Most have been in this situation before: one of the providers or services we use is a victim of a data breach and we want to determine if our personal user data has been impacted. This is where fully homomorphic encryption (FHE) comes into play. With FHE, the encrypted, personal password is compared against the data set of stolen user data and potential matches are identified without ever revealing the user’s password.

Use cases for this type of privacy-enhancing technology (PET) are numerous. They range from applications in medicine, where third-party service providers can analyze health data without compromising a patient’s privacy, to performing machine learning and AI algorithms on encrypted data, allowing organizations to derive insights from sensitive data sets without exposing the data to potential breaches or privacy violations.

How It Works

Fully homomorphic encryption allows calculations to be performed on encrypted data without having to decrypt it first. Confidentiality is maintained, as even the results are encrypted and can be viewed only with the appropriate decryption key. Further techniques for processing encrypted data are multi-party computation (MPC) and trusted execution environments (TEE).

Mathias Kohler, research manager at ���Ĵ�ý Security Research, outlines the differences: “While FHE is the most known of the encryption technologies, MPC is the ideal candidate if working with several parties exchanging encrypted data across company borders. And it can be substantially faster than FHE.” While both are software-based technologies, TEE is hardware-based, which makes it the fastest choice. The downside: TEEs, unlike MPC and FHE, require decrypting the data for processing. While decryption happens in a trusted hardware environment isolated from the operating system, it can allow data leakage via side-channel attacks. Notably, PETs do not need to be considered in isolation and can augment each other. For example, MPC can encrypt and distribute an FHE decryption key, protecting the FHE key and ensuring no single party can decrypt everything.

Why It’s Relevant

There is a demand for this kind of technology. By 2025, 60% of large organizations will use at least one privacy-enhancing computation technique in analytics, business intelligence, or cloud computing, according to .

Fully homomorphic encryption has numerous applications, especially in scenarios where privacy and security are paramount, such as secure computation in the cloud, privacy-preserving data analysis, and secure outsourcing of computations. As long as one party is performing the data processing centrally, FHE is the encryption method of choice. FHE enables organizations to share encrypted data with partners or third parties for analysis or monetization purposes while maintaining data confidentiality. This is particularly relevant in industries such as advertising and market research.

Interesting use case scenarios from ���Ĵ�ý’s perspective could be secure benchmarking and predictive maintenance.

Secure Benchmarking

Companies often assess their competitiveness relative to industry peers and compare business-relevant KPIs, such as automation rate or return rates, with peers and even competitors. With fully homomorphic encryption, all participating parties can share encrypted KPIs without revealing individual data. As a result, they learn about relevant statistics, such as averages or medians, to assess their relative competitiveness and decide where to improve and invest.

Predictive Maintenance

Predictive maintenance is a machine learning technique to forecast demand for maintenance or spare parts based on historical data. “In certain industries, required data, such as usage patterns and failures, is considered sensitive and is not easily shared with data scientists or maintenance operators,” says Anselme Tueno, senior researcher at ���Ĵ�ý Security Research. By computing on encrypted data, however, no sensitive information is revealed while still allowing for the required insights to be gathered for prediction tasks.

Carbon Footprint Calculation with Multi-Party Computation

While it is early days from a product availability perspective, ���Ĵ�ý is working on potential use cases with customers and partners. One key example is calculating carbon footprints of products.

Prime examples for complex collaborations are today’s supply chains, intricate networks that encompass various levels of suppliers, manufacturers, and processed goods. Unfortunately, there is often a lack of comprehensive visibility across the entire process – either for technical reasons or because businesses are often reluctant to share sensitive data across supply chains that often include direct competitors.

However, to accurately assess and disclose a product’s carbon footprint, sensitive production details and associated carbon costs for production-relevant parts and materials are required. Here, MPC can reveal only the required carbon footprint without disclosing associated, proprietary manufacturing details with other supply chain participants.

Currently, ���Ĵ�ý is working with Bosch on cloud-native software for secure multi-party computation called .

“���Ĵ�ý participates in this open-source project and supports the development of Carbyne Stack’s storage and processing services and the deployment of Carbyne Stack on Amazon Web Services (AWS),” Kohler explains. “For Bosch, Carbyne Stack is a type of cloud-native operating system for MPC workloads that manages resources to run as efficiently as possible in multi-cloud deployments.” This effort can help ���Ĵ�ý in the long run to integrate MPC as technology into ���Ĵ�ý solutions and services while running in a cloud-native environment.

What’s Next?

Despite all the benefits around processing data, encryption introduces significant computational overhead due to the complexity of performing operations on encrypted data. Slow processing speeds, especially for complex operations and large data sets, makes fully homomorphic encryption impractical for real-time applications or large-scale data processing. Although the performance of FHE has greatly improved in recent years, its practical adoption is still limited due to the processing overhead and performance considerations. Ongoing research is focused on the design of FHE-specific hardware accelerators.

“PETs for computing on encrypted data have the power to amplify data-driven business collaborations and reshape the future of cloud computing,” explains Jonas Böhler, senior researcher at ���Ĵ�ý Security Research. By safeguarding data, they enable access to previously untapped information while minimizing privacy risks and thwarting data breaches. The future of computing is encrypted.

Someone who seized this opportunity is Jonny Rowland. Having spent a big part of his childhood in Uganda, he saw the opportunity in the growing thirst for sustainable, high-quality coffee – and an opportunity to ensure that coffee farmers would receive their fair share in this business. Together with his sister Beth, he founded , a profit-for-purpose business that works directly with local farmers to achieve higher yields and the high quality coffee that is sought after on the global market.

With the support of local experts working for Agri Evolve, farmers have improved their productivity and increased income for themselves, their families, and their communities. The main idea of this social enterprise is to use digital technology to improve established supply chains for coffee cherries in the Rwenzori Mountains, one of Uganda’s key habitats for Arabica coffee plants, and share the latest agricultural practices with local farmers.

Click the button below to load the content from YouTube.

Always allow YouTube

A Long-Lasting Tradition: Coffee in Uganda

Uganda is one of the few countries in the world where coffee plants are native. The processed beans are an integral part of Uganda’s export economy, making the country one of the 10 largest coffee producers in the world. In recent years, Uganda has made a name for its specialty Arabica, which thrives in the local climate with humid days and cool nights. But Arabica coffee, a climate sensitive plant, is beginning to struggle as global warming shortens the cooler phases it needs to thrive. This requires adjusting the traditional ways of cultivating the cherries.

Over 1.8 million households in Uganda grow coffee, and coffee contributes nearly a third of the country’s export earnings, paying for critical infrastructure like roads, hospitals, and schools. While many families grow coffee, it was hard to make a living out of it in the past and motivation to produce high quality was low as prices on global markets wouldn’t justify the effort.

Joyce Birungi is one of the coffee farmers living high in the rainforest of the Rwenzori Mountains who registered with Agri Evolve as a supplier. Like most farmers in this remote area, she took care of the plants based on traditions passed down over generations. The cherries were often picked too early and only ripened while drying in the sun for longer periods, which caused a loss in quality and therefore lower prices. Climate changes, like increases in floods, droughts, and heat waves, put additional pressure on the traditional ways of farming.

“In the past, I used to pick coffee cherries of poor quality. So, I lost a lot of money,” Birungi said, explaining her challenges as a small local producer before working with Agri Evolve. “Transportation was a big challenge. I had no way of selling the coffee and I didn’t know about best practices in farming and agribusiness.”

Collins Kifula is a field coordinator at Agri Evolve and an expert for sustainability and quality. He works closely with the farmers, registering each farmer using a mobile app based on the ���Ĵ�ý Rural Sourcing Management solution and collecting data like the size of the farm and the state of the coffee trees. Using the mobile app saves time – time he can use to talk to the farmers and solve issues they might be facing or suggest improvements to their current setup. One important focus of his work is education on soil protection. With floods happening more frequently these days, farmers benefit from terraces and planting grass patches to avoid the fertile soil being washed away by rain.

As Agri Evolve is specialized in high-quality, sustainable coffee, certification is required. “We need to hold up to a certain level of quality and we do that by getting, for example, the . To receive this, we need to raise data and have the transparency. Here, the mobile app within ���Ĵ�ý Rural Sourcing Management is of huge help,” Kifula explained.

“Technology Is Changing Things”

The biggest challenges the farmers faced before starting to work with Agri Evolve was the quality of the cherries and the lack of price transparency. Sometimes there was even fraud along the process of collecting the cherries. “Technology is changing things,” Kifula said.

Today, farmers still deliver their cherries to so-called middlemen who check the quality, but they now document the weight using the ���Ĵ�ý app. The farmers instantly receive an SMS message showing the delivered quantity and its value. “This price transparency is a big motivation for the farmers and creates trust,” he said. Further, ���Ĵ�ý Rural Sourcing Management enables the calculation of the accrued delivery quantities and therefore the yield a farm is producing, which are used as the assessment basis to provide loans to the farmers.

Keeping Track of Data

“Our mission is to do sustainable farming for future generations. As a social enterprise, Agri Evolve works with a growing number of farmers, which can quickly add complexity to our supply chain. Today, we deal with around 22,000 smallholders. We collect a lot of data that needs to be analyzed,” Roset Biira, field supervisor at Agri Evolve, said. “We needed a system that collects and analyzes our data in an efficient way – to be able to be transparent to all our farmers and even our clients.” Using ���Ĵ�ý Analytics Cloud, Biira showed how Agri Evolve can identify if coffee trees are getting old or are weakened by pests early on. Based on these predictions, they consult with the farmers on planting new trees or suggest trainings on pest control.

Realizing Change

Agri Evolve is building a network of trust with the farmers and middlemen, making sure to give the best price possible to the farmers so that the local societies participate in the higher prices sustainable coffee can generate on the global markets.

“I’m now able to increase my coffee production and send my kids to school, and I hope that they will graduate from university and have better lives. We used to live in semi-permanent houses, but now we are moving to permanent housing,” Birungi said when asked what impact this new way of collaboration has had on her and her family.

Talking about the future of the growing profit-for-purpose business and its network of local coffee producers, Beth Rowland, co-founder of Agri Evolve, said: “We will continue to push the boundaries of supply chain transparency through digitalization to allow us to provide the best service to smallholder farmers in the Rwenzoris.” For the family business, it is important to ensure connected farmers have access to a fair and transparent market for their coffee cherries.

While sharing sustainable practices to counteract the impact of environmental changes is a requirement for the production of certified sustainable coffee, it will also make sure that the tradition of coffee production will live on in the Rwenzori mountains of Uganda.

It is the nature of artificial intelligence not to give any further importance to the emotional side of this explanation in the first place. Yet it’s precisely this emotional aspect that makes a discussion on the opportunities and risks of the metaverse or virtual realities so complex. Opinions here can vary from extreme enthusiasm to outright rejection. Some people are avoiding the topic altogether because they don’t know much about it or simply aren’t interested; others see massive potential and opportunity.

���Ĵ�ý’s latest art exhibition – The Metaverse: Dreamland or Dystopia? – shines a spotlight on these differing standpoints. Thirteen artists each present their take on how reality and virtuality – and art and technology – are converging. Working in a variety of media, they playfully explore space and time, traversing the real, the augmented, and the virtual. You can choose to visit the exhibition in the real world at the ���Ĵ�ý International Training Center in Walldorf or you can experience it in a . The exhibition runs until September 1, 2023.

Alexandra Cozgarea, curator of the exhibition, takes a close look at the impact of virtual worlds and artificial intelligence on art. “Artists have always used tools to create their work, be it hammer and chisel, paint and brush, or, in the age of technology, graphic design software, algorithms, and artificial intelligence. Today’s rapid advances in technology create tools that enhance our physical and mental abilities. The possibilities for the future are endless,” said Cozgarea during her speech at the opening of the exhibition, stressing how important it is for ���Ĵ�ý to use innovative technologies responsibly.

But for Cozgarea and the artists featured in the exhibition, the deeper question is whether digital and physical art can coexist. Is art on Instagram still art or must art only exist in the real world? But then isn’t Instagram also real? “How we perceive the metaverse is not only a technical question, but one about our vision of a desirable future and our ability to shape it ourselves,” added Cozgarea.

Her intent was to create a hybrid exhibition, grounded with one foot in the real world and the other in the virtual world – two places with blurring borders. The works impart their thoughts on how we as a society want to define ourselves as humans and how machines fit in.

During the opening of the exhibition, Jörg Kraus, who was instrumental in putting together the exhibition and is himself one of the artists featured, gave a speech about the metaverse, in which he connected current artistic viewpoints to all the ambivalent questions about the relationship between man and machine.

In his view, whether the metaverse is a dreamland or a dystopia comes down to our beliefs and our mindset. “Digitalization has already changed our lives and it will continue to do so at high speed,” he began. “Are you ready to contemplate the impact this might have on you?”

“As artists, we wonder whether the abundance of images that our society is already exposed to will continue to grow and level everything out. Will it really matter to anyone whether a work of art was created by a person or by a machine? Will we still have a notion of beauty and be critical of things – or will that all fade into the background?”

When asked where he stands on the metaverse, Kraus replied, “A key question for me is: where do I direct my attention when multiple elements are competing for it? We can focus on only one thing at a time.”

The Metaverse: Dreamland or Dystopia? is about art and therefore mainly about the artists’ perceptions and the aesthetics with which they express their thoughts. Images, sculptures, installations, videos, and a take us on a journey through the real and the digital, the two dimensional and the three dimensional.

When speaking to the artists about their work, it became clear that digital art, media art, or online art are challenging the notions of authorship and possession that underpinned the art market in the past. Virtual worlds not only offer artists new tools and opportunities, they also force them to face the ongoing discussion on how we as a society will use and consume creative content in the future. Has our society reached the point where the debate is no longer just about the pros and cons of technological progress? The exhibit suggests that we as a society should ask about the “deals” between man and machine, between man and economy.

So, is the metaverse a dreamland or dystopia? Even among the showcased artists, there is no single answer to this question.

A few of the exhibiting artists share their thoughts:

Michaela Schrabeck uses AI to create her work.

“For me, digital technology is a contemporary expression of human creativity. I’m drawn to machine learning and the fanciful artworks that artificial intelligence dreams up. Do the pieces I taught the machine to create pass as art and will they be accepted by my audience? Or must they be translated into traditional art forms?”

Paul Hirsch creates his sculptures twice: out of wood and in a virtual space.

“Didn’t virtuality and reality converge a long time ago? We already had powerful technical tools. Have we just added more – or do they represent a shift into a new and different world? Wrong questions – and much too passive. I think it depends on what we make of it. The space between the virtual and the real – and where the two overlap – that’s what I find so exciting.”

Christiane Rath builds “human nests” out of branches and leaves and invites the observer to make themselves comfortable inside of them. She reflects on what we might lose in the virtual world: the haptic experiences and smells that etch themselves into our memories.

“The virtual world is exciting and thrilling. It opens up a wealth of possibilities for new fantasy worlds and sensory adventures. Yet whenever I catch myself becoming too enthusiastic about the latest innovations, I think about what we might lose along the way.”

Jörg Kraus works with location data from Google Maps.

“The entire world is being rasterized and converted into virtually reproducible data. I’m relieved whenever I discover a gap – that even the “machine” misses something. That gives it a human-like quality. I’m worried, though, that the gaps are closing and that those fleeting escapes that we can still enjoy today will disappear.”

Image copyright: Klaus Kirchner
Disclaimer: This article is not about technological details, nor is it an assessment of artificial intelligence, virtual realities, or the metaverse. Instead, it contains different responses to the questions that using these technologies has raised – with a special focus on how they affect artists.

Even as far back as the 1950s, computer scientists were already experimenting with this concept of data modeling. In the 1980s, knowledge graphs were a much-discussed topic in the context of expert systems, and in the 2000s they once again came to the attention of the scientific community, as they took on a fundamental role in Semantic Web. Finally, in 2012, it was a search engine that gave knowledge graphs their big appearance among a wider audience: Google announced that its users would be able to search for “things not strings.” Rather than just trying to match keywords from a query, the search engine would also place them in the right context – all with the aim of delivering better, more intelligent results.

“Put very simply, knowledge graphs are a technology that seeks to turn data into machine-interpretable knowledge,” says Michael Burwig, innovation engineer at Potsdam. Their key focus is to model the relationships between objects, which are shown as a network of interconnected points.

“Knowledge graphs allow us to map how we humans understand the world – how we go through life, accumulate knowledge, and how we contextualize that information in our minds,” Burwig explains. This ability allows humans to draw conclusions that produce fascinating “aha” moments. Ultimately, this is also the goal of business software: to connect knowledge and find solutions, ideally in an automated way. There are many potential use cases of knowledge graphs, such as knowledge and data management and chatbots. Detecting insurance fraud is one example, but, in short, they can be used in any scenario in which patterns are examined to identify exceptions and anomalies.

The Potential of Hybrid AI

Burwig believes, however, that automated insights like these have not yet become part of everyday practice. “Hybrid AI is a major step in this direction. It combines semantic technology, in other words knowledge graphs, and static machine learning, which is now being used for a host of scenarios of this kind,” he says.

Dr. Jan Portisch, lead architect at Value Accelerator Delivery, describes the difference between classical machine learning concepts and hybrid AI machine learning as follows: “Classical machine learning methods use extracts of existing databases. These methods draw on such data sets, which are effectively snapshots without any context, to create machine learning content. With knowledge graphs, on the other hand, new data can be added at any time so that they keep on ‘learning’ and, unlike static methods, stay current.”

Even though knowledge graphs are not new, it took the massive increase in computing power that we have seen since the 2000s to unleash their full potential, Portisch explains. And since this technology is not yet widely taught at universities, few developers know much about it. “Another difficulty is that designing graphs is highly complex, with developers having to think beyond their own application. The modeling behind the graphs has to be complete and semantically accurate. Nonetheless, their potential is huge,” he says.

As an architect, Portisch is involved in creating a central graph for process knowledge at ���Ĵ�ý. The ���Ĵ�ý Signavio Value Accelerator Delivery team collates and models the process knowledge ���Ĵ�ý has built up over the past 50 years to give customers an integrated view. In the future, the accelerator will offer a semantic search where users will be able to formulate a business problem they’d like to solve and the system will display a list of processes and data that are impacted by this problem, Portisch explains. Further, this capability could be a helpful transition tool in migration projects or could provide presales teams with the visibility they need when tailoring solutions to customers’ specific situations.

Felix Sasaki, expert in Knowledge Graphs & Semantic Technologies in the ���Ĵ�ý AI unit, explains additional benefits: “Standards-based knowledge graph technologies facilitate the modeling of business scenarios. So-called constraints complement the existing logic-based modeling. Since constraints can easily capture the knowledge of business experts, modeling becomes easier. In addition, knowledge graph-based vocabularies like schema.org have found widespread adoption and thus help to find a more ubiquitous language.”

Decoupling Business Expertise and Application

For ���Ĵ�ý, the power of this technology lies in how graphs can be combined in different ways and, therefore, in how data models – and ultimately the applications themselves – can be integrated and composed.

Portisch explains how it’s important to recognize that RDF – Resource Description Framework, a syntax used to model metadata for Internet resources – is a publicly acknowledged standard and that knowledge graphs are already widely used outside the corporate world. One of the best-known knowledge graphs is ; another example of a large-scale graph is . “These graphs are public resources that can also be used commercially,” Portisch says.

That creates interesting use cases. For example, private corporate data could be combined with public data. A supplier system running on ���Ĵ�ý software could automatically import metadata from a graph about companies it interacts with, such as the type of business, company logo, and who their managing directors are.

“Semantic Glue” – Integration by Design

Burwig sees another advantage of these graphs in their “semantic glue.” Because of their flexible structure that can be enhanced in real time, individual graphs can be “glued” together to link up data silos. Unlike graphs, the table structure of a relational database has to be defined at the beginning – and changing that structure later requires a lot of work. But graphs can store and link metadata in a semantic data layer that is separate from the data stored in an application’s tables. This makes it possible to consolidate data across different products and data silos.

Graphs offer huge advantages over relational databases in certain scenarios, such as calculating the shortest route or associating a specific material with a customer material. “If such a scenario is based on a relational database, the query would have to touch applications from across the entire ���Ĵ�ý world. But with a graph, it would be a simple problem to solve,” Portisch says. While relational databases usually offer greater performance within their applications, graphs always have an advantage when it comes to creating bridges, known as joins, between data structures and recognizing the correct context.

Because it is possible to extend knowledge graphs, they represent a perfect data model for businesses to start small with and then, as an iterative process, to roll out to more locations and build on. For example, a company could begin with just one use case or department and then gradually roll the model out to the rest of the organization. Any additional data models are “glued” to the graph as new nodes and edges.

The Situation Knowledge Graph: One of the First Knowledge Graphs at ���Ĵ�ý

One of the most advanced knowledge graphs to date at ���Ĵ�ý serves as the basis for the Explore Related Situations app, which is part of the Intelligent Situation Automation service on for situation handling in ���Ĵ�ý S/4HANA.

The business background: around 4% of the automated business processes within a business require manual intervention because of an unforeseen event, such as a late payment, a delayed delivery, or a transport issue.

The situation knowledge graph links these exceptional events to their business entities, enabling the user to better understand the situation in a business context and therefore helping solve the problem and optimize business processes. Further analyses of these situations often reveal relations of issues to certain materials or partners. “Today, this kind of knowledge is often hard-coded,” says Dr. Torsten Leidig, an architect in ���Ĵ�ý’s Situation Handling team. The knowledge graph that underpins the Intelligent Situation Automation service enables a business expert or a key user to model processes and understand them within a comprehensive business context. A problem that has been detected can be resolved automatically based on simple rules and without additional programming.

Dr. Knut Manske, engineering lead for Situation Handling and Responsibility Management at ���Ĵ�ý, describes the graph as a layer that stretches across various ���Ĵ�ý applications. To summarize the capabilities of situation handling, he explains: “The knowledge that is embedded in ���Ĵ�ý applications is extracted and made usable for algorithms. Situation handling runs alongside the application, analyzing data and reacting to information about data changes or events. The aim is to show solutions that span various applications or business areas.” Customers and partners can define situations themselves without ever touching the application. A selected group of customers is currently validating the solution.

Looking Ahead

Currently just a prototype, the ���Ĵ�ý “Business Decision Simulator” innovation project simulates the effects of internal and external factors on a company and presents the user with potential future scenarios and recommended responses. “One possible question could be: what does a bushfire in Australia mean for my global pharmaceutical company’s value chain and thus the achievement of our targets in Europe?” Burwig explains. “On the one hand, knowledge graph technology can help express the complex relationships between real-world events and business-related processes. Additionally, an extensive knowledge base makes it easier for the user to choose the best possible response to opportunities and risks.”

When asked about the possibilities of graph technology, Burwig says: “The potential of knowledge graphs – at least that’s the dream – is that someday there will be extensive graphs modeled on deep knowledge and experience that can be accessed by programs. This means that with every change in technology, we’d only have to recode the application and not the knowledge stored inside it.”

While the decoupling of data from applications and the accompanying scalability of solutions and programs are important aspects for an IT company, Burwig also sees a vision that goes beyond the ���Ĵ�ý context: symbolic AI, artificial intelligence that is linked to a large number of interdisciplinary knowledge networks, could have the potential to generate new knowledge that has not been explicitly modeled. Linking scientific content could have the power to accelerate the discovery of new drugs or forge innovations between different science disciplines that don’t interact on a large scale today.

In the history of humanity, things have often been – and still are today – “invented” multiple times, at least in part from lack of knowledge about other relevant findings. Connecting global research projects within one given discipline has increased sharply within the past years, thanks to the availability of extensive digital content. Uniting scientific knowledge across the individual disciplines – for example, mathematics, biology, medicine, and chemistry – could uncover new insights and lead to exciting interdisciplinary inventions, Burwig concludes.

The Virtuous Circle Between Knowledge Graphs, Machine Learning, and Natural Language Processing

Academic communities of symbolic AI and machine learning in the past had few touchpoints in terms of research methods and researchers. This is changing – among others leading to so-called hybrid AI. This results in various virtuous cycles in which knowledge graphs, machine learning, and natural language processing (NLP) cross-pollinate each other. Johannes Hoffart, head of the Chief Technology Office in the AI Unit at ���Ĵ�ý, explains the relation between machine learning and knowledge graphs: “Knowledge graphs enable data scientists to work with complex and heterogeneous data sources. Their flexible schema can be easily extended and contains powerful data validation capabilities. At the same time, knowledge graphs facilitate access to and exploration of data, as they represent data and its schema such that it’s easier for humans, and also large language models, to understand.”

Christian Lieske from ���Ĵ�ý’s Language Experience Lab adds about the relation to natural language processing: “Knowledge graphs can feed NLP and can be fed by NLP. Take the detection of new business entities as an example: a knowledge graph can inform NLP about known entities and NLP can add additional entities to a knowledge graph.”

To get further insight around knowledge graph technology, read this study co-authored by Portisch: .