Speaker: Prof. Dr. Stefan Dietze
GESIS – Leibniz Institute for the Social Sciences & Heinrich-Heine-University Düsseldorf

Abstract: Scientific discourse is scattered across unstructured scholarly publications and increasingly takes place online, e.g. in news or social media. Understanding the state-of-the-art in specific research fields, involved data, software or methods, and their impact in both science and society requires substantial efforts and has become increasingly challenging. At the same time, societal debates about topics such as COVID or climate change have demonstrated the impact of science discourse on public opinion, policies and society as a whole. This talk will provide an overview over a range of works that use deep learning-based NLP, such as PLMs and LLMs, to construct and use knowledge graphs about scientific discourse. These include, on the one hand, approaches that extract metadata about scholarly entities, such as code, data, tasks or machine learning models from scientific publications to enable machine-interpretable research information and understand dependencies between scholarly artefacts. On the other hand, we introduce NLP methods and knowledge graphs that enable an understanding of societal discourse about science, e.g. on Twitter/X, and facilitate interdisciplinary research into (mis-)representation and -information of scientific claims and findings in societal debates.

Short Bio: Stefan Dietze is Professor for Data & Knowledge Engineering at Heinrich-Heine-University Düsseldorf (HHU), and scientific director of the department Knowledge Technologies for the Social Science (KTS) at GESIS – Leibniz Institute for the Social Sciences. He also is affiliated member at the Heine Center for Artificial Intelligence & Data Science (HeiCAD), the Düsseldorf Institute for Internet & Democracy (DIID) and the L3S Research Center of the Leibniz University Hanover, Germany. Previous positions include the Knowledge Media Institute (KMI) of The Open University (UK) and the Fraunhofer Institute for Software and Systems Engineering (ISST, now part of Fraunhofer FOKUS). His research interests are at the intersection of information retrieval, knowledge graphs, NLP and machine learning, and his work is concerned with the extraction, fusion and search of knowledge and data, in particular, on the Web. Beyond being a data scientist, he is a freelance writer for magazines & blogs.

Abstract: Creating intelligent applications that valorise complex domain data such as in the scientific, technical, and legal domain often calls for solutions that combine learning and symbolic artificial intelligence (AI) methods. In line with such developments, in the first part of this talk, we focus on describing a new sub-area of AI that focuses on combining Machine Learning components with techniques developed by the Semantic Web community—Semantic Web Machine Learning (SWeML). We report on the results of a systematic mapping study during which we analysed nearly 500 papers published in the past decade in this area, where we focused on evaluating architectural and application-specific features of such systems. In the second part of the talk, we describe the development and evaluation of a concrete SWeML system that aims to extract key elements from official Austrian permits, including the Issuing Authority, the Operator of the facility in question, the Reference Number, and the Issuing Date. We hope that our lessons learned both about this area as a whole (through the survey of SWeML systems) and the concrete system we built will provide inspiration for researchers and practitioners working with such complex data as in the legal domain and beyond.

Short Bio: Prof. Dr. Marta Sabou is a professor for Information Systems and Business Engineering at the Vienna University of Economics and Business (WU) and the Head of Institute for Data, Process and Knowledge Management (DPKM). She holds a PhD in Artificial Intelligence from Vrije Universiteit Amsterdam, for which she won the IEEE Intelligent System’s Ten to Watch Award in 2006. During her career, she performed Artificial Intelligence (AI) research as Research Fellow at the Open University UK, Assistant Professor at MODUL University Vienna, Key Expert in Semantic Technologies at Siemens and FWF Elise-Richter Fellow at the Vienna University of Technology.

Prof. Sabou leads the Semantic Systems research group, which performs foundational and applied research at the intersection of the Semantic Web, Machine Learning and Human Computation research areas. Her group’s research topics range from knowledge engineering (knowledge graphs and their evaluation, data integration) to the development of novel intelligent systems that combine both symbolic and sub-symbolic AI techniques, i.e., neuro-symbolic systems. This foundational research underpins an active involvement in applied research in terms of developing advanced functionalities (e.g., system explainability and auditability) in application areas ranging from tourism and cultural heritage to mission critical domains enabled by complex cyber-physical (social) systems such as smart grids, smart buildings, smart factories (as part of Industry 4.0-5.0). Increasingly, the group addresses topics in the area of Digital Humanism such as the auditing of AI systems and the involvement of human stakeholders in the design of intelligent systems. Prof. Sabou is an accomplished academic (close to 150 peer-reviewed papers, h-index 46) and takes an active role in the Semantic Web research community as an editorial board member for two journals (SWJ, NAI) and conference organiser.