To celebrate the annual international Ada Lovelace Day and women’s achievements in Computer Science, we interviewed Dr Valentina Tamma, Assistant Professor in Computer Science at the University of Liverpool. (5 minute read)

With an impressive career in Computer Science research, in the areas of Artificial Intelligence, ontologies and knowledge sharing, we asked Dr Valentina Tamma to share more about this fascinating and fast-growing AI industry and how ontologies and knowledge sharing can help us all.

Originating from Ancient Greek roots, the word ‘ontology’ refers to the philosophical study of existence and being and explores what entities exist and how they are related. Nowadays the term ‘an ontology’ has been adopted into our modern language within the context of Artificial Intelligence, where it refers to a structured way of representing knowledge, defining key concepts in a domain and how they connect to support understanding and knowledge sharing between systems.

As Valentina explains “in AI, ontology can be seen as a shared vocabulary for a subject area, it defines the things that are important for that area. It could be people, locations, events, concepts, but most importantly it explains what these things mean and how they relate to each other. 

For example, a Lego set’s blueprint instructions don’t just tell you what the Lego pieces are, but also tell you how you put the pieces together to build a Lego castle or rocket etc.”
The individual Lego set is an ontology and the blueprint instructions the knowledge sharing.

Valentina continues “Knowledge sharing is simply the process of making information understandable and reusable by others, so it is an agreement in using the same vocabulary (language) and using it consistently across different people and systems.”

This is a relatively new research area and industry. The use of ontologies in AI grew in the 1990s as researchers looked for better ways to represent and share knowledge across systems. Around 2000, this idea became central to the vision of the Semantic Web, which aimed to make online information understandable to both humans and machines. These efforts laid the groundwork for today’s use of ontologies and knowledge graphs (which use ontologies to link real-world data into meaningful networks) to connect data and enable smarter, more interoperable (and ultimately intelligent) systems.

But how does this affect us in the real world? Valentina explains that without a shared understanding of the meaning of information and data, even the most advanced systems can fail.

She recalls the 1999 NASA Mars Climate Orbiter mission, which was lost due to a simple but costly misunderstanding: one piece of software used imperial units, while another used metric. As a result, the spacecraft entered the Martian atmosphere at the wrong altitude and was destroyed: a $125 million loss caused by the different NASA systems interpreting data inconsistently. “Ontologies aim to prevent exactly these kinds of problems by making the meaning of information explicit by defining concepts, relationships, and measurement units in a common, structured way so that both humans and machines can interpret information in the same way and avoid hidden assumptions.”

Valentina goes on to clarify that this same principle, i.e. making the meaning clear and shared, is now driving some of the most exciting progress in AI. “Ontologies and knowledge graphs, which are essentially ontologies filled with real data and connections, are becoming a key part of how intelligent systems learn and reason. They give AI a structured understanding of the world, helping it combine logical thinking with what it learns from data through Machine Learning.”

This field of research is developing rapidly, and new trends are emerging. Valentina is involved with the latest developments in knowledge graphs and neuro-symbolic artificial intelligence, “which combines symbolic reasoning (using logic and structured knowledge) with machine learning’s ability to spot patterns in large amounts of data. This approach helps AI systems go beyond recognising patterns to explaining their conclusions and reason about the world,” 

Recently, Valentina’s been collaborating with the British Music Experience Museum, the UK’s museum of popular music, to develop a knowledge graph that brings their vast collection to life.  They hold artefacts related to popular music, from musical instruments and original manuscripts of lyrics of songs to concert costumes and memorabilia.  

Valentina shares that “Our goal has been to connect these items with information about artists, genres, and key moments in music history, creating a rich web of relationships that helps visitors and researchers explore how British music has evolved over time.
“A good example is a drum kit used by the band Queen, which we can explicitly link in the knowledge graph to the Live Aid 1985 concert, and to David Bowie who performed with Queen during the concert.  This same drum kit could now feature in an exhibition marking the 40th anniversary of Live Aid where the knowledge graph connects all these facts, showing how artists, events, and artefacts are related, and links them to other facts about some other artefacts. The knowledge graph provides an interactive way for people to explore and query the museum’s catalogue.”

Valentina also makes us aware of other examples that will impact many of us directly. “Mobile phone company Samsung is starting to implement personal knowledge graphs within their mobile phones.  The knowledge graph gathers all the personal information that we keep on our mobile phone, and it connects appointments, emails, pictures, locations, etc with all the facts that we store on the device.”  [watch the launch of Samsung's phone incorporating ontology https://www.youtube.com/watch?v=HinL5jCy_oI ]

So how does Ontology help AI decide between facts, fiction or recognise bias? Valentina animatedly explains that “One of the most exciting developments right now is the use of ontologies and knowledge graphs to detect and counter bias or incorrect inferences in systems like ChatGPT, Google, or Alexa etc. These models tend to generalise from the most common things they’ve seen and forget what we call the ‘long tail’, the rare or unusual cases, which are often the most interesting ones. Knowledge graphs give AI a kind of factual backbone. They connect verified, structured facts and relationships, so the system can check what it’s saying against real context, instead of just guessing from patterns.”

During our interview it’s obvious that Valentina is passionate about the subject and her research, and it’s clear how much she enjoys sharing her expertise.   As an area editor of the newly established journal ‘Transactions on Graph Data and Knowledge’ Valentina is immensely proud of the publication and that it is a Diamond Open Access journal, meaning it is free for both authors to publish in and readers to access.  The journal helps democratise research by making high-quality work available to everyone, and it publishes original research articles and surveys on graph-based approaches to representing data and knowledge. https://drops.dagstuhl.de/entities/journal/TGDK 

As a researcher, Valentina works closely within the Ontology and knowledge graph community. “We all just want to advance the research and work with others because collaboration is essential to research. Research can't be insular. It needs to be a shared endeavour.

“This field of research is really important for the future. Right now, AI systems tend to make broad generalisations, and they’re being used by almost everyone. So, we need humans and AI to work together, with safeguards that make sure decisions are based on facts, not just patterns in data. Knowledge graphs are one of the key ways we can do that, they help AI stay anchored to real, verifiable information.”

In addition to her research and teaching, Valentina’s also a LivWiSE (Liverpool Women in Science and Engineering) role model and is inspiring future generations of women and girls to study computer science.  

She notes that “I'm very privileged because in my research community, the female representation is very high.  Having a role model is not enough, we need to present to all women (in academia and beyond), a wide variety of role models who are relatable, approachable, and able to provide examples of how women can overcome challenges. This is why I believe it is so important to start inspiring girls early and start this conversation with girls when they are still at school, which is why I decided to become a governor at West Kirby Grammar School for Girls.”

Biography - Dr Valentina Tamma, Assistant professor, University of Liverpool
Valentina Tamma is an assistant professor in the School of Computer Science and Informatics at the University of Liverpool. Her research focuses on AI ontology engineering and, in particular, in designing, aligning and evolving ontologies to support autonomous decision making in rational agents and other autonomous systems. More recently, Valentina has been working in the area of ontology engineering and on industry-led projects exploring the use of ontologies and semantic technologies for managing complex data and knowledge systems. She has served as a reviewer for major AI, multi-agent systems, and semantic web conferences and journals. She is an associate editor for the Journal of Web Semantics, an Area Editor for the newly established Transactions on Graph Data and Knowledge journal, and currently co-chairs the Knowledge Graphs Interest Group at the Alan Turing Institute.

For more information about Valentina, visit her LivWiSE profile