The Data, Knowledge, and Web Engineering group at Aalborg University is a leading research group that focuses on quantum computing, natural language processing (NLP), knowledge, web, and advanced machine learning methods.

We participate in a number of cross-disciplinary research efforts in different areas, including healthcare, bioscience, education, and security, and closely collaborate with other groups and departments.

Our joint ambition is to bring meaning to large amounts of heterogeneous data and exploit it in the best possible way for a broad range of use cases and applications.

QUANTUM

DKW has been working in quantum computing and quantum information since late 2022. Our focus is on basic research at low Technology Readiness Levels (TRL 1-2), emphasizing modeling quantum noise, error correction coding to reduce physical-to-logical qubit ratios, and sensitivity analysis of quantum methods.

We integrate tools from signal processing with quantum challenges, leveraging extensive simulations and experiments using quantum hardware and high-performance computing.

Our goal is to develop new methods, algorithms, and software to address fundamental quantum problems, including:

• Quantum Noise and Error Correction: Modeling quantum noise and improving error correction methods to optimize physical-to-logical qubit ratios.
• Quantum Algorithms and Simulations: Designing and validating new quantum algorithms through extensive simulations and experiments.
• Sensitivity Analysis: Investigating the impact of quantum state and output variability on implemented methods.

NATURAL LANGUAGE PROCESSING (NLP)

DKW contributes to fundamental research in Natural Language Processing, with a particular focus on multilinguality and low-resource languages. Our work bridges computational linguistics, AI, and data science to address real-world challenges, promoting equitable language technology and robust AI systems.

We carry out fundamental research in the areas of:

• Multilingual NLP: Improving the performance and fairness of NLP systems across typologically diverse linguistic landscapes.
• Linguistically Informed NLP: Leveraging insights from linguistic theory and linguistic typology to improve model performance, interpretability, and robustness.
• NLP Security: Developing methods to detect and mitigate attacks on NLP technology, including generative language models.
• Large Language Models (LLMs): Investigating the factuality, behaviour, security, and ethical deployment of LLMs in diverse applications.
• NLP for Education: Designing NLP systems to support feedback, assessment, and equitable learning environments.

KNOWLEDGE

DKW researches knowledge engineering and machine learning, with a particular focus on knowledge graphs, representation learning and data privacy. We aim to bridge data analytics and AI to address real-world challenges, including applications in healthcare and biology, promoting reliable intelligent systems. We study how to derive actionable knowledge and enhance decision-making. We carry out fundamental research in the areas of:

• Knowledge graphs: Focusing on constructing knowledge graphs, ontology engineering and querying to enhance data integration, retrieval and reasoning across large amounts of heterogeneous and dynamic data.
• Representation learning: Developing novel methods to learn and manage representations of various data modalities, including data sequences (especially genomic sequential data) and graphs, and utilising these representations for data analytics tasks such as pattern mining and data fusion.
• Data privacy: Exploring the use of generative AI and differential privacy to synthetise multimodal datasets, protecting the sensitive information of individuals while preserving utility in various data analytics and data science tasks.
• Semantic technologies: Leveraging and extending semantic web technologies to enable accurate and context-aware analysis.

WEB

DKW has been working on the topics of Web Science and Engineering since the beginning. Our research in Web science concerns the Web as a subject of research and the Web as a technological infrastructure. Our focus is on AI-based methods for learning user preferences from user explicit and implicit data, from side information, and methods for recommending items and users for different tasks on the Web. We are actively advancing the state of the art in Recommender systems, Personalization, User Modeling, (Personalized) Information Retrieval, and Web social networks analytics. The main directions are complemented with the publication, linkage and alignment of open data with other datasets on the Web.

Our goal is to develop new methods, algorithms, and software to address fundamental problems in the areas such as:

• Information Overload: The Web provides access to a large volume of data. AI technologies for personalization and recommendation help to expose the most relevant information.
• Cold Start: Finding additional side information and similar users to remedy situations when there is not enough direct evidence on a particular user or item.
• Explainability: Providing reasons why certain information or items are exposed to a user based on review data or semantic web representations.
• Heterogeneity and Modalities in the Data: Methods for recommendations, alignment, integration and data retrieval when the data comes from diverse sources and different representations including tabular, graph, and text data.

RESEARCH APPROACH

The research approach is primarily constructive in nature: theoretically well-founded, purposeful artefacts such as frameworks, data structures, indexes, algorithms, languages, tools, and systems are prototyped and subjected to empirical study.

Further, the research is mostly driven by novel and challenging real-world applications, with primary application areas being web querying, bio science, and healthcare.

The research has impact on (at least) SDGs 3, 4, 12, 15, and 16.