Jan Van Den Bussche
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Jan Van Den Bussche has a total of 93 co-authors. Here are the top co-authors by number of shared publications:
- Bart Bogaerts
KU Leuven
10 shared publications - Jasper Steegmans
University of Antwerp, Antwerp, Belgium and Hasselt University, Hasselt, Belgium
5 shared publications - Maxime Jakubowski
TU Wien
5 shared publications - Dirk Van Gucht
Indiana University
4 shared publications - Jan Hidders
Birkbeck, University of London, London, United Kingdom
4 shared publications - Tom J. Ameloot
Hasselt University and Transnational University of Limburg, Hasselt 3500, Belgium
4 shared publications
Latest Publications
- Open Access
- 30 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 1
Models are Data Too: Towards Expressive Querying of Machine-learning Models
There is a case to be made for the querying of machine-learning models. In the data science enterprise, a large number of models are being produced. This is due to testing, hyperparameter tuning, continual learning, and also, simply, the large number of different projects and applications taken up by the enterprise. We argue that these models are valuable, intensional data and should be managed as such. In particular, we want to be able to query these data, along with the more conventional, extensional data. There are several reasons for this: calling models within conventional queries; treating models as infinite relations; analysing models (e.g., pruning); interpreting and explaining models; and supporting verification and trustworthiness. In this keynote we will present different forms that model-query languages can take, investigate their expressive power, and discuss implementation challenges. We note that Marcelo Arenas's keynote at PODS 2024 already addressed a similar subject, but focusing mostly on their approach towards querying boolean models [1]. In contrast, here we will mostly consider neural network models [2-4].
- 30 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 1
Open AccessModels are Data Too: Towards Expressive Querying of Machine-learning Models
There is a case to be made for the querying of machine-learning models. In the data science enterprise, a large number of models are being produced. This is due to testing, hyperparameter tuning, continual learning, and also, simply, the large number of different projects and applications taken up by the enterprise. We argue that these models are valuable, intensional data and should be managed as such. In particular, we want to be able to query these data, along with the more conventional, extensional data. There are several reasons for this: calling models within conventional queries; treating models as infinite relations; analysing models (e.g., pruning); interpreting and explaining models; and supporting verification and trustworthiness. In this keynote we will present different forms that model-query languages can take, investigate their expressive power, and discuss implementation challenges. We note that Marcelo Arenas's keynote at PODS 2024 already addressed a similar subject, but focusing mostly on their approach towards querying boolean models [1]. In contrast, here we will mostly consider neural network models [2-4].
- Open Access
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4 (2) , 1-20
Recursive Querying of Neural Networks via Weighted Structures
Expressive querying of machine learning models--viewed as a form of intensional data-enables their verification and interpretation using declarative languages, thus making learned representations of data more accessible. Motivated by the querying of feedforward neural networks, we investigate logics for weighted structures. In the absence of a bound on neural network depth, such logics must incorporate recursion; thereto we revisit the functional fixpoint mechanism proposed by Grädel and Gurevich. We adopt it in a Datalog-like syntax; we extend normal forms for fixpoint logics to weighted structures; and show an equivalent ''loose'' fixpoint mechanism that allows values of inductively defined weight functions to be overwritten. We propose a ''scalar'' restriction of functional fixpoint logic, of polynomial-time data complexity, and show it can express all PTIME model-agnostic queries over reduced networks with polynomially bounded weights. In contrast, we show that very simple model-agnostic queries are already NP-complete. Finally, we consider transformations of weighted structures by iterated transductions.
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4 (2) , 1-20
Open AccessRecursive Querying of Neural Networks via Weighted Structures
Expressive querying of machine learning models--viewed as a form of intensional data-enables their verification and interpretation using declarative languages, thus making learned representations of data more accessible. Motivated by the querying of feedforward neural networks, we investigate logics for weighted structures. In the absence of a bound on neural network depth, such logics must incorporate recursion; thereto we revisit the functional fixpoint mechanism proposed by Grädel and Gurevich. We adopt it in a Datalog-like syntax; we extend normal forms for fixpoint logics to weighted structures; and show an equivalent ''loose'' fixpoint mechanism that allows values of inductively defined weight functions to be overwritten. We propose a ''scalar'' restriction of functional fixpoint logic, of polynomial-time data complexity, and show it can express all PTIME model-agnostic queries over reduced networks with polynomially bounded weights. In contrast, we show that very simple model-agnostic queries are already NP-complete. Finally, we consider transformations of weighted structures by iterated transductions.
- Open Access
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4 (2) , 1-19
A Logical View of GNN-Style Computation and the Role of Activation Functions
We study the numerical and Boolean expressiveness of MPLang, a declarative language that captures the computation of graph neural networks (GNNs) through linear message passing and activation functions. We begin with A-MPLang, the fragment without activation functions, and give a characterization of its expressive power in terms of walk-summed features. For bounded activation functions, we show that (under mild conditions) all eventually constant activations yield the same expressive power—numerical and Boolean—and that it subsumes previously established logics for GNNs with eventually constant activation functions but without linear layers. Finally, we prove the first expressive separation between unbounded and bounded activations in the presence of linear layers: MPLang with ReLU is strictly more powerful for numerical queries than MPLang with eventually constant activation functions, e.g., truncated ReLU. This hinges on subtle interactions between linear aggregation and eventually constant non-linearities, and it establishes that GNNs using ReLU are more expressive than those restricted to eventually constant activations and linear layers.
- 12 May 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 4 (2) , 1-19
Open AccessA Logical View of GNN-Style Computation and the Role of Activation Functions
We study the numerical and Boolean expressiveness of MPLang, a declarative language that captures the computation of graph neural networks (GNNs) through linear message passing and activation functions. We begin with A-MPLang, the fragment without activation functions, and give a characterization of its expressive power in terms of walk-summed features. For bounded activation functions, we show that (under mild conditions) all eventually constant activations yield the same expressive power—numerical and Boolean—and that it subsumes previously established logics for GNNs with eventually constant activation functions but without linear layers. Finally, we prove the first expressive separation between unbounded and bounded activations in the presence of linear layers: MPLang with ReLU is strictly more powerful for numerical queries than MPLang with eventually constant activation functions, e.g., truncated ReLU. This hinges on subtle interactions between linear aggregation and eventually constant non-linearities, and it establishes that GNNs using ReLU are more expressive than those restricted to eventually constant activations and linear layers.
- Open Access
- 16 April 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Database Systems
- Vol. 51 (4) , 1-30
The Relational Completeness of the DNA Query Language DNAQL
DNAQL is a query language for databases implemented as DNA molecules in solution, with well-known DNA wetlab procedures as operators. DNAQL focusses on faithful implementability in DNA, and, as a result, does not include powerful, but difficult to implement, features like nonterminating/recursive hybridization needed for general computation (i.e., Turing-completeness). In this article, we show that DNAQL is relationally complete, i.e., the full relational algebra can be performed by DNAQL programs.
- 16 April 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Database Systems
- Vol. 51 (4) , 1-30
Open AccessThe Relational Completeness of the DNA Query Language DNAQL
DNAQL is a query language for databases implemented as DNA molecules in solution, with well-known DNA wetlab procedures as operators. DNAQL focusses on faithful implementability in DNA, and, as a result, does not include powerful, but difficult to implement, features like nonterminating/recursive hybridization needed for general computation (i.e., Turing-completeness). In this article, we show that DNAQL is relationally complete, i.e., the full relational algebra can be performed by DNAQL programs.
- 20 March 2026
- Published by Springer Nature
- p. 195-210
Analyzing the Behavior of Database Applications Through Size Descriptions
- 20 March 2026
- Published by Springer Nature
- p. 195-210
Analyzing the Behavior of Database Applications Through Size Descriptions
- Open Access
- 6 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on the Web
Incremental View Maintenance for SPARQL Queries: Adapting the Counting Algorithm
The counting algorithm is a classic approach to incremental view maintenance for queries on relational data. We adapt this algorithm to SPARQL queries on RDF (Resource Description Framework) datasets. In the decentralized Web, data are customarily stored in RDF, which is linked data that can be queried using SPARQL. The efficiency of retrieval of query results can be improved through maintaining views on the data incrementally as the underlying data change. SPARQL operators give rise to heterogeneous sets of solution mappings and involve multi-set semantics. We develop a theory of SPARQL algebra on annotated sets of solution mappings and show how to handle operators traditionally perceived as difficult, including LeftJoin, Diff, and Minus. We discuss the implementation of our methods—based on the counting algorithm—and assess the feasibility using data and queries from the Berlin SPARQL Benchmark. Performant incremental view maintenance will be instrumental when querying the decentralized web at scale.
- 6 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on the Web
Open AccessIncremental View Maintenance for SPARQL Queries: Adapting the Counting Algorithm
The counting algorithm is a classic approach to incremental view maintenance for queries on relational data. We adapt this algorithm to SPARQL queries on RDF (Resource Description Framework) datasets. In the decentralized Web, data are customarily stored in RDF, which is linked data that can be queried using SPARQL. The efficiency of retrieval of query results can be improved through maintaining views on the data incrementally as the underlying data change. SPARQL operators give rise to heterogeneous sets of solution mappings and involve multi-set semantics. We develop a theory of SPARQL algebra on annotated sets of solution mappings and show how to handle operators traditionally perceived as difficult, including LeftJoin, Diff, and Minus. We discuss the implementation of our methods—based on the counting algorithm—and assess the feasibility using data and queries from the Berlin SPARQL Benchmark. Performant incremental view maintenance will be instrumental when querying the decentralized web at scale.
- Open Access
A Logical View of GNN-Style Computation and the Role of Activation Functions
We study the numerical and Boolean expressiveness of MPLang, a declarative language that captures the computation of graph neural networks (GNNs) through linear message passing and activation functions. We begin with A-MPLang, the fragment without activation functions, and give a characterization of its expressive power in terms of walk-summed features. For bounded activation functions, we show that (under mild conditions) all eventually constant activations yield the same expressive power - numerical and Boolean - and that it subsumes previously established logics for GNNs with eventually constant activation functions but without linear layers. Finally, we prove the first expressive separation between unbounded and bounded activations in the presence of linear layers: MPLang with ReLU is strictly more powerful for numerical queries than MPLang with eventually constant activation functions, e.g., truncated ReLU. This hinges on subtle interactions between linear aggregation and eventually constant non-linearities, and it establishes that GNNs using ReLU are more expressive than those restricted to eventually constant activations and linear layers.
Open AccessA Logical View of GNN-Style Computation and the Role of Activation Functions
We study the numerical and Boolean expressiveness of MPLang, a declarative language that captures the computation of graph neural networks (GNNs) through linear message passing and activation functions. We begin with A-MPLang, the fragment without activation functions, and give a characterization of its expressive power in terms of walk-summed features. For bounded activation functions, we show that (under mild conditions) all eventually constant activations yield the same expressive power - numerical and Boolean - and that it subsumes previously established logics for GNNs with eventually constant activation functions but without linear layers. Finally, we prove the first expressive separation between unbounded and bounded activations in the presence of linear layers: MPLang with ReLU is strictly more powerful for numerical queries than MPLang with eventually constant activation functions, e.g., truncated ReLU. This hinges on subtle interactions between linear aggregation and eventually constant non-linearities, and it establishes that GNNs using ReLU are more expressive than those restricted to eventually constant activations and linear layers.
- Open Access
- 30 June 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Database Systems
- Vol. 50 (3) , 1-38
Expressiveness within Sequence Datalog
Motivated by old and new applications, we investigate Datalog as a language for sequence databases. We reconsider classical features of Datalog programs, such as negation, recursion, intermediate predicates, and relations of higher arities. We also consider new features that are useful for sequences, notably, equations between path expressions, and “packing”. Our goal is to clarify the relative expressiveness of all these different features, in the context of sequences. Towards our goal, we establish a number of redundancy and primitivity results, showing that certain features can, or cannot, be expressed in terms of other features. These results paint a complete picture of the expressiveness relationships among all possible Sequence Datalog fragments that can be formed using the six features that we consider.
- 30 June 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Database Systems
- Vol. 50 (3) , 1-38
Open AccessExpressiveness within Sequence Datalog
Motivated by old and new applications, we investigate Datalog as a language for sequence databases. We reconsider classical features of Datalog programs, such as negation, recursion, intermediate predicates, and relations of higher arities. We also consider new features that are useful for sequences, notably, equations between path expressions, and “packing”. Our goal is to clarify the relative expressiveness of all these different features, in the context of sequences. Towards our goal, we establish a number of redundancy and primitivity results, showing that certain features can, or cannot, be expressed in terms of other features. These results paint a complete picture of the expressiveness relationships among all possible Sequence Datalog fragments that can be formed using the six features that we consider.
- Open Access
- 22 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-5
SQL4NN: Validation and Expressive Querying of Models as Data
- 22 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-5
Open AccessSQL4NN: Validation and Expressive Querying of Models as Data
- Open Access
SQL4NN: Validation and expressive querying of models as data
We consider machine learning models, learned from data, to be an important, intensional, kind of data in themselves. As such, various analysis tasks on models can be thought of as queries over this intensional data, often combined with extensional data such as data for training or validation. We demonstrate that relational database systems and SQL can actually be well suited for many such tasks.
Open AccessSQL4NN: Validation and expressive querying of models as data
We consider machine learning models, learned from data, to be an important, intensional, kind of data in themselves. As such, various analysis tasks on models can be thought of as queries over this intensional data, often combined with extensional data such as data for training or validation. We demonstrate that relational database systems and SQL can actually be well suited for many such tasks.