Vojtěch Svátek and Ondřej Zamazal: Ontologies Support Flexible Data Exchange – Something the Circular Economy Also Needs

Vojtěch Svátek and Ondřej Zamazal, experts in ontological engineering from the Prague University of Economics and Business, have joined an ongoing Horizon Europe project called Onto-DESIDE with their team through the so-called hop-on mechanism. Their expertise in ontology development and transformation is expected to help advance the circular economy by enabling more flexible data exchange between partners in circular supply chains, which ontologies facilitate.

What personally attracts you to the field of ontology? Why did you choose this discipline?

Zamazal: Ontologies caught my interest because they are a versatile tool for knowledge representation in artificial intelligence. Moreover, ontologies can be used in a simple form as well as in a complex form with reasoning capabilities, making them applicable in many areas. Today, AI is dominated by generative approaches, particularly large language models. However, knowledge representation, ontologies, and knowledge graphs still have a lot to offer.

Svátek: I would express my connection to ontologies in an even more elevated way. We have all encountered the term “ontology” in a philosophical context, where it refers—very simplistically and imprecisely—to the study of “being”—what constitutes the world, what types exist in it, and so on. And research on ontologies in computer science—if we are able to perceive it—retains many of these profound philosophical aspirations. Specifically, in our research, we deal with questions such as when a given concept should be understood as an “individual,” something that exists uniquely, and when it is instead a “type” concept that includes multiple instances. Or whether a certain property of an object, such as being a product, is a permanent characteristic, or whether the object can lose that property—such as when it becomes raw material for the production of a new product, even though its external form remains unchanged.

For me, our specialized subfield of computer science, known as ontological engineering, is like the “queen” of computer and engineering disciplines. In addition to engineering approaches (which are also necessary—we work with ontologies as data structures encoded in formal, machine-processable languages), it also extensively incorporates mathematical-logical and philosophical aspects.

This doesn’t mean that ontology is merely an intellectual exercise for the satisfaction of a contemplative mind. Its real-world impact can be significant. Allow me to borrow an example frequently used in lectures by one of the most renowned researchers in the field of information ontologies, Giancarlo Guizzardi from the Netherlands. After the 2001 attacks on the World Trade Center in New York, a legal question arose, the answer to which determined an insurance payout difference of several billion dollars: Were the impacts of the two planes on each tower two separate insurance events, or a single event? Legal statutes and regulations alone did not provide sufficient grounds to resolve this question—ontological analysis involving concepts such as identity, unity, and causality was needed, whether explicitly or implicitly.

Ontologies and Ontological Engineering

Ontology as a discipline focuses on structuring and describing knowledge about the world—for example, identifying which objects exist, what properties they have, and how they relate to each other. In computer science, ontologies are used as advanced data schemas to clearly describe information so that both humans and computers can understand it. Ontologies allow machines to comprehend the meaning of data and process it efficiently. The goal is to create clear structures that facilitate knowledge sharing, system integration, and data processing—such as in the field of the circular economy, which is at the heart of the Onto-DESIDE project.

You are also working with ontologies within the Onto-DESIDE research project. What was the main impulse for you to apply for involvement in this project? What was your motivation to enter such a research project?

Svátek: We had known about the Onto-DESIDE project beforehand thanks to direct contact with its coordinator, Eva Blomqvist from Linköping University in Sweden. Nearly two years earlier, we had collaborated with her on an Erasmus+ project proposal, which, to some extent, had a similar focus—it aimed to enhance the teaching of ontological engineering methods, using the circular economy as a model domain. We assumed that for today’s students, particularly those with an ecological mindset, this could be an attractive and engaging way to learn. That Erasmus+ project was not approved—but we never imagined that we would soon enter the world of ontologies for the circular economy through another door, a scientific rather than an educational one! But when that European-labeled door opened slightly, it was only natural to reach for the handle.

You joined the project via the so-called hop-on mechanism. What led you to this approach, and what were the main advantages?

Svátek: I must admit that this mechanism was new to us, and when I first heard about it in general terms, I was rather skeptical. In an ongoing project (we had participated in several under Horizon 2020 and earlier in FP6 and FP7), partners already have defined roles, have shared a lot of knowledge, and have gone through complex procedures related to grant agreements and consortium contracts. Adding another partner inevitably involves significant administrative overhead, and I assumed that existing partners would rarely be keen on such additions. However, when our university’s project support team independently identified possible Horizon Europe projects for hop-on participation, and Onto-DESIDE was at the top of the list, it was irresistible!

Although I was not far off in estimating the complexity of the joining process (it took several months longer than expected), the challenges also became advantages. By the time we joined, the project was already beyond its halfway point, meaning that some dead ends had been explored and abandoned, and there was already a solid data foundation upon which our newly developed methods could be applied.

Hop-on Mechanism

The hop-on mechanism allows new partners to join ongoing projects funded by European grants, particularly under Horizon Europe. Its goal is to increase participation from countries and organizations traditionally underrepresented in these projects (so-called widening countries). This fosters broader collaboration, knowledge sharing, and more balanced research development across the European Union.

You already mentioned the collaboration with the Swedish university Linköping. What role did it play in integrating VŠE into the Onto-DESIDE project?

Zamazal: It played a key role, both in the phase of preparing VŠE’s accession proposal to the project and in the phase of engaging in activities. A big thanks for this goes especially to the project coordinator, Eva Blomqvist. Our prior contacts with this university, which my colleague already mentioned, definitely helped us significantly in joining the project.

Svátek: I would also explicitly highlight the excellent organization of the catch-up bilateral meeting they arranged for us. Right after the start of our participation in the project, the two of us traveled to Linköping, where we received a very warm welcome. We presented topics from our current research to colleagues from Eva Blomqvist’s team as well as to other senior researchers and established a concrete bilateral collaboration. Thanks to this, just half a year later, we presented the first tangible results of our collaboration at a European conference in Amsterdam—experiments with using large language models to generate names for newly created concepts in ontologies.

And what specific expertise does the VŠE research team bring to the Onto-DESIDE project?

Zamazal: VŠE has contributed expertise in the area of ontology mapping and transformation. Both of these specialized areas address a common challenge in today’s global environment: data published or processed by different entities often have different structures or use different terminology, even when they refer to the same subject matter. The first topic, ontology mapping—linking concepts used by different entities—was primarily represented in the project by the Swedish university, and VŠE’s involvement enabled mutually beneficial international collaboration. The second topic, ontology transformation—where we have a single system of concepts (ontology) but modify it to fit a new problem or to describe independently created data—was not previously represented in the project and is primarily the domain of the VŠE team.

Svátek: That is exactly what I consider key when joining an ongoing project: when we have significant overlap with our partners in some areas, we understand each other from the start (both in terms of terminology and the state of knowledge) and can work as a tight-knit team, while in other areas we are complementary and can mutually enrich each other. Specifically, in the case of ontology transformation, its necessity was recognized within Onto-DESIDE—after all, if ontologies are rigid in their structure and do not adapt to the needs of a specific application, some data may become difficult to access, or their presentation to users may become confusing. However, before our entry into the project, transformation activities were performed manually and unsystematically, without a formal methodological foundation. We already had some experience with automatic transformation based on formal models from a previous national project supported by GA ČR. We revisited these models for Onto-DESIDE, significantly expanded and revised them, and most importantly, incorporated an entirely new element that did not even exist during the previous project—large language models.

This, by the way, is something worth a special mention, especially for those interested in the future of artificial intelligence. Ontologies primarily belong to the conceptual world of symbolic artificial intelligence, which is often seen as an outdated competitor to AI based on deep learning and neural networks, particularly large language models. However, in the Onto-DESIDE project, as in many other projects worldwide, it is becoming increasingly evident that these two approaches complement each other well. I believe that an increasing number of AI-driven projects in the coming years will adopt some form of hybrid, neuro-symbolic solution.

Circular Economy

The circular economy is an economic model that aims to minimize waste and efficiently utilize material resources. Instead of the traditional “produce, use, dispose” approach, it focuses on reusing materials, repairing, recycling, and extending product life cycles. This approach promotes sustainability by reducing the need for extracting new raw materials, lowering environmental impact, and contributing to the conservation of natural resources. The circular economy connects environmental goals with economic benefits, helping businesses save costs while protecting and preserving the environment.

When we return to the topic of the circular economy, how can the Onto-DESIDE project and ontologies contribute to a better understanding and development of the circular economy? How specifically will the VŠE team help achieve this goal?

Svátek: The circular economy is an example of a system in which a large number of independent economic entities collaborate or compete. These entities exchange data—about products, materials, technological processes, etc. Value chains (traditionally referred to as supply chains) can be very long and complex precisely because of the principles of circularity. At the same time, there is often no hegemon among stakeholders who can impose a single data format on others, making the environment relatively democratic for now. These factors create favorable conditions for sharing data and metadata described through ontologies. Compared to strictly structured database models and data exchange formats, ontologies—used as schemas for graph data—offer greater flexibility and facilitate mutual understanding of different data formats used by individual entities. Participants in a circular marketplace for reused components and recycled materials do not need to fully agree on every detail of their data structures, yet they can still exchange information effectively, ensuring that the commodities being traded meet expected criteria.

The ontology mapping and transformation methods developed by the VŠE team represent some of the most advanced tools for such flexible data exchange. As I mentioned, ontologies—even without transformations and mappings—already allow a certain degree of adaptability in structuring data. However, if we also make the ontologies themselves more flexible, we can achieve data exchange that flows as needed, without being constrained by rigid, predefined schemas. At the same time—and this is where symbolic representation has an advantage over purely neural, numerical approaches—transformations are precisely documented. On the other hand, incorporating language models enhances user-friendliness and brings a degree of creativity when generating names and descriptions for newly proposed ontology concepts. Of course, there is a certain risk of errors, so-called hallucinations, but in this case, they can be mitigated more effectively than in many other applications of generative AI, as the textual sequences involved are relatively small and well-targeted.

How do you expect participation in the Onto-DESIDE project to impact the development of your research team’s expertise? What professional or scientific opportunities and benefits do you foresee from this international consortium partnership?

Zamazal: In general, international collaborations advance researchers both professionally and scientifically, and this project is no exception. Specifically, we have co-authored publications with Linköping University and co-organized an international competition in the field of ontology mapping. As we prepare for the next project meeting in Ghent, Belgium, hosted by another academic partner, IMEC, we also plan to engage in detailed bilateral discussions with their team, which could lead to further international joint publications. More broadly, our experience with managing our first Horizon Europe project will be invaluable in future efforts to participate in European research initiatives.

Svátek: I believe these international partnerships are especially crucial for our PhD students. No matter how many times we explain from past experience how such projects operate, nothing can replace their firsthand participation. Attending meetings, receiving assigned work packages, observing how foreign research labs function both organizationally and interpersonally—these experiences are irreplaceable.

If you were to assess the Onto-DESIDE project in terms of societal impact, what outcomes would you consider the most significant for society or the business sector? What will be the project’s practical contribution?

Svátek: It’s difficult to predict how soon the project’s results will be directly applicable to the business sector. Given the project’s short duration (and this applies even more to the research conducted by VŠE than to that of other partners), the main outputs will likely be partial technologies tested on small-scale case studies. The most practical application, I anticipate, will come from the circular economy ontology network, which—like many ontologies—can be used in ways we cannot fully foresee at this stage. As for our initial work on ontology transformation methods, we are still in the early stages. The robustness required for industrial applications is not yet there. Our presentations at the European conference in November attracted interest from several major corporations, but as an intermediate step, we would likely need to collaborate with a startup—either Czech or international—that could transfer these methods into practice more effectively than we could as a purely academic institution.

In such an abstract field, what does “transformation into practice” mean? How might the project’s outputs be relevant to businesses?

Zamazal: A practical impact could be cost reduction for companies, particularly in cases where businesses (such as actors in the circular economy) use different ontologies for their graph data and need to integrate them with other graph data structured according to different ontologies. Ontology mapping techniques provide a systematic, documented, and reusable solution. When the ontologies themselves change, ontology transformation techniques come into play, offering another cost-saving mechanism through systematic adaptation.

Do you have any final thoughts or challenges for readers, particularly regarding ontologies?

Svátek: I would like to take this opportunity to encourage early-career researchers in computer science not to shy away from ontologies as if they were something mystical and unscientific, but also not to confuse them with arbitrary data format specifications that anyone can invent without regard to reality. Ontologies are both rich and profound while also being intuitive to the way humans think. Analyzing the world through an ontological perspective is truly exciting!

Author: Žaneta Mrkvičková, with copywriting support from the editorial team of https://vedavyzkum.cz/

Ondřej Zamazal is an associate professor at the Department of Information and Knowledge Engineering at VŠE. His main research focus is ontology mapping and transformation. In his teaching, he specializes in artificial neural networks.

Vojtěch Svátek is a professor and deputy head of the Department of Information and Knowledge Engineering at VŠE. His main area of expertise is the application of ontologies and the “ontological way of thinking” in various fields of computer science, including data schema design, data and text analytics, and information visualization.