My research is at the intersection between AI and social computing. I study and build digital systems that combine data and algorithms with human and social capabilities. I have always been fascinated by collective and hybrid intelligence systems as a means to tackle some of most important challenges of our times, whether that’s in the form of citizen-science platforms like the Zooniverse and Eyewire; participatory-sensing apps that create added value through ingenious mixes of data and crowd contributions; universal knowledge bases like Dbpedia and Wikidata; or open-innovation ecosystems for data and AI.

I had the chance to lead 16 research and innovation projects and work on 28 more, often in collaboration with researchers and industry from all over the world. My personal highlights in this journey are INSEMTIVES, RENDER, QROWD, and MediaFutures. These projects have a lot in common – they look at the web or adjacent large, decentralised systems as sociotechnical artifacts, which require an understanding and appreciation of the experiences, motivations, and incentives of their users, in addition to sophisticated technologies, to deliver value.  For example, in QROWD, I led a team that built a hybrid, crowd-AI architecture for smart transport, which enables apps and services for traffic management, navigation and, parking.  In the same space, I contributed to SOCIAM, a large UK programme studying the design and evolution of social machines, including citizen-science platforms, online communities, and social networks. As a scientist, I have always found citizen science a fascinating application of hybrid and collective intelligence:  in ACTION, my lab worked with 16 citizen-science communities around Europe fight pollution, and we’ve scaled that model of engagement to more than 125 initiatives to date in the programme Impetus.

I have a background in semantic web and symbolic AI. Since my PhD I have worked on sociotechnical methods to enhance data access, quality and use –  well-governed, machine-readable data are critical to better AI systems. This started with research on socioeconomics aspects of knowledge bases when I developed the first statistical model to estimate the costs of building and maintaining knowledge bases. Knowledge bases (and knowledge graphs, as knowledge bases are called today) are an important category of AI datasets, which capture information about a domain in a formal way, facilitating reasoning and explanations, reducing hallucinations of large language models, and improving machine-learning performance when training data is scarce. Given my interest in hybrid and collective intelligence, a lot of the work I do in this space is motivated by, and applied to, collaborative knowledge bases such as Wikidata. This is the subject of my Hans Fischer Senior Fellowship, funded by Siemens and undertaken in collaboration with colleagues at my alma mater, the Technical University of Munich. Among others, I work on methods to mitigate biases in knowledge graphs, and in making the entire knowledge graph life cycle more transparent and accountable.

Between 2015 and 2020, I led two European data incubators, ODINE and Data Pitch, helping almost a hundred small and medium businesses from more than 20 countries innovate with data. Both programmes taught me a lot about how people use open, closed, and shared data. I’ve tried to apply the lessons I’ve learned to new domains – from 2017 to 2021 I was the principal investigator of Data Stories, an EPSRC funded grant that develops concepts and tools that make data  easier to engage with in everyday situations; this line of research has taken me into arts-inspired approaches to data and technology engagement, as in the European programme MediaFutures. 

I have been part of many amazing initiatives of the science and tech community – I served as chair and programme chair of the European and International Semantic Web Conference series, the European Data Forum, the European Semantic Technologies conference, the AAAI Conference on Human Computation and Crowdsourcing, and The Web Conference. I am currently the president of the Semantic Web Science Association. Over the years, have led many summer schools in Europe, Asia, and the US and pursued a range of education activities, from projects like EUCLID, which developed an open curriculum for Linked Data, to the Southampton Data Science Academy, which offers professional training in data science and AI. Since 2023, I co-chair of the MLCommons working group Croissant, which developed the first machine-readable open vocabulary to standardise how AI datasets are described, so as to make them easily discoverable and usable across tools and platforms. More than 700k datasets are annotated using Croissant to date.

In 2023 I have ventured outside academia and joined the Open Data Institute as director of research. Among others, I’ve set up a programme of research and advocacy on data-centric AI. Seeing research insights making their way into applications and having an impact on practices, standards, policy making is a rewarding, but humbling experience and I’m enjoying every minute of it.