Prototype winner using ‘natural language processing’ to solve journalism’s commenting problem

technology / Article

Photo of Francis Tseng courtesy of Liya Safina. 

In November, Reuters was among the largest news organizations to announce that it will close online user comments on news stories. It wasn’t a huge surprise in our contemporary digital environment where the comments section can often disintegrate into the depths of politically misguided, racially insensitive and profanity-laced conversation.


"Prototype Fund winner Max Ogden building a data-sharing ecosystem" by Vignesh Ramachandran, 03/26/15 

"Prototype Fund wnner aims to break 'cycle of poverty' with social services information platform" by Vignesh Ramachandran on Knight Blog, 02/12/15

"Knight Prototype Fund supports 10 new data & storytelling projects" by Chris Barr on Knight Blog, 02/10/15

Knight Prototype Fund winner Francis Tseng wants to change that with machine learning and artificial intelligence.

Tseng, a Knight-Mozilla OpenNews Fellow, received $35,000 in support last year from Knight Foundation to build a prototype for his news aggregation application, Argos. Argos seeks to automate the creation of timelines out of news events and news stories. So instead of a person manually curating how content fits together, the right kind of automation could allow a computer to recognize articles from the same event and organize them.

With the funding, Tseng worked on improving Argos. He says the original idea was to package Argos as a consumer app, but now he plans to make it an open service that others can build on. The secret sauce behind his technology is natural language processing – which Tseng now wants to use to solve news organizations’ commenting dilemmas.

“For me, the biggest takeaway wasn’t the [Argos] software itself but that I really had an opportunity to familiarize myself with a field [journalism] that I didn’t really know much about and now has become sort of my main passion, my main interest,” he said.

Tseng earned his bachelor’s in cognitive neuroscience from the University of Pennsylvania in 2011, and then worked for a year as a product manager for a company in Beijing. In 2012, he moved over to renowned design and innovation consulting firm IDEO in San Francisco and New York, where until last summer, Tseng says he worked on design problems in various industries. Now, he wants to apply those chops to journalism.

“Journalism is particularly interesting because it’s sort of a perennial and very important institution in any society,” Tseng said.

As a 2015 Knight-Mozilla OpenNews Fellow, Tseng will be embedded this spring at The New York Times. He’ll be working on The Coral Project, a Knight Foundation-funded collaboration between The New York Times, The Washington Post and Knight-Mozilla OpenNews. The idea is to improve communities on news sites with open-source software.

Since The New York Times’ comments are human-moderated, Tseng is looking forward to working with the Times’ high-quality data set of what makes a good comment and what makes a bad one. Eventually, Tseng hopes, technology can auto-filter the offensive comments on news sites to save moderators from what can be an “exhausting process.”

At the Times, Tseng wants to get a better sense of the news production process and to see how Argos works in a newsroom – an environment he looks forward to experiencing first-hand.

“It’s kind of weird building journalism products from outside without really any sort of intimate knowledge of how that world works on the inside.”

Vignesh Ramachandran is a frequent contributor to Knight Blog.

The Knight Prototype Fund accepts applications on a rolling basis to help take ideas from concept to demo. The next deadline is May 15, 2015. Apply here.

Sign up for our newsletter

Submit your email. Receive updates and the @knightfdn newsletter.

Subscription Options

What does a post-broadcast world mean for public media?

technology / Article