Radboud Universiteit
Zoek in de site...

Looking for a search engine that does not know everything about you

Datum bericht: 3 september 2019

If you search for something on the internet, you use Google. The fact that this allows the company to learn quite a bit about you is bothering more and more people. Can it be done differently? Certainly, thinks Djoerd Hiemstra, professor of Federated Search at Radboud University. If you allow different independent search engines to work together, you as a user do not hand over all your data to a single party, and thus reduce the risk of misuse of your data.

Djoerd Hiemstra

If you let Google answer every question, that search engine will learn a lot: about your health, your finances, your sexual preferences... Google also knows where 'home' is via other apps, such as Google maps, as well as where you like to go out. Before you know it, a single company is controlling your behaviour, or selecting the answers to the questions you ask.

Search engine expert Djoerd Hiemstra feels rather strongly about this: “In the Netherlands there is one monopolist in the field of search engines: Google. Now, I have no problem with that company – they make amazing things – but it’s becoming a problem in our society that one party possesses so much information about so many people; and that that party can also determine which answers you get to your search questions. That can result in not everyone having the same information and that is undemocratic.”

Collaborating search engines

Hiemstra conducts research into federated search engines, which pose fewer risks in terms of manipulation and privacy. Such a search engine consists of different independent search engines that work together, without one machine having all the data. That is different from how Google works: it downloads all the data it can find on the internet to its own servers, where it is easier for them to search.

That method is not necessary for the correct search result, says Hiemstra. “Imagine that I want to take the train from Nijmegen to Rotterdam tomorrow and I have to be there by 3 p.m. I could ask Google that question, but I actually don't need much data from railway operator NS to find out what time I have to leave and if there are any particulars I should be aware of. Via the NS app I can go directly to their search engine with my question, and then only NS and I are aware of my plans – and in principle no unsolicited offers from the hospitality industry in Rotterdam pop up. Unless, of course, NS use Google applications in their app.”

“Email is another nice example of a federated system. My employer, the university, has an email server and I can simply email someone with a mail address from another provider without being offered advertisements. The great thing is that you can’t buy email, because nobody owns it.”

In practice: Searsia

Hiemstra’s research is not all theory: in 2016 he developed a prototype of a federated search engine, Searsia. The University of Twente, where Hiemstra worked until this summer, uses this machine for its own website. “Searsia is not web scale – you cannot answer as many questions with it as with Google and it is, therefore, certainly not yet an alternative. But Searsia is self-learning, so the more it is used, the greater the scale on which it can work. I will spend less time on Searsia in Nijmegen, but here, too, my research is about developing models and methods to improve such systems.”

Contributing to a better world

Hiemstra has come to Radboud University because here he has more colleagues doing research on search engine technology. The Artificial Intelligence & Data Science research group is also affiliated with the Dutch Responsible Data Consortium, and responsible data handling is a goal that Hiemstra wholeheartedly endorses. “I’m no stranger to idealism. Look, you can use your knowledge of statistics and technology to get better advertising to people and have a very nice job doing that. But as an academic researcher you have the ability and the responsibility to explore other directions with your research. We have to contribute, or at least want to contribute, to a better world. So if my research can help break through the power monopoly of the internet and social media giants, if we can show that there is another way, it’s worth it.”