Mathieu d’Aquin is a Professor of Informatics specialised in data analytics and semantic technologies at the Insight Centre for Data Analytics of the National University of Ireland Galway. He was previously Senior Research Fellow at the Knowledge Media Institute of the Open University, where he led the Data Science Group. In this interview, he speaks about research on semantic web technologies and specific application of web data technologies, which are two key areas of his work interest.
You have been working for years on Semantic Web/Linked Data technologies. What will shape our future the most?
There have been strong trends in the Semantic Web and Linked Data communities in the last 15 years. The future of those technologies relies on the way they remain relevant when so many different approaches have been suggested with different properties. In other words, the benefits of the Semantic Web and Linked Data are more than ever relevant, but need to be extracted so they integrate into the practices of developers today, which, even if there has been some transfer, still remain far away from the ones valued by Semantic Web and Linked Data practitioners.
Which technological innovation would you love to see happening?
The trendy answer would be to mention machine learning and things like this. However, what I think the Semantic Web really needs now in terms of technological development is to bring the things that really address the semantic layer into useable forms. Much of the research 15 years ago focused on those aspects, but they didn't make it to the frontline. The tools developed back then do not work in the current developers' environments. Can I get a knowledge engineering tool that would enable me to abstract the very large amounts of data being used in a data analytics process as an ontological model? Can machine learning models be made meaningful by connecting their structure to known ontological knowledge? Can I get an epistemic logic reasoner for OWL that would work on Facebook graph data?
What is your vision of semantic technologies and intelligent applications?
The vision for intelligent applications is that they integrate, and more importantly they consume the Semantic Web. However, this should not be an objective in itself. As alluded to above, Semantic Web technologies need to become tools in the toolkit of developers. This implies (still) that they need to be efficient, robust and have clear benefits. It also means that they should naturally integrate with other approaches, whether it is in data management or data analytics. Doing machine learning with Linked Data should be a natural thing when the data is from multiple sources, complex and heterogeneous. More than ever, to achieve this, we as Semantic Web researchers need to focus on the semantic layer of the data: How do we represent, capture and distill the meaning of data coming in and out of data analytics process?
Which application areas for semantic technologies do you perceive as most promising?
This is a hard question as there are many. I currently work a lot in education, and in the humanities. Humanities research has a lot of potential as the data being manipulated there are not necessarily large, but are very rich, collected through many sources and many different approaches, and focus a lot of relations (between people, organisations, places, artefacts, etc.)! I started exploring Semantic Web technologies, a long time ago, through an application in medicine. Biomedical data still represent a challenge for Semantic Web technologies, and a lot of the work we do at the Data Science Institute and the Insight Centre in NUI Galway is about addressing those challenges, which exist also in many other application domains.
Did you experience a remarkable moment with semantic technologies?
When I started my PhD, the Semantic Web was barely a community, and many people didn't believe in it at all. The only thing I knew, coming from a traditional AI, knowledge representation and reasoning background was that I had to connect to it. There was a paper entitled "DAML+OIL is not enough" by Sean Bechhofer, Carole Goble and Ian Horrocks at the time that was basically saying “if this is what we want to achieve, we need to go beyond the technologies we are building". This paper is still relevant today I believe, but it was really important to me as a young PhD student as it captured clearly some of the discussions in the community about the objectives and the ambitions that were driving those technology developments: There was research to be done towards something really grand! There have been many remarkable moments since then, mainly in showing what in specific cases those technologies enable, but the most important ones to me come when the vision and ambition of what we are trying to achieve become apparent.
You have worked on applying your Semantic Web/Linked Data technologies in areas like medicine and education. What are your results so far?
The results are great applications. In those domains, Semantic Web and Linked Data technologies are really one aspect: They are an enabler. So far they have mostly been helpful as development support, to make the integration of data and knowledge easier, more efficient and more flexible. It is a great result, but as expressed above, I think we can do better and if this is what we want to achieve, we need to go beyond the technologies we are building
In what way can technology be used to do better?
We need more integration not only in the data, but also in the approaches. There has been too much of a divide between different paradigms: Semantic Web technologies on the one side, the rest on the other. Those technologies do not do everything. They have specific benefits that complement, rather than replace others.
What are your expectations about Semantics 2018 in Vienna and the DBpedia Community Meeting?
DBpedia is a great resource, that has helped a lot in showing where Linked Data could go. It has especially been a great research resource, helping Semantic Web researchers in validating, testing and studying specific challenges with the technology or with its use in various domains. Through Semantics and the DBpedia Community Meeting, I hope to see the discussion around this being lifted up beyond providing resources to raising our ambition regarding what Semantic Web research can deliver at a more fundamental level.
Discuss the potential of semantic web technologies with Mathieu at SEMANTiCS 2018 or listen to him at his keynote presentation at the DBpedia Day on September 10, 2018. Register now!
Many thanks to the DBpedia Association, in particular Julia Holze, who supported the preparation of this interview.
The annual SEMANTiCS conference is the meeting place for professionals who make semantic computing work, and understand its benefits and know its limitations. Every year, SEMANTiCS attracts information managers, IT-architects, software engineers, and researchers, from organisations ranging from NPOs, universities, public administrations to the largest companies in the world. http://www.semantics.cc