Breaking through the glass ceiling of NLU with Upper Ontology

What is an Upper Ontology?

An established approach to transfer knowledge in a computer-readable format is the use of ontologies. Ontologies are graph-based structures that represent knowledge in a similar way as taxonomies do. You define relationships of concepts and definitions. Often an “is-a” relationship is used to show the connection between different words. So, a wide network of relationships is created. These relationships create meaning of known things. A daily encountered ontology is the google knowledge graph.

Here we can see knowledge covering 570 million entities and 18 billion facts.

The concept of an ontology is used widely for different domains. Each ontology covers a specific field of interest. And you can find a lot of ontologies on the internet. They cover science and research as well as day to day communication like news, blogs, and wikis. So a wide variety of knowledge is available in the form of ontologies.

Image, Semantic Interoperability with Upper Ontology:
The Foundation Ontology as a Basis for Semantic Interoperability Patrick Cassidy MICRA, Inc., Plainfield, NJ

The basics of Upper Ontology

An Upper Ontology provides the necessary semantic interoperability to infer meaning across multiple domains. Interoperability can be achieved by offering general concepts that provide the most fundamental understanding of our world. This is not an easy task, as schools of thinking and perception of order are different amongst scholars and practitioners. However, in general, there is a common understanding that the being itself and the world consists of categories and individuals, time and space, and objects and processes. At this foundational level, the discussion quickly turns into a philosophical one, which leads to disagreement on details or fights between schools of thinking. Often questions to describe useful aspects of the Upper Ontologies help to understand the school of a specific upper ontology.

Image, Outline of categories of the DOLCE Upper Ontology
Ontology Engineering Lecture 6: Top-down Ontology Development I By Maria Keet,

Individuals and their categories

When talking about an upper ontology, also called top-level (or meta) ontology, or upper model, one main point is the definition of categories. As a precise definition of what a category seems to be a complex endeavor, upper ontologies focus on the distinction between categories and individuals. Individuals, also called particulars, are defined as concrete, spatiotemporal entities. They could be seen as persons, material objects, or events that have a definite place in space and time. In addition to that, things that could be identified through individuals are also individuals themselves. These are abstract things like thoughts or physical expressions of emotions like smiles.

All things are in time and space

Our perception of the world is defined by the continuity of time. We experience the past, present, and future. This reflects in our language. For an understanding of time, it’s important to define time as an interoperable concept. So, different expressions of time in domain-specific ontologies can be aligned. In this way, interoperable artifacts like schedules expressed in natural language can lead to an unambiguous understanding of time. Here two different convictions prevail. One is the continuity of time in the sense of intervals; the other one is focused on points in time. With this foundation, a unified time concept is created through an Upper Ontology.

The existence of Objects and Processes

When we look at natural language understanding, the understanding of objects and processes is quite important. Things change over time, they move from one place to another. And we need to keep track of that to understand. Things are falling down, and they obey physics laws. People change their names when they get married, and they still remain the same. Titles are awarded, and products and companies get new names due to marketing.

It’s not easy to decide which Upper Ontology is the right one

Upper ontologies help NLU in offering universal general categories for semantic interoperability. They define concepts that are essential for understanding meaning. But the challenge we are facing is the multifaceted offer in Upper Ontologies. There are Descriptive Ontologies for Linguistic and Cognitive Engineering (DOLCE), Basic Formal Ontology (BFO), General Formal Ontology (GFO), Suggested Upper Merged Ontology (SUMO), and the list goes on.

Image, Upper Ontology architecture for NLU:
The Foundation Ontology as a Basis for Semantic Interoperability Patrick Cassidy MICRA, Inc., Plainfield, NJ



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Constantin Kogan

Constantin Kogan


Dad, truth seeker, sociologist, smart securities & blockchain enthusiast. Researching and contributing to the sharing and value economies 🚀