Category: Group Portraits

Smart and human-centered handling of information, knowledge and language

The Data and Knowledge Engineering lectureship of HAN University of Applied Sciences focuses on practical applications of existing knowledge. Projects include automating work instructions for SMEs and tackling energy poverty by providing insight into unstructured data.

Whereas academic research groups develop new knowledge, research groups in higher vocational education (HBO), called lectureships, make existing knowledge better and more publicly applicable. Stijn Hoppenbrouwers leads the Data and Knowledge Engineering lectureship at HAN University of Applied Sciences, a group of 12 researchers.

‘What binds the group together,’ says Hoppenbrouwers, ‘is a love of developing computational applications to deal with information, knowledge and language in a smart and human-centered way. For us, the question is how to apply existing generic knowledge in our teaching and in our collaboration with parties in practice. And then the focus is not on the leaders, which includes companies like ASML, but about the peloton, for example small and medium-sized enterprises and municipalities.’

A good example is the two-year RAAK SME project Flexible work instructions. This project investigates how work instructions can be made and used in a context-sensitive and flexible way. Hoppenbrouwers: ‘Suppose a heating technician needs to repair a boiler. What then is the best way to present a work instruction? From my lectureship, we are investigating how to generate work instructions automatically from various sources. The Media Design lectureship investigates how the information can best be presented so that it is usable for different types of boilers, for example. Ultimately, we want to offer SMEs a toolbox to support flexible work instructions.’

Another lectureship Hoppenbrouwers works closely with is Applied Science & AI. ‘That lectureship emerged from our group early this year, because it started to grow too big’, he says. ‘The new lectureship is more on the data analytics side. We are more on the linguistic side with knowledge representation, data quality and things like that.’

When it comes to data-driven work, there is still a big gap between what the big tech giants do and what, for example, SMEs do. To close that gap, the SPRONG project DEMAND was created, a collaboration between various institutes, lectureships and business partners, supported by the governing body SIA. The project is led from the HAN. ‘In this project,’ Hoppenbrouwers explains, ‘we can use our expertise when it comes to data quality, data availability and data management. DEMAND is also starting to act as a magnet, prompting external parties to come to us with their questions.’

Ambiguous concepts

Senior researcher Maya Sappelli started five years ago in Hoppenbrouwers’ group as a specialist in language technologies like text mining and information retrieval, in which she had gained her PhD. ‘What I like about our research group’, Sappelli says, ‘is that we have a lot of interaction with other parties. At a university, researchers are more concerned with just their own little research.’

For example, Sappelli is doing a project together with Alliander, the company that develops and manages energy networks. They want to improve the use of information within their own company. ‘Part of this is a conceptual framework: what terms are used within the company? And how do employees use those terms in practice? Take a concept like the Dutch term ‘aansluiting’. In Dutch this is unambiguous: a physical connection. At Alliander, however, they also use ‘aansluiting’ in the interpretation of a ‘customer’. Detecting those terms with multiple interpretations are important to make knowledge graphs and large language models more usable. Eventually, Alliander wants to build a kind of chat interface that allows employees to efficiently query the company’s data sources.’

Searching spider

The headquarters of the Data and Knowledge Engineering lectureship is at in the Arnhem part of the HAN. That’s where the researchers and lecturers meet physically, but a lot of work is also done from home or from workplaces in the Nijmegen part of HAN. ‘We are very flexible about that’, says Marijn Siebel, another senior researcher in the Data and Knowledge Engineering lectureship. ‘We have a lot of freedom in who and how we want to collaborate with, including with the other lectureships within HAN.’

As a specialist in query-driven knowledge systems, Siebel’s work includes the project SCEPA: Scaling up the Energy Poverty Approach. With energy bills rising in recent years, more and more European citizens have run into financial difficulties. Siebel: ‘Within the SCEPA project, we are investigating the smartest way to tackle energy poverty in a country, municipality, city or neighborhood. The project runs until 2027. By then, we want to have a well-grounded framework on how to look at energy poverty. Which information from which sources do we need? How can we make the access to this information easy and intuitive? I like to think of it as creating a spider that searches for you in a web of information. Ultimately, we want to have a beta version ready, which can be used to take better measures against energy poverty, tailored to the local situation.’

Group passport – Lectureship Data & Knowledge Engineering

Research fields

Data engineering, data management, information systems, text mining, knowledge engineering, knowledge representation, knowledge systems, knowledge extraction

Institution

• HAN University of Applied Sciences

Websites

• Lectureship Data & Knowledge Engineering
• SPRONG-project DEMAND
• Project ‘Flexible work instructions’
• Project SCEPA (Scaling up the Energy Poverty Approach)

By Bennie Mols
Images Ivar Pel

Developing high quality evolving software

The Software Engineering and Technology group at Eindhoven University of Technology investigates how software can achieve high quality throughout its lifetime.

The software industry faces significant challenges in the form of rapid software growth, declining software quality and increasing societal expectations. The Software Engineering and Technology (SET) group at Eindhoven University of Technology develops methods and tools for time- and cost-efficient evolution of high-quality software systems: from their inception, through development and maintenance, to phase-out.

‘Our group does not quite conform to the classic organisation of a research group’, Michel Chaudron says. ‘We have three full-time professors who form a kind of partnership in which the three of us make the important decisions.’ The trio consists of Chaudron himself, Professor of software engineering with a focus on software architecture, design and modelling, Mark van den Brand, Professor of software engineering who focusses on model driven methods, and Alexander Serebrenik, Professor of social software engineering, who focusses on social factors in developing and using software.

‘Software engineering has largely become an empirical study,’ says Chaudron, ‘which is why we think the link to practice is very important. Our group is good at collaborating with companies and we have a great connection to the industry in the Brainport Eindhoven technology region.’

An example of research that Chaudron himself is working on is recovering the architecture from its implementation and creating visualisations to help understand the architecture. Chaudron: ‘Software is actually an edifice of layered abstractions that continuously evolves. Practical software consists of millions of lines of code, nowadays even in several programming languages. If you want to know how that software works, you first want to understand its structure: what relationships exist between different components? That’s a big puzzle, and software visualisation helps solve it by making the relations between pieces of the puzzle better visible.’

The most important change facing software engineering in the coming years, and thus also the research in the group, is the introduction of AI, says Chaudron. In industry ChatGPT is already being used by software developers for many of their tasks. Chaudron: ‘AI is going to turn everything within software engineering on its head: from requirements analysis, creating designs, generating source code and debugging, to writing documentation. We know that AI in software engineering can lead to a major productivity improvement, but the big challenge is to ensure that this does not lead to a lower quality because we start to rely too much on AI.’

One of the ideas that Chaudron has developed together with fellow professor Jurgen Vinju to prevent this is to create a national or even international research infrastructure aimed at connecting different software analysis tools.

Complementing curiosities

Jacob Krüger joined the Software Engineering and Technology group as an assistant professor in September 2022. Krüger received his undergraduate and graduate training in Germany, and he notices the difference between the culture in the SET group and the academic culture in Germany: ‘It’s so much less hierarchical than I was used to. In SET it’s much more about collaboration. Michel, Alexander, and Mark, the three group leaders, are trying to create a group in which people can work on their own favourite topics but in such a way that all the topics complement each other. And everybody is involved in deciding the development of the whole group.’

Krüger studies human factors in software development, in particular economical and psychological factors. ‘A typical team of software developers consists of, let’s say, four to fifteen developers’, he tells, ‘and often you have multiple teams working in parallel. Potentially hundreds of people are involved, which creates a complex social environment.’

Improving software quality costs time and thus money, but it might save money by delivering more reliable software. By studying economics of software development, Krüger noticed how important it is to also look at how developers approach their work cognitively: ‘How do they understand code? What do they memorise and what not? One of the results that we have found, is that developers are good at memorising high level abstractions, architectures and features, but not at memorising lower level code. Based on these findings we try to build tools to support them in recovering what has happened over time with the software.’

Examine evolution

Lina Ochoa joined the SET-group in April 2023 as an assistant professor. She studies the evolution of software ecosystems. Ochoa: ‘Let’s say a team of software developers has released a software project, but after some time starts changing some features. A new version of the software is released and the changes will impact other teams of developers that rely on that project. One of the main challenges is to understand the people working on the software and the values they have.’

On a more technical level Ochoa develops tools that make software more robust against changes. Ochoa: ‘For example, when software libraries evolve, they incorporate new features like bug fixes and security patches. These changes might break the contract previously established with its clients. As a result, clients may hesitate to upgrade their software. An analysis tool might help library developers to understand and anticipate the impact of their changes.’

Like Krüger, Ochoa appreciates the non-competitive and collaborative environment in the SET group: ‘People are very good at what they do, but at the same time they are humble and willing to teach you things that you don’t know yet. I feel they provide the support that makes you grow.’

Group passport – Software Engineering and Technology group

Research fields

Model-driven software engineering, digital twins, software evolution and maintenance, human and social aspects of software engineering

Institution

• Department of Mathematics and Computer Science of Eindhoven University of Technology

Website

• Software Engineering and Technology

By Bennie Mols
Images Ivar Pel

Looking for smarter ways to test software

Software testing is paramount for many applications, but because of its difficulty and invisibility, it is not the most popular part of software engineering. Professor Tanja Vos wants to change that by developing more intelligent testing tools and building better educational modules.

The Software Testing group led by Tanja Vos is unique in spanning two locations in two countries: the Open Universiteit (OU) in Heerlen and the Universidad Politécnica de Valencia (UPV) in Spain. Vos has been an associate professor at UPV since 2003, and in 2016, she also became Professor of Software Engineering at the OU.

‘The Open Universiteit has place- and time-independent teaching education at its core’, says Vos, ‘and then it does not matter whether you collaborate with people working from different locations in the Netherlands or with people working in Valencia. The four PhD students I have at the OU all work from different locations in the Netherlands. In Valencia, there are four people in the group. We communicate with each other in the same way online. We use various chat groups for different purposes and, of course, we meet in person regularly.’

The research group of Vos is specialised in automated testing of desktop, web and mobile applications at the Graphics User Interface (GUI) level. ‘Our flagship is a testing tool called TESTAR,’ says Vos, ‘which we have continuously developed since 2010. TESTAR is a form of scriptless testing in which you do not have to write a test script in advance, but in which TESTAR decides which actions to select to create a test. We construct a very large collection of all possible actions in a given software state, and then we choose one of them to go the next state.’

In its default configuration, TESTAR only does random testing. This may sound strange but in practice, it turns out to be complementary to scripted testing. Vos: ‘We have tested software with TESTAR at several companies, including the mobile banking app of ING, and we found that TESTAR covers different parts of the software than human written tests do.’

Still, the question remains whether there are better ways to test than defining a script in advance or choosing random test actions. ‘Indeed’, says Vos, ‘that is why we are investigating if and how we can let TESTAR itself learn the best way to select actions. This involves using techniques from artificial intelligence. Another future research direction is to see whether we can use a ChatGPT type of tool trained on a large number of existing test scripts to predict the best test action to select next. We are constantly looking for smarter ways of testing software.’

Agents and models

Beatriz Marín is one of the senior researchers in the Valencia part of the Software Testing group. She was a professor in Santiago (Chile) for many years, and in 2021, Vos persuaded her to come and join the research group in Valencia as an expert in software quality. ‘During the past few years, I have worked on a project that combines the use of intelligent agents and models in testing’, says Marín. ‘An additional reason Tanja asked me to join the group is that I had experience using serious games to motivate students for software testing. Whether you work in a company or in academia, software testing is often considered difficult, and too few people want to do it. But of course, good testing is extremely important, so our group sees testing and quality as inseparable parts of the education in software engineering.’

Marín has just started working on a new three-year EU-funded project called ENACTEST, to improve the education of testing. The goal of this project is to identify and design seamless teaching materials for testing that are aligned with industry and learning needs. Marín: ‘We have four people from our research group working in ENACTEST. Nine different European partners are involved, including universities, vocational centres and small and medium enterprises. In the long run, this project will improve the software quality on which our digitalised society relies.’

Better education

As a PhD student of the Open Universiteit, Niels Doorn is researching how to improve testing education. Doorn works on his doctoral project three days a week; the other two days, he works as a lecturer and researcher in IT and computer science at the NHL Stenden University of Applied Sciences in Emmen. ‘The goal of my PhD project is to create effective methods for teaching software testing that are supported by scientific evidence’, says Doorn, ‘because at present, we know very little about what works well and what not.’

Doorn uses both a theoretical and a practical approach. For example, he compares how test experts test software to how students do that. Doorn: ‘What conceptual knowledge and experiences do they use? How can we best integrate testing into programming assignments without compromising the programming concepts we want to teach students? Among other things, we have already created a website with exercises ready for use by teachers.’

Doorn does the day-to-day supervision and most meetings online, but the group members also meet up several times a year. ‘In May, we had a computer science day where the group members got together. We conduct workshops for OU students. And we take occasions like the opening of the academic year, an inauguration or a conference to get together for one or two days. Although the group is spread across two countries, the collaborations are going really well.’

Group passport – Software Testing at OU and UPV

Research fields

Software testing, automated testing, software quality, education in software testing

Institution

The Software Testing group is spread over two locations: Open Universiteit (OU) in Heerlen, and Universidad Politécnica de Valencia (UPV).

Websites

• Prof. Tanja Vos
• Open Universiteit (OU)
• Universidad Politécnica de Valencia (UPV)
• TESTAR (Automated system testing of desktop, web and mobile applications at the Graphical User Interface Level)
• ENACTEST

By Bennie Mols
Images Ivar Pel

Data science for the common good

The Data Science group at Radboud University develops theory and methods for machine learning and information retrieval with a strong focus on social responsibility.

With the surge in the production of digital data and the explosion of machine learning applications over the past decade, it is no wonder that the Data Science group at Radboud University has grown significantly to some forty researchers. One of the group’s key characteristics is its strong focus on social responsibility in general and a strong connection with applications in the health domain in particular, the latter via close cooperation with the Radboudumc hospital.
‘Despite the growth of the Data Science group in recent years, we have decided to stick to our three core themes’, says group leader Tom Heskes. ‘We focus on causal reasoning for machine learning, biomedical applications of machine learning, and information retrieval and recommender systems. With the growth of the group, we considered whether we should split into more subgroups, but we decided not to do so precisely because there is a strong social cohesion running through the subgroups. Even though people are doing different things in terms of content, we do a lot of social activities together.’
One of the weaknesses of most machine learning applications developed and applied in recent years is that they are poor at reasoning about cause and effect, so-called causal reasoning. A machine learning application may conclude from data on smoking and lung cancer that the two are correlated but does not automatically understand that smoking can cause lung cancer. ‘Machine learning techniques are based on making associations’, says Heskes, ‘but because they are bad at causal reasoning, they don’t know what happens if you do an intervention, like banning smoking in public spaces.’ In the field of causal reasoning for machine learning, the Data Science group is one of the largest research groups in the Netherlands.
Heskes and his colleagues try to extract more information from the data to reason about cause and effect. Heskes: ‘One of the basic ideas that we use is that a model that goes from cause to effect is likely less complex than a model that goes from effect to cause. This essentially goes back to the philosophical thesis of Ockham’s razor, which states that the simplest explanations are usually the best ones. For example, we have applied this idea to data about attention deficit hyperactivity disorder, ADHD. The data show that the attention deficit causes the hyperactivity and not the other way around. When machines get better at causal reasoning, it makes them much smarter and more robust in many applications.’

Artificial immune system

PhD candidate Franka Buytenhuijs has worked since 2020 in the second big theme of the Data Science group: biomedical applications of machine learning. She is part of the computational immunology subgroup. ‘I work on a project called Artificial immune systems’, she states. ‘Just like the brain, the immune system is also a learning system. The brain was the inspiration for the development of neural networks. We are now looking for a system to describe the immune system’s behaviour. How does the immune system learn? How does it remember? How does it forget? For example, we want to use the insights to study how the immune system determines which cells are harmful and which ones are harmless. My research focuses on a specific type of immune cell called T cells. From experimental data from Canadian colleagues, I am trying to find features that determine how strong T cells bind to viruses and the body’s own cells.’
Before starting her PhD research in the Data Science group, Buytenhuijs had completed her Master’s project in the same group. ‘I have a background in AI, but I like to apply this knowledge in the medical domain’, she says. ‘Furthermore, I enjoy the broad diversity of topics in the Data Science group, the ease with which everybody can be approached and the minimum of hierarchy. And despite the diversity of topics, most group members use some form of AI technique, so we can still learn from each other. We get a chance to share our results in our bi-weekly seminar.’

Learning from clicks

The bi-weekly seminar is co-organised by assistant professor Harrie Oosterhuis whose research focuses on optimising ranking systems for search engines and recommender systems. His work is part of the third theme of the Data Science group: information retrieval and recommender systems. ‘We develop statistical methods that learn from the click behaviour of users’, says Oosterhuis. ‘One of the applications is that search or recommender results that are displayed lower in the results list, but that people click on frequently, get an extra push to the top.’ In recent years, the work of Oosterhuis has won three best paper awards at the top conferences in the information field.
People sometimes ask Oosterhuis if improving search and recommendation systems is not already solved by companies like Google or Microsoft. ‘Of course, they are working on that as well’, replies Oosterhuis, ‘but companies like to solve it best for themselves and don’t like to share their results. In our group, we think it is important that what we develop is freely available and open for investigation and improvement.’ For example, group members Arjen de Vries and Djoerd Hiemstra are working on a European search engine that does not depend on large American tech companies. ‘We do not focus solely on publishing papers but also aim to contribute to such socially responsible initiatives’, says Oosterhuis.

Group passport – Data Science at Radboud University

Research fields

Causal reasoning for machine learning, Biomedical applications of machine learning, Information retrieval and recommender systems.

Institution

The Data Science group is a section of the Institute for Computing and Information Sciences at Radboud University.

Labs

• ICAI Lab ‘Radboud AI for Health’
• AI for Precision Health, Nutrition & Behavior
• AI for Energy Grids
• AI for Parkinson

Websites

• Data Science group
• Radboud AI (campus-wide initiative connecting all activities on Artificial Intelligence and Data Science within Radboud University and Radboudumc)

By Bennie Mols
Images Ivar Pel