Back

Humanist takes on the computer

Post-doctoral researcher Tuomo Hiippala studies, for example, photos of Helsinki uploaded to Instagram by tourists. Photo: Heta Koski

Humanist takes on the computer

Heta Koski

Post-doctoral researcher Tuomo Hiippala is clearly enthusiastic. Before I even have time to ask, he is already singing the praises of the CSC's cPouta cloud service he uses: "All I needed to do was to fill in one form and suddenly I had all the resources I needed at my fingertips!"

Hiippala has not been using information technology more powerful than his laptop to help him in his research for more than about a year, but he is already talking like an old hand. At first glance, you wouldn't believe that it is, indeed, a born and bred humanist sitting in front of you, denying being that interested in computer science, although terms such as programming languages, servers and algorithms trip easily from his tongue.

Hiippala, currently working at the Centre for Applied Language Studies of the University of Jyväskylä, is particularly interested in the interaction of language and imagery in various documents. For his dissertation, approved in 2013 in the University of Helsinki, he collected a unique dataset that is thus far the most extensive one applying the Genre and Multimodality model developed in the University of Bremen. The openly published research data[1] comprises travel brochures published by the City of Helsinki between 1967 and 2008, and their contents, appearance, layout and interrelations, have been defined in a database using the XML markup language.

Computer vision and machine learning

This summer, Routledge published a book written by Hiippala based on his dissertation research[2] that discusses precisely the structure of multimodal documents and their research. "The book helps researchers who are planning to collect or research multimodal corpora", Hiippala describes. He says that he himself learned to program on his own through trial and error. So, what does a humanist do with a cloud service intended for scientific computation?

"In my latest study, I applied an algorithm that compares the similarities between two images.  I used it for an analysis of the layout of bilingual texts. For example, by placing the bilingual content on both sides of a spread, readers can be directed not to pay attention to the other side, if their language skills are not sufficient to understand both. The person who originally presented this hypothesis was extremely pleased when I contacted them and said that I had been able to study this empirically as well", Hiippala recounts.

"Naturally, I could have gone through my entire data set comprising 1,500 spreads of Finnair customer magazines manually, but it is easy for a computer to analyse symmetry. So, why should I not let a machine do it for me?", Hiippala asks. In his view, the need for external computation resources was apparent – an analysis that took up to couple of hours on his own laptop is completed in ten minutes on CSC's servers.

The best ideas come from unbridled experimentation

The best thing about the cPouta cloud service isn't even the powerful computation resources; it is the ease and freedom of its use.

"It's great that starting to use it has been made so easy. The use is free of charge for university researchers, and you get to try things out immediately after getting your access rights. I have had my best ideas only after I started using the service, as I've been able to freely test and try out my ideas. The browser UI is so flexible that it feels just as if I were working on my own computer – everything is just much more powerful", the researcher rejoices.

At the moment, Hiippala says, he is primarily developing methods the use of which could be expanded in research in the future. He has even been tempted by a return to his old dissertation materials, although he confesses that, after several years, he has become somewhat tired of travel brochures. However, with new resources, new perspectives could be found in old data as well. Machine vision, for instance, could be used to quickly determine whether Helsinki Cathedral appears in the brochures more often than Uspenski Cathedral.

Selfies from Senate Square are next under the microscope

Next, however, the researcher will set his sights on the social media. In early summer, the City of Helsinki gave him a grant for analysing photographs shared by tourists on the Instagram service. The purpose is to find out what tourists wish to show to the world of Helsinki.

 

"Let's take a beautiful summer day or week, and take a look at all the photographs taken on Senate Square and shared. Do people take photographs of the church or of other people; are they landscape photos or selfies, or something completely different? The data will include thousands of photographs", Hiippala envisions.

In addition to academic interest, the research also sees a commercial potential in the machine vision and humanist tourism research applications. He is unwilling to reveal all of the business ideas formulating in his head, but he gives one hint: "By developing this method, tourists in a new city, for example, could be recommended to visit sites based on what kinds of sites they have photographed in their home town."

So, there you go, enthusiastic software start-up entrepreneurs – contact Hiippala, he is sure to tell you more.



comments powered by Disqus