In the fourth and final article of this series, Adam Parr joins us to discuss his journey to become a founder of Oxford Semantic Technologies and his insights into the semantics world. If you missed the previous articles, you can find them here: Professor Ian Horrocks, Professor Boris Motik and Professor Bernardo Cuenca Grau.
“I moved to Oxford in 2006 to join the Williams Formula One team where I worked for five years, before getting into venture capital. I was a founding investor of a company called Mind Foundry. In 2016, through the Mind Foundry Professors, Steve Roberts and Mike Osborne, I met Ian Horrocks. At this time, I hadn’t heard of a knowledge graph. Ian told me about the spin-out he was about to launch with two colleagues Bernardo and Boris. I loved the technology and the Professors. My co-investor and I decided we would back the initial round and I ended up chairing the company.”
“The beauty of RDFox is being able to ask, and importantly answer, the questions that people want to ask. To do this, it is about understanding what clients are trying to do logically, rather than understanding the software. In most businesses you have one or more sets of data and you may have a business intelligence system which allows you to ask certain questions. But the questions it allows you to ask are the questions that it’s able to answer.
Using a knowledge graph involves asking the questions the clients want to ask and thinking about how to answer them. RDFox helps you turn that model into code using rules (i.e. reasoning) bringing in the necessary data from various data sources to answer the question, in a very smart way.
I love two other very important things. First, the team — we have three great academic founders and a small team of developers, knowledge engineers and data scientists who are fantastic. Second, the clients. Their problems are varied and intellectually stimulating. Many have tried to solve their problem with other databases or solutions, which quite honestly, weren’t fast enough, or good at reasoning. We are the kind of company that you go to when you have an interesting problem, and that makes for a very enjoyable business.”
“When we invested in this company, we knew that it was cutting edge technology — not all knowledge graphs do reasoning, or do it well! As companies acquire more and more data and need more value from that data, logically they’re going to need smarter database technology that can integrate and interpret it. I expect to see the demand rising and people’s understanding of this field increasing. We’re like a surfer that’s just getting up on the board — the market is coming in our direction!”
“Totally, there is another level of intelligence in knowledge graphs compared with relational databases. The declarative structure of knowledge graphs allows flexibility to shape knowledge and adapt data models as variables change.
Knowledge graphs reflect how our brains work. We don’t have tables with rows of information, instead our minds use a series of relationships and connections. As Tim Berners-Lee says, all information is relationships; information without relationships doesn’t make sense. Storing information as relationships reflects a natural way of processing information.
The world we live in is a networked structure, we can see this throughout nature. This is also evident in the human world, for example, Apple has become a 2 trillion dollar company through network effects. Google and Facebook are the same. Society operates through networks and relationships. Our knowledge graph technology aligns with this whole way of thinking.”
“The thing I find very exciting about RDFox, and about this field, is that the challenge is not a technical one in the sense of being able to read C++ or code; instead the challenge is intellectual.
Understanding reasoning and the logic behind knowledge graphs is what is challenging. For example, I find the preparation of data interesting. In a slightly deceptive way, we often use terms without fully defining them. Take your coffee example, if I said ‘I love a good coffee’ and you agreed, we might not be referring to the same thing. Is a good coffee 100% Arabica beans, ethically produced, does it have subtle cereal notes, or does it just taste delicious? RDFox makes the user become more rigorous in their discipline and thinking about concepts, which I think won’t do any harm.
A concrete example is the work we did with Mind Foundry and HSE. Safety is a very interesting problem intellectually. To model the data in order to predict injuries and determine whether businesses have a good safety culture, we first must define what a good safety culture is. Is it few injuries, lots of engagement with staff or lots of inspections? By thinking through and designing concepts, we are closer to achieving real answers to real questions.”
“I think getting information into a knowledge graph at scale is a challenge. We can create triples at scale and quickly from structured data, but I would love to see this same functionality with unstructured data. In line with the vision that Tim Berners-Lee had for the semantic web, where all the information on the Internet is machine readable. For this we need knowledge graph technology and the ability to bring unstructured data into a knowledge graph. The power of this would be phenomenal.”
“If you know the question that you want to answer and explore, and if you want to integrate data, then my tip is come and talk to us because we love solving these problems.”
Thank you to all our founders for their insightful contributions to the ‘Meet the Founders’ series. This concludes the series, however, if you have any questions for our founders or want to get in touch, email firstname.lastname@example.org