people do.”

With the introduction of the Knowledge Graph, and the Hummingbird, RankBrain, and BERT algorithms, search became capable of dealing with ambiguous language, providing summaries, and connecting queries to other relevant information. Thanks to Schema.org’s vocabulary, strings of text could be understood as things containing unique properties and relationships with other things on the web.

With this process in mind, Google’s Knowledge Graph is best defined as a compilation of information from one or more pages that presents answers to search queries in the rich results, enhanced SERP results, and knowledge panels we’ve come to associate with trusted content.

Google’s Knowledge Graph is best defined as a compilation of information from one or more pages that presents answers to search queries in the rich results, enhanced SERP results, and knowledge panels we’ve come to associate with trusted content.

How a Knowledge Graph Works

Learning to implement schema markup is much like learning another language, one that allows your web content to be understood by machines. The team at Schema App understands the inner workings of these processes on an intimate level. Our co-founder Mark Van Berkel is, himself, a Semantic Technologist. We understand that every time you add markup, you are asserting that something exists and defining how it relates to other things in the world. 

To illustrate our approach, we’ve created a conceptual illustration of the main processes contributing to a Knowledge Graph: Data, Standardization and Connection.

Data 

Content on your website is created and presented to be accessible to a human audience. Be it an automotive business, a medical office, or a software company, all of these things have unique capabilities, features, and teams of people making it all happen. Your web presence should reflect that. Without structuring this data, machines have difficulty accessing and processing it.

Standardization

The Schema App tools have been built around the Schema.org vocabulary to ensure that your markup follows the standards expected for semantic search. From within our Editor tool, you can see the hierarchy of Schema.org types and expand them to find more precise terms.

While the Schema.org vocabulary is quite robust in some areas, they explicitly state that they are not a “universal ontology.” For this reason, certain industries, such as automotive and medical, have created extensions for more specific terms. We have imported those extensions to better accommodate the needs of our clients.

Every time you create a data item, our Editor lists all of the properties available to that type. It also notes which properties are required and recommended in order to be eligible for Google’s rich results. Beyond including the correct properties, adhering to the syntax of the code is essential. One misplaced comma or period can result in an error.

Using Schema App’s tools, you can rest assured that your markup won’t risk the syntax errors that come with manually coding. Standardization is at the heart of structured data and we’ve made sure to embed it into our tools at every step.

Connection

But applying an ontology can only get you so far. In an interview with Steve Macbeth of Microsoft, he notes that “Semantics without the ability to connect to other data is almost as valueless as no semantics…[S]emantic data [is] only valuable in my opinion when it can be bridged to other data.” This is another factor we take into account in our tools. Because of our background in semantic technologies, we are passionate about connecting your content. 

This is why we developed the Schema Paths Tool. Using Schema.org can feel like learning a complex language with its own grammar. The Schema Paths tool functions as a kind of translator, simplifying the process, saving time, and providing options for how real world entities, or data items, can connect to one another.

By simply inserting the two types you want to connect, you can see every possible predicate to connect them, and choose the one that most appropriately articulates their relationship in the knowledge graph of your content.

We also know that a key differentiating factor of knowledge graphs is the interconnectedness within and between your concepts. Our tools help you create consistent URIs so you can begin to reuse and reference your data items as more than simple labels for your content. When it comes to knowledge graph management, we make your data easy to maintain. Any changes or improvements you make to your reusable data items will only further benefit all of the web pages that refer to them. 

Not only do we want to help you create a knowledge graph of your own data, but we want you to be able to connect with other knowledge graphs as well.

This is why we advocate connecting your data items to terms from Wikidata. Wikidata can be accessed to help provide more specific definitions about your schema markup. Moreover, by linking to other knowledge graphs like Wikidata, your content can inform and be informed by other structured data on the web. Thanks to this collaborative effort, new knowledge may be inferred and accessed through semantic search.

Conclusion

The term Knowledge Graph has a history of being notoriously difficult to define, and like many other terms in the domain of structured data, the definition changes depending on who you ask.

At Schema App, we understand a knowledge graph to be a collection of relationships between things defined using a controlled vocabulary, from which new knowledge may be gained by way of inferencing. It isn’t a simple definition. If nothing else, it reflects the fact that all of the domains contributing to knowledge graphs are complicated: theoretically, technically, linguistically, mathematically.

There are as many approaches to knowledge representation as there are perspectives on the nature of reality which means that current and future use cases are only limited by our own logic (see the recent DeepAi article A Survey on Knowledge Graphs: Representation, Acquisition and Applications for more information).

But just because the architecture contributing to meaning-making is complicated doesn’t mean the tools have to be. Schema App takes care of the technical aspects so that you can take full advantage of what structured data can do without getting mired in the weight of the work.

a knowledge graph [is] a collection of relationships between things defined using a controlled vocabulary, from which new knowledge may be gained by way of inferencing.

The dream of the Semantic Web was one of connecting domains, things and people, and making human information understandable for machine learning. In engaging with schema markup, you are contributing to a global collective intelligence whereby data can be transformed into knowledge.

Understanding the significance of this work, and the possibilities it allows, is the driving force behind Schema App. We want to help people be understood so they can be connected with the right people, at the right time, through the right channel. Yes, at its most basic level we want to help you make your content more findable to increase interaction and engagement, but more than that we want to facilitate the process of reflecting what exists in our world to the machines that connect our global village.