A beginner’s guide to the difference between machine learning algorithms and exapsecting them.
What Are Few Exampes Of Online Learning Algos In Machine Learning
Journalism on a machine. Photo by Johannes Simon/Getty Images
The idea of machine learning is scary to me. As someone who spent a good amount of time in grad school diagnosing mood swings and swings in relationship stability, I was more used to adjusting than to the constant adaptation of my own mind. Having to follow machine learning, which does not always take the information that I give and fit it exactly to myself, seemed dangerous.
Then I started working at it.
The center of journalism news at the moment is an experiment in how data can help us anticipate and predict trends in the news. Often, that requires knowing how one entity affects another—like when a big corporation tries to influence the organization that will cover it. The data exists, but analytics managers (developers who use data analytics to predict trends in the data) use it for different purposes. For some, that means crunching numbers to decide which editor to use on a particular story. For others, it’s training algorithms to pick and choose the best story from thousands to determine what gets written.
And then there’s machine learning. As the name implies, machine learning is the process of making computers “learn”—like if you teach your dog to do tricks. The idea behind machine learning isn’t that your dog will always understand how to fall for the ukulele. Instead, it’s that the machine can “play” with the data; it can learn to predict the behavior it needs in order to respond appropriately.
Machine learning is already a tool that journalists use. And yet machine learning data analysis is still in its infancy in journalism. Machine learning could someday help us anticipate what will happen based on the behaviors of people, the underlying information of public documents, and the data from your own search habits. In this moment in news consumption, where we can actively access public sources like Twitter, it is more important than ever that journalists find different ways to be informative with analysis. Whether that is the human brain, the apps that we’re already using, or machine learning—the growing use of analytics to predict things from data, whether it’s trending topics or biases in how information is interpreted.
1. Image tagging
There are signs of machine learning at The New York Times, including a recent article by The Atlantic called “Can Computers Predict What a News Story Should Be About?” The article says, “From the wars in Iraq and Afghanistan to global revolutions to natural disasters like the earthquake that destroyed Christchurch, New Zealand in 2011, how we decide what to cover has significantly shifted over the past decade. A lot of what we see online today depends on heavy machine learning, which relies on personalization algorithms to constantly crunch incoming content and plots the various benefits of each piece of news.”
Virtually all journalism starts as an image—more than you realize. The Guardian’s recent viral piece “A Shocking Invocation for Life” described how Marwan Barghouti, a Palestinian terrorist, was convicted of killing civilians in a café in Israel in 2002. But the horrifying detail that struck me was that he also tortured someone to get a confession. The image of the terrorist, the Hamas flag, the fighter, the terrorist execution—all of these could have been used by a human reporter to humanize Barghouti and inform the story. But a picture is very effective at humanizing, at making readers feel like they are getting up close to an action. A computer cannot make such humanizing elements in a photo; human reporters can. Journalists themselves must cultivate these humanizing elements in their images to make a narrative.
3. Showing versus telling
Hugh Grant is a ridiculously charming British actor who has a lot of movies you should see. But what’s more important than how he looks in a movie is how he looks in an interview. In a recent interview, he said, “If you are writing and directing as well, the excitement of the moment means more than what was happening on screen that day.” What a charming idea! You are so much more excited about what you’re talking about, and show up for a story, than what just happened.
In a very engaging New York Times Op-Ed from October called “What the CFA exam is missing,” Darby Warren, a former CFA charterholder, writes: “Equities can go up as well as down. In general, they do not turn around and start going down.” And yet there is no way to encapsulate this information with a cogent answer. That’s why a good analyst will deconstruct a news event in a way that makes sense and makes it interesting.