The COLUMBIA JOURNALISM REVIEW explains machine learning with examples of how it’s being used in newsrooms.
Author Nick Diakopoulos describes four ‘flavours’ of machine learning. Each varies by the amount of input by humans. For example, in supervised learning, the data to ‘train’ the system needs to be labelled. The labels can be as simple as naming a document ‘interesting’ or ‘uninteresting’ but nonetheless need to be there as clues, helping the algorithm assess what kind of future documents are a match for ‘interesting.’
The other three forms require less human input, in varying degrees. One of the challenges in journalism, he suggests, is matching the flavour to the objective, beginning with the question, will any be useful?
Diakopoulos points to bias and uncertainty as potential pitfalls, again using examples from news work.
- Nick Diakopoulos is a leading figure in the field and this is an authoritative summary worth saving. He’s a professor at Northwestern, a fellow at Tow, presented a course about algorithms in journalism funded by the Knight Foundation, and has a book on the subject forthcoming.
- Bias is widely named as a weakness when working with algorithms.
- Uncertainty is less often flagged, yet equally important for journalistic uses. Findings by learning-based algorithms need validation, too, and as their insights become more sophisticated, verification will become more challenging.
‘I’m bullish on the capabilities and opportunities that machine learning presents to editorial work, but also cautious enough to remind readers that machine learning is not the answer to every journalistic task.’
What is machine learning and why should I care?
COLUMBIA JOURNALISM REVIEW | April 25, 2019 | by Nicholas Diakopoulos