# What is the place of ontologies in artificial intelligence?

Very much a general question here, from a somewhat uneducated perspective.

I'm currently part way through an MSc in AI, and at the minute I am taking a module on Knowledge Engineering and Computational Creativity. The professor taking the class obviously does research in this area and is saying that ontologies are becoming very important in the world of AI, or it may be more accurate to say he is suggesting they are becoming more important.

I intend to look into the work he does, and ask him a few questions, but, generally, I was wondering where this type of research sits in the world of AI. Is it something being worked on a lot? Is it becoming bigger?

I am interested because I do find the topic interesting, and I will have a research project coming up soon, and while I do want to work on an interesting topic, I also want to work on a relevant topic, so any information would be great.

• Hello. Could you please focus on 1 specific topic (either ontologies, semantic networks, or logic, but not all 3 at the same time) at a time? I am asking you this so that people can focus on one topic at a time. Maybe someone only knows the answer to the question "Where are ontologies being used nowadays?" (which seems to be one of your questions, if I interpreted your post correctly; if not, let me know) and knows nothing about semantic networks. You can ask the related questions in a separate post and I would recommend that you do that.
– nbro
Oct 22, 2021 at 13:16
• @nbro - Hello! Yes of course, not a problem. I will edit my post to focus on a single topic. And yes, you have interpreted my question correctly. Thanks for the feedback too. Oct 22, 2021 at 18:11
• I found a definition: "In AI, an ontology is a specification of the meanings of the symbols in an information system." To me this relates to semantics and surely some area computational linguistics. It's in this view it's a technical subject, related to input and output, but it's also a deeply philosophical subject, and has been heavily treated in the speculative literature.
– DukeZhou
Oct 23, 2021 at 4:53
• An example of the philosophical dimension is the "Grecian Room" where two subjects who can't see each other converse in Ancient Greek. Neither subject can know if the other actually understands the language, or is just following a set of instructions to simulate knowledge. From a technical perspective, directed graphs are and n-dimensional network topologies are examples. (The Symbol grounding problem is an example of a technical challenge that is thought to have philosophical implications.) So I'd say ontology is pretty central to AI in both domains.
– DukeZhou
Oct 23, 2021 at 4:59
• Your question is somehow unclear/ambiguous. What exactly are you asking? Are you asking 1. "Where are ontologies being used in the context of AI?" or are you asking 2. "Is research on ontologies being done in AI, and for what purpose?". I guess these questions are related, but "place of ontologies in AI" to me seems ambiguous and your questions in the body of the post suggest that you're only interested in whether people in AI are doing research on ontologies. So, please, edit your post to clarify your question.
– nbro
Oct 23, 2021 at 14:32

Computational creativity is not something I know anything about. However, I work in knowledge engineering. This falls into the areas of knowledge representation and reasoning known as semantic web technologies in general. More recently Google has popularized the name "knowledge graphs" and now many people tend to talk of knowledge graphs rather than the semantic web, even though, strictly speaking, knowledge graphs are a small subset of semantic web technologies.

This is a massive field of research, which is widely used in bioinformatics, healthcare, property management, and many other fields to enable semantic search. I myself work in bioinformatics where we use 265 ontologies spanning close to 7 million concepts that are used to enable semantic searches across around 300 petabytes of data. In fact, it is knowledge graphs that are at least in part used to enable Google searches and information provided Google info boxes. Hence, many people are already using the results of knowledge engineering without knowing it.

So what is an ontology? An ontology defines concepts and relations between concepts. This has been done in computer science for ages, so what makes ontologies and the semantic web so special?

• Each concept and relation is assigned globally unique identifiers, e.g. URI, IRI, PURL
• Ontologies are defined in a machine-readable syntax, e.g. XML, JSON-LD
• The semantic web defines a generic data model able to describe arbitrary data called RDF triples. In particular, this helps people to define their data before having to define a schema (as is the case with relational databases).
• An RDF triple <s, p, o> expresses that subject (s) and object (o) is related via predicate (p).
• SPARQL is a query language for querying RDF triples
• Ontologies are equipped with formal semantics based on mathematical logic, which enables artificial intelligence reasoning procedures to infer implicit knowledge from explicit knowledge. E.g. RDFS and OWL.
• JSON-LD, RDF, SPARQL, RDFS and OWL are W3C standards.

Here is a book that can give you a reasonably gentle introduction to aspects of the field and how it is used in practice: The knowledge graph cookbook.

• Hi! That is a fantastic answer and does help a lot, and gives me other things to look into. I don't have the reputation to upvote your answer, but when I do, I will. Thanks for the reply and the information! Oct 23, 2021 at 11:33
• Hello. Thanks for contributing to our site! Your answer seems good. You may be interested also in this question.
– nbro
Oct 23, 2021 at 14:36