• Technical solutions are only worthwhile if they benefit the user
News

Urška Herbst is a mother, athlete, dance instructor, natural scientist, product manager at Zemanta and co-organiser of ProductTank in Slovenia. She graduated from the Faculty of Computer and Information Science (FRI) with an undergraduate thesis titled ‘Predicting authors’ emotional orientation in online comments’. In this interview, she talks about the challenges she faced at FRI and those she faces now at Zemanta.


Why did you decide to study at FRI?

I’d loved natural sciences at primary school, but made my final decision when I was at Gimnazija Vič [grammar school]. This was partly down to Alenka Krapež, who was then an information sciences teacher and is now head teacher, and who presented the subject in a very interesting and inspiring way, and partly down to the fact that it was starting to become evident that computer science was the profession of the future. So the best option seemed to be to study something one enjoyed and something that would train one for a promising career.

 

What was the structure of study at FRI?

It was 80% mathematics in the first year, which was great, and 20% programming. I struggled a bit with the programming, but during my second year the process of converting problems into a mathematical-logical model suddenly became clear to me. From then on, things became a lot easier. I always saw programming as a tool for achieving other, higher goals, which is why I chose bioinformatics as my fourth-year option. I was fascinated by the fact that you could use programming knowledge for the benefit of mankind, for example by using the human genome to predict who was susceptible to a particular disease.
 

It was down to competitions held on the Kaggle platform and to Prof Blaž Zupan that I became more and more interested in extracting knowledge from data. His lectures were always given with such enthusiasm that I simply devoured the subject, and wanted more and more of it. Under his supervision, I conducted a semantic analysis of online comments for my undergraduate thesis. I selected articles from the rtvslo.si website and tried to predict whether the comments would be positive or negative. Using certain identical phrases, my system was able to recognise related and different articles very well, but I did not get the desired results when I analysed the comments. Online commentators mostly do not use ‘official’ Slovenian, they write in a very superficial manner with lots of typing errors, and they are often sarcastic – something my system was unable to recognise. It was also the case that the comments were mostly not exclusively positive or negative, but in one of the many places in between. In this case, logistic regression showed that my system was only a little better than random selection. It correctly identified just over 60% of cases, which meant that it was still mistaken in just under 40% of cases. It was a good thing that I wasn’t attempting to solve a more serious problem, with lives depending on my predictive model.

 

Computer science is a wide field, so it is important to find a narrower, more specific direction during or after one’s studies. Since this is such an interdisciplinary science and is woven into every day of my life, I finally understood what the dean had repeatedly said to us – that study primarily helps you to develop a way of thinking. So in every challenge, whether in my work or private life, I try to tackle problems in an interesting and different way.

 

How did you hook up with Zemanta?

Zemanta were always very well-known to me as an innovative and advanced startup in which you not only created a career, but also and at the same time ‘created yourself’. The people at the top, Andraž Tori and Boštjan Špetič, are inspirational in themselves on account of their knowledge and way of thinking. And because I like to learn, I like to be surrounded by people who know how to motivate me. Although I had been very happy in my previous job, I was delighted to accept Zemanta’s invitation. This was a tech company, which meant that I would be able to put my knowledge to good use and to improve it. There was some overlap too, as I had previously worked in a sector that used products such as those created by Zemanta. When planning and building new functionalities, I find it much easier to identify with the user because I used to be one myself.

 

What do Zemanta do?

In advertising, you obviously want to reach as many people as possible. Digital publishers come together in various networks, each of which has its own reach and traffic volume, and its own platform on which advertisers create advertising campaigns. Zemanta are developing a platform that enables advertisers to straightforwardly distribute advertising campaigns across the more than 70 platforms of which we are part. This increases reach in a very simple manner, and reduces the amount of manual and bureaucratic work required. There is no need to enter or amend data in the 70 systems, but just in one. This makes it easy to optimise the financing and results of advertising campaigns.

 

How would you describe the role of product manager?

As product manager, you follow trends on the market, talk to customers and keep an eye on the competition. A product is not meant to be ‘standalone’ but to be of use to customers. You identify needs or problems on the market and try to describe and define them. Only then are the engineers able to design solutions to resolve the problem or meet the need. You can have a great technical solution, but every minute devoted to it that does not resolve a problem, that makes the solution unusable for users or makes users’ everyday lives more difficult is a minute wasted.

 

What is ‘native’ advertising?

Native advertising is advertising that melds together so well with the content of a webpage that the user does not notice it, thereby making the advertising more effective. Adblockers exist because people get annoyed by classic advertising banners, as they disturb a user’s experience, or even bring it to a premature end. In this way, we ensure that readers who reach the content they wish to read free of charge can coexist with advertisers aiming to reach audiences in order to promote their products or services via the content of the page.

 

Why do you do what you do?

‘To build products that people love’. Helping to create a good product that makes users’ lives easier or solves their problems gives me great pleasure. This is particularly tough in the advertising industry, which makes it an even greater challenge.

 

(This is a promotional article from Zemanta d.o.o., the main sponsor of the 2018 FRI Summer School.)