Jiří Materna’s research team helps you find what you need to know
To be successful outside the university walls, graduates must be able to think contextually.
Who would think that the search results of a web search engine are shaped by a young introvert, who was a closet writer of short stories at the time when the Seznam.cz web portal was born? Jiří Materna, a graduate of the MU Faculty of Informatics, has always been fascinated by artificial intelligence and natural language processing. His enthusiasm won him the position of research leader at always. His bloodshed in society Seznam.cz.
You are the leader of the research team of the largest Czech search engine. Does that mean that you decide what search results appear for specific keywords?
Yes, we create the algorithms that control the order of search results. While this work isn’t as visible as creating a new logo for the landing page, it’s a fascinating job and we really enjoy it. Improving the quality of the search engine by several per cent is a huge success for us although the average user may not even notice it, because changes are implemented continually.
How can you measure if a search was successful or not?
We hire temps who help us calibrate this by evaluating whether the page that was found corresponds with the query. The queries are chosen based on what users actually look for. The evaluation tells us what sites should appear in the results for a specific search. We then compare this information with the actual results and tweak them to get closer to the desired level and improve the search engine.
How did you get to a position with such a big impact on technology development before you were thirty years old?
Part of it was luck. The Brno office of Seznam opened right before I finished my thesis. At that time, I was already interested in things that are relevant to search engines, from computer linguistics through artificial intelligence to machine learning. I wrote my thesis on a topic suggested by Seznam and then started working for them. After several years, I was offered and accepted the position of researcher leader.
Do you think that a student who wants to get a job similar to yours should already start working towards that goal at secondary school?
I think that at secondary school, you should try to broaden your horizons and not specialise in a very narrow field unless you want to start working in a trade right after you finish. University helps you gain more in-depth knowledge in a specific area, but to be successful outside the university walls, graduates must be able to think contextually. As a student of informatics, I never thought that I would need to be able to parse the syntactic structure of a sentence in my job.
And is it a piece of cake for you now?
I don’t remember exactly the way they teach you to parse a sentence at secondary school. The parse trees that help computers understand language are a bit different but the basic principles are very similar.
Does your job influence the way you read? Do you sometimes get stuck on a sentence or word and try to find a way to process it?
That happens to me all the time, trying to analyse sentences in the same way a computer would process them. I can get stuck on details that a normal reader wouldn’t think twice about. For example, if I saw a mathematical paper titled “Varsovian Models”, I would ponder over the ambiguity of the title and how to make the computer interpret this reliably as mathematical models and not fashion models.
In the academic part of your research, you focus on machine recognition of the correct meaning of ambiguous words in a text. Do you believe that one day, machines will learn how to write articles or even books and we will not be able to tell the difference?
There are already systems that can generate sports news and you can’t tell that they are produced by a machine, but these are a far cry from artificial intelligence. They use a large number of ready-made sample sentences and just fit the specific information into them. Before a computer can insert its own thoughts into a text, it has to have a sense of self-awareness; however, computers are currently very far away from that.
You did not spend your studies just in front of a computer; you were also a member of the faculty theatre group. What was it like when IT students started giving theatre performances?
A lot of people think that theatre is completely out of place at a faculty of informatics but I’m convinced that the opposite is true. One of the problems of the Czech education system is that the students are incredibly passive and unable to present their thoughts in public. This is largely because they are not asked to do that at the lower stages of education. Students frequently try and sit as far away from the teacher as possible and don’t participate in discussions at all because they are afraid they might say something embarrassing. Later on, when they present their work at a conference, they mumble and cannot sell their topic, even if it’s a potent one. I am an introvert myself and acting in front of a packed lecture room was a huge help in this respect.
So you applied to a theatre group to learn this the hard way?
That was definitely part of it, but I also just like theatre, which is why I stayed in the group for three years. My fondest memories are of my very first role in Dürrenmatt’s The Visit, where I played a perverse priest.
Do you have any other fond memories of your university years?
I had a great time at university. As a student, you have got your newly acquired freedom, you have a lot of opportunities, and your responsibilities are minimal. The only thing I had to do was pass all my exams once a semester; otherwise, it was one long joyride.
To conclude with, let’s look at the future through your eyes. What do you think search engines will look like in 2030?
That’s an incredibly complicated question. Information technology is developing in a very dynamic way. It’s not so long ago that Nokia was the giant among mobile phones. Then Apple came with its touchscreens and turned the whole market upside down. The same thing can happen with search engines – somebody will come up with an unexpected idea and everything will suddenly change. The search engine user interface, for example, has been the same for almost 20 years: an input field to type your query in and the ten most relevant pages as a result. It’s just a matter of time before somebody comes up with something better.
Would you like to be that person?
Of course (laughing).
The interview took place in 2013 for the magazine Muni Extra.