Personalize News

A COMM 281/CS 206 Project
by Kaveh Danesh, Jihyeon Janel Lee, André Natta, and Titus Plattner

About the Project

When we tell stories, we don't tell them the same way to our grandparents and our college friends. Depending on whom we're talking to, we vary the amount of context, emphasize certain details, or tell an inside joke, all in an effort to make the story more relevant and engaging for the listener.

Today’s news organizations know a lot about their consumers. And they could potentially know a lot more given the large amounts of data generated online. But they continue to use a one-size-fits-all model—presenting the same article whether the reader is young or old, female or male, from the United States or abroad—thereby missing significant opportunities to engage existing readers and attract new ones.

Our project helps solve this problem in two ways. One the writer end, we offer a tool that makes it fun and easy to personalize content for readers of different characteristics. And on the user end, we input a consumer's data and output a personalized article.

We believe our tool can help newsrooms better inform readers by making their content more readable, relevant, and engaging. Ultimately, we hope to change the way news is written and consumed.

Lessons Learned

Each week, our group went to class with a specific vision for our project. And after receiving feedback from the teaching team and our classmates, we left with a slightly (or sometimes very) different vision.

This was not always easy. After working on a particular idea or prototype all week, we would grow attached to it. In-class feedback wasn’t necessarily pleasant to hear.

But over time, we grew accustomed to the highs and lows of product development. We learned to not be attached to any particular version of our prototype, to listen carefully to feedback, and to pivot as necessary.

For example, an early version of our tool focused on suppressing unnecessary context (so that readers knowledgeable about politics would see “Mr. Ryan” instead of “Mr. Ryan, speaker of the House of Representatives”). At the time, this seemed like a useful and scalable contribution. But the teaching team felt we weren't thinking big enough and sent us back to the drawing board, where we first vented our frustrations but eventually came up with a more general form of personalization.


The idea is as follows. Suppose you’re a journalist writing an article on how breast cancer is one of the most common cancers among women. You want to provide female readers with information on how to conduct a breast self-exam, but you realize that male readers would benefit little from this information; they might like to know what cancer is most common in men. So you write two versions of the sentence and agonize over which to include.

We considered this example, and then we realized: why should you have to choose? We could build a tool that lets you include both versions, and gives readers the one that matters most to them. That way you can have your cake and eat it too.

Experience working on an interdisciplinary team

Two journalists, a computer scientist, and an economist walk into a room. What happens next?

While this sounds like the start of a bad joke, it isn't all that far off from our first day as a team: four nearly perfect strangers meeting on the fourth floor of McClatchy Hall.

What happened next was subtle. The journalists described the elements of writing a story. The computer scientist described what could and couldn't be accomplished with code. The economist described the inner-workings of supply and demand.

We listened. We learned to think more like each other. And we had fun doing it. That enabled us to apply the most powerful aspects of our respective fields toward a common goal.