“Learning the language of data technologies for agile operations”

© 2023 EPFL

© 2023 EPFL

Dimitris Kiritsis, Professor Emeritus of Information and Telecommunication Technology for Sustainable Manufacturing at EPFL, points out the power of shared language around data technologies for agile and resilient operations.

For today’s technologies to effectively promote resilience, agility, and sustainability in our value chains, the data we have to gather and analyze is just one piece of the puzzle. Data must then be interpreted to become information and given shared meaning across its owners and users to become knowledge.

In part 1 of this series, one of the insights Kiritsis invited us to reflect on is that data is more than just a set of organized and exchanged information. He argues that data has meaning, meaning that requires trust and common understanding produced by shared language and communication between its owners and users. In short, the potential of data as an enabler for sustainable resilient value networks can only be realized if the people who wield it can build a trusting relationship with it and one another. In this second extract from our interview, Kiritsis takes us deeper into the world of connectivity & the human factor in data management, industrial AI and cognitive digital twins.

Semantics Matters as Much in Data as It Does for Human Communication

Trust is an important factor in any fruitful stakeholder collaboration. Collaborations built on data are no exception. In Kiritsis’ words:

“There are two dimensions of trust in today’s data-driven business operations landscape. The first one is the obvious technical one, which deals with questions of cyber security, encryption and server protection. The second is one of human trust. Both are linked however, by what my research is focused on: semantics.”

The power of semantics as an agent of trust between people is more apparent. We require shared language to converse with one another and build mutual understanding. The language we speak in meetings or with business partners, as well as the technical language with which we converse about sector specific topics, forge deals, and run our value chains. Kiritsis points out that such language is required when dealing with interoperability of data and connectivity across businesses. Much like with spoken language, we see shared vocabulary emerge, standards and protocols smoothing the path towards advanced analytics, traceability, circularity, and collaboration.

There is another parallel between digital and human systems Kiritsis shared with us:

A lot of our wellbeing is based on information, guiding us in taking decisions in a given context. Our access to data needs to be combined with a capability to analyze it, understand its meaning, and make it explainable. This is one of the issues with artificial intelligence. The data it provides is like a black box. It is informative, but it cannot explain why the information it provides is correct. If we can explain a thing, we can trust it, right?”

Our CAS in Value Chain Data Technologies was created to empower teams with the language needed to become translators between Operations requirements and Data Technology opportunities to drive agile and traceable operations through system-wide connection.

Cognitive Digital Twins and Value Networks

There is a lot of hype around the digital domain, its ability to visualize data and run simulations for us. Kiritsis reminds us that we should look beneath the hype to understand the true value of these technological advances. This is what he has to say regarding his field of research, the (cognitive) digital twin:

The digital twin allows for closer communication between the virtual world, where we do predictions and simulations, and the physical world in which we live. It is different from normal simulations, because it permanently connects an object or process in the real world to its virtual counterpart, constantly sharing information back and forth, constantly analyzing. One step further is what we call the “Cognitive Digital Twin”. It is a collective of digital twins across, for example, a supply chain process, all capable of interacting with one another in a complex system. This system is able to develop cognition, develop reasoning, and achieve autonomous decision-making.”

Kiritsis further illustrates this point with the following example:

Some time ago, a company came to me with a question. They wanted to learn about digital twins and needed a model to manage the supply chain for a particular product, from design to delivery, point of sale, and customer interaction. This is a complex system with many stakeholders. They would have a digital twin for production, transportation, and delivery to retailers, including forecasting on what, when, and how to deliver to the shop. To run this process successfully, we need access to all the different sets of information provided across the entire chain, from all the different digital twins. This step forms the cognitive digital twin, a platform that allows us to manage and combine individual digital twins for key elements, analyze them, create graphs and make recommendations for the final decision-maker.”

The semantic models that Kiritsis described earlier in this interview are the key to smoothing the connections between these digital twins, allowing information to flow among them and to be read out uniformly across the chain. What these semantic models and enabled cognitive digital twins will furthermore allow us to do, is move away from thinking of our processes as chains, and start operating them as networks.

In a chain, one element is linked to two others maximum. Remove that element, and the chain breaks. In a network, each element has multiple connections to others. While this more accurately depicts the reality of the value networks we operate in today, it also makes managing them more complex. A complexity that we are seeking to manage by developing cognitive digital twins connected through shared semantic models.

Companies aiming for traceable value chains and seamless customer experience must recognize their responsibility to provide learning opportunities for their Operations professionals. They are the ones best equipped with the business knowledge to design and adopt the right data technologies to fit their particular value network business needs.