Meet: Professor Elena Simperl

Elena Simperl's journey in artificial intelligence began with a fascination for knowledge engineering. This is an area of AI that takes a distinctly different approach from today's mainstream machine learning systems, by valuing curated, trusted knowledge, represented neurosymbolically, to facilitate advanced AI capabilities such as abstraction, analogy, or planning. An engineer by training, Elena was always passionate about building knowledge-based AI that works, solving real problems and making a difference in the world.

Her work with Wikidata, a project of the Wikimedia Foundation, exemplifies this. Wikidata serves as the central symbolic knowledge base that powers those fact boxes you see on Wikipedia pages.

"In machine learning, you rely very much on available data to understand patterns and generate predictions," she explains. Knowledge engineering brings a grounding with real-world facts, organised in the same way as we as humans think about knowledge, in terms of entities and links between them. This helps make AI systems more precise, especially when there isn’t a lot of data available to learn patterns from.

For example, knowledge graphs, a knowledge engineering technique, are used extensively to respond to search engine queries about entities of interest such as people, businesses and events, delivering accurate responses with a clear provenance, which anyone can readily understand and verify.

"That's what we mean by grounding," Simperl notes. "When you consider this kind of background knowledge, curated by people, to a system, hallucinations that we often see in generative AI start to disappear."

After studying in Germany, Simperl moved to the UK in 2012 as an Associate Professor at the University of Southampton. At that time, while researchers had developed sophisticated methods for creating AI knowledge bases, they faced a critical challenge: insufficient data to apply to these knowledge bases. This realisation led Simperl down an unexpected path, investigating ways to encourage organisations to make their data more accessible, and to participate in community initiatives to develop shared knowledge bases.

Now as a Professor of Computer Science at King's College London, she is currently working with Siemens and the Technical University of Munich to ensure knowledge bases, which these days rely on a mix of human and algorithmic capabilities, follow responsible research and innovation practices and make sure they are in line with regulation. On the same theme of doing AI responsibly, she also led on the development of standardised vocabularies for machine learning datasets to help organisations get their data AI-ready, hosted by MLCommons. Elena was named as an Association for Computing Machinery 2024 Distinguished Member for her contributions to knowledge engineering, knowledge graphs, and the semantic web.

Elena is also Director of Research at the Open Data Institute, and continues to champion open approaches to AI development. She is leading on examining how generative AI can support Wikipedia communities in different languages without compromising their knowledge integrity or cultural autonomy.

"We're looking at how to support these committed teams with AI in ways that empower them to use the technology and contribute to AI data commons in multiple languages, which are currently missing." The four-year project involves careful consideration of everything from dataset creation and curation to community involvement in system auditing and retaining participation across varying levels of AI literacy and trust.

Simperl's dedication to openness extends to her role on the OpenUK's AI Advisory Board, where she's working to promote genuine transparency in AI development and advance open AI ecosystems. "We hear a lot about 'open AI models,' but when you look at what information is actually available, most companies only publish selective details," she observes. "For an open approach to work, we need full transparency about training data, processing methods, and decision-making throughout the development process."

At King's College London, where she's been since 2020, Simperl particularly values the interdisciplinary environment. Coming from primarily STEM-focused institutions, she finds King's genuinely multidisciplinary approach inspiring and essential for addressing the complex challenges of AI development.

"When you work in a volunteer-based context or support open ecosystems, the results are often for the public good," Simperl reflects, summarising her career-long commitment to open, community-driven approaches. As AI continues to evolve, her work ensures that human knowledge and community wisdom remain at the heart of technological advancement.

Professor Elena Simperl is a Professor of Computer Science in the Department of Informatics at King's College London.