A washing machine can learn

The 50 most important research questions for the Netherlands


Can machines help us to create knowledge from mounds of information?

Can machines help us to create knowledge from mounds of information? Yes, is the answer given by Jeroen Donkers, a knowledge engineer who works as a lecturer for the department of Educational Development & Research at the Faculty of Health, Medicine and Life Sciences. “It is already happening on a large scale, every branch of science is using it. The strange thing is that as soon as a technique from the field of knowledge engineering becomes public property, people forget that this is where it has come from. Take Google, for example. From billions of pages of texts, on numerous unrelated subjects, Google retrieves data in a second. There is a whole lot of clever computer technology behind that.”

The question is, Donkers says, whether you can call a list of search results "knowledge"? Would that not be just information? Or is it data? “Data, or loose numbers, words or fragments of text that only becomes information when you give it meaning. Imagine you measure the lengths of a number of individuals. Then you get numbers. If you interpret those numbers as the lengths of persons, you have information. To acquire knowledge, we have to go one step further. If you subsequently compare the lengths of men to the lengths of women, calculate averages, and so determine differences, then that is a type of knowledge. The more information you can connect, the more knowledge you obtain. It is all about conclusions, deduction and predictions.”

Can a computer acquire its own knowledge? “It is difficult for a computer to interpret, but it can process data in such a way that we can use it as knowledge. These days it is possible, for example, to input a large number of scientific articles and instruct the computer to look for new connections. This is called text mining, which is a subcategory of data mining: a method of filtering probable hypotheses from data. You can see examples of this in molecular genetics and bioinformatics. Researchers then have to try and actually demonstrate that there is such a connection. This is how knowledge is created.”

Donkers points at another skill that computers have: they can learn. “Humans give it the capacity to learn, but then the technology gets down to work. An example is the thermostat that can calculate independently what time to start the heating so that the living room is warm at seven o’clock in the morning. Or the washing machine that measures the temperature of the water and knows how long the washing programme should be.”

Dealing with tons of information, that is what computers are good at, Donkers emphasises. For example, technology helps the police to search the PCs of suspects in child pornography cases. But one can also use knowledge engineering to present large amounts of data in more useful ways. Donkers types on his keyboard and shows the Gapminder World site. A moving graph shows in approximately twenty seconds the development of prosperity in the different continents from 1800 until today. “At a glance.”





Riki Janssen

The Royal Netherlands Academy of Arts and Sciences (KNAW) made a list of the 50 most important research questions for the Netherlands. Every week a UM researcher gives an answer.

A washing machine can learn

Add Response

Click here for our privacy statement.

Since January 2022, Observant only publishes comments of people whose name is known to the editors.