About the Project
'HSE University's Age-Mates'
2022 marks the 30th anniversary of the founding of HSE University. Many of the university’s peers—those born in 1992—now work and study here. Thirty-year-old HSE graduates work in various fields, from business and fintech to IT and contemporary art. As part of the new ‘HSE University's Age-Mates’ project, some of them have shared their stories and talked about what they like about the university.
In his youth, HSE University graduate Sergei Mikhaylov wanted to become a diplomat, but after earning a degree in history, he ended up working in computational linguistics. In this interview with Age-mates, he talks about working with pharmaceutical companies, John Austin’s speech act theory, and how one infinity can be greater than another.
How did you end up at HSE University?
I was born in the town of Balakovo in the Saratov region and dreamed in my youth of moving to St. Petersburg. This dream stayed with me when I was choosing a university, although I also considered going to Moscow. I had the strange idea of becoming a diplomat, but I wasn’t admitted to the Moscow State Institute of International Relations. However, I was among the first group admitted that year into the Department of History at the HSE University campus in St. Petersburg. That was in 2012. I felt as though everything was falling into place perfectly and that the HSE Department of History was a good alternative in terms of education. However, over the course of my studies, both my values and my understanding of what I want to do have changed dramatically. By the end of my first year, I had completely forgotten about the plans I had made as a school student.
Are you still interested in history?
That’s a difficult question. It’s one thing to take an interest in history at school, watch history programmes on TV, read something on the subject, and play video games with historical settings, but quite another to study history along with other early-admission students at a department that takes a very scientific approach to history. It was interesting, but not easy: there was a lot of homework and many projects.
Sometime during my second year, I tried to arouse the scientifically-minded historian in myself. I needed to find a topic that I could study my whole life—that’s exactly what was expected of us. As a result, I wrote my term papers and undergraduate thesis on the Russian Empire’s role in the creation of the Universal Postal Union. I can’t say that I found such academic work very interesting, although I did work as both a research assistant and a teaching assistant for a long time. It was pretty good, but I didn’t want to do it my whole life.
I think in practical terms. Whenever I do something, it is important for me to see the end result. And in science, there is no endpoint, no boundary: you can study something forever and never get to the bottom of it. Obviously, every researcher contributes to the global sea of science, but this isn’t a way of life that satisfies me.
Still, I took a regular part in all of our department’s academic activities. And, while working in a research and educational group for creating geographic information systems, I met with computer linguists from the HSE University Moscow campus. This inspired me to try out the field, so I joined the master’s programme in Computational Linguistics at HSE in Moscow.
What did you like about it?
First of all, I liked the topic of learning languages, the apparent connection between consciousness and language. I found this topic very intriguing, even when we studied John Austin’s speech act theory or the works of Ludwig Wittgenstein as undergraduates. What’s more, computational linguistics is no longer about a philosophical approach to language, but about the practical aspect: for example, you go on an expedition, collect data, and then digitize it and build analytics on it.
I also like the symbiosis of using digital methods to study the humanities. We took mathematics in our graduate programme, and it really expanded my awareness: I began to think about things that I had never thought about before. In our first math class, we compared two infinite sets of numbers and I remember driving home and trying to grasp the idea that one infinity could be greater than another.
Did you know how to program?
I began learning Python as a fourth-year student, but of course, my fellow students in the master’s programme had much more serious backgrounds: some had switched directly from technical majors. But I made a dramatic shift in the focus of my life and took the chance to learn something new. The first year of my master’s degree, I tried hard to get into the swing of things, to catch up to my classmates. It was tough, but I discovered a whole new universe of digital methods, programming, and linguistics as a science. I got a big kick out of the process. I liked learning new things, learning how the language works, and how it was studied. I found it all fascinating and exciting.
What was your first job as a computational linguist?
Early in the second year of my master’s programme, my research advisor, Irina Efimenko, invited me to come work with her. This was when her company, Semantic Hub, had started recruiting people for the computational linguistics department. I was one of the first employees and haven’t changed jobs since. What does our department do? In short, Natural Language Processing (NLP) for medicine. Put differently, we write programs that extract medical facts from open source texts (social networks and forums) in different languages.
You’re saying that you can write a program for a specific language without even knowing that language?
That’s the whole point: a computational linguist doesn’t have to know the language like a native speaker. It is more important for us to know how the language works, what methods we can use to analyse it, how we can approach it technically in order to extract or recognise something.
When we work with foreign languages, we usually bring in language experts. We ask them to give us a crash course on the language and tell us the things that interest us. First, how does the morphology work? Syntax? Are there any peculiarities associated with slang expressions in certain topics or language constructions that change meaning based on the medical context? We also evaluate the language quantitatively. If necessary, we can calculate the frequency of words and recognise specialised vocabulary. In addition to working with experts, I use online translators and read breakdowns of the grammar. You need all this to write programs.
Did they teach you this at HSE University?
In computational linguistics, we studied various methods that enabled us to do this. We usually worked with Russian and English, but other languages were also used as examples. Python has libraries that make it possible to work with very different languages. If you have mastered such a library for working with English, there’s nothing to prevent you from immediately switching over to working with, say, German.
Who needs your services and why?
Our main customers are pharmaceutical companies. Let’s say Company X is about to launch a new drug for disease Y. To make the launch successful, Company X wants to collect as much information as possible about patients: what their needs are, how disease Y affects their quality of life, what specialists they see, what symptoms cause the most discomfort, etc. With these questions, Company X comes to us. What do we do? From open sources where people discuss these topics, we ‘crawl’ (collect) large amounts of textual data and write programs that extract different facts. For example, a person writes that there is no specialist in his city on disease Y. This is an example of a fact that we extract, along with many others, and then our analytical department builds analytics, infographics on these facts and passes this to Company X. Pharmaceutical companies use such analytics as part of their decision-making process. They can use this data for their own purposes: for example, to acquaint doctors with specific features of the patient’s vocabulary in order to ask them more accurate questions about their condition. Or decide that a region should be supplied with larger or smaller volumes of a particular drug.
Do you have competitors or is this a unique offering on the market?
There are companies that do things which are similar, but not exactly the same. Our service is very complex and comprehensive, so we manage to avoid competing directly. In general, pharmaceutical companies are only just beginning to really use the possibilities of big data; the interest in this is really significant.
But the idea that you can analyse open data from the Internet and gain new knowledge is not exactly new. More and more companies and industries are realising that the Web is an important source of useful and commercial research information that can be tapped. However, it is not widespread yet. So for the pharma industry, we are innovators.
What sort of things are your fellow master’s students doing?
We’re all into NLP or related topics, so we’re still on the same wavelength in terms of friendships and work experience. We sometimes ask each other for advice, despite the fact that we work in different areas. Even if I work with text and medicine and my friend works with pictures and neural networks, we can offer each other insights and new ideas.
Do you feel anything was lacking in your education?
I had to put in so many hours of study during the bachelor’s programme that I sometimes missed participating in student life. I played What? Where? When? and attended various student events, but our faculty was separated from most student activities. This was also because our classes were held in several rented offices near the main educational buildings and we went there as if we were going to work. But I wanted to be close to the other faculties, to eat in a large common cafeteria with others, to walk down corridors crowded with other students... But it took time to set all of that up. Student life was better in the master’s programme: we formed a wonderful community and are still friends.
Did you feel that the Higher School of Economics campuses in Moscow and St. Petersburg were part of the same educational institution?
Yes. LMS is the same everywhere—the same standards, symbols, and merchandise. Teachers from HSE Moscow came to our history department, and we cooperated with Moscow students for the work our scientific and educational group did. But the Moscow campus feels more strict; the St. Petersburg campus has a feeling of greater freedom, in the broadest sense of that word.
What would you wish HSE University on its 30th anniversary?
I would like HSE to remain a university that isn’t afraid to take on new challenges in sciencee, and that loves and appreciates innovation. And that this desire for innovation always paid off and kept HSE on the path of progress.
How would you describe a typcial HSE University student?
He or she is a person with an inquisitive mind who is never satisfied with standard answers and is always looking for ways to learn something new, even about seemingly ordinary things. Taking a scientific approach, using a critical eye when examining the source, and looking at every issue from all sides—these are the characteristic features of an HSE student.