Reuters and the Japanese Evolution of a Natural Language Newsroom
The Thomas Reuters Corporation’s drive to exploit new technology to facilitate the speed of news is being challenged by a Japanese corporation that seeks to automate journalism using natural language machine learning.
Highlights from Shawn Malhotra, the VP of Thomas Reuters Technology Centre’s piece in Policy Magazine:
- Reuters was founded by Canadian Roy Thomson in 1934
- “In 1992, we launched the first commercial natural language search engine. It was the first search engine in-market to introduce probabilistic ranked searches for natural language queries—a form of machine learning.”
- “At Thomson Reuters, we now process and collect more data in a single day than we did in a month five years ago. Even as we move to the cloud, we still store 60,000 terabytes of data in our data centers.”
- “Globally, we invest more than $3 billion per year on technology. We have more than 12,000 software engineers, systems architects and data scientists…”
Compare this with the news coming from JX Press Corporation, which is reporting to have a new Fast Alert system run by robots that can “filter out 99% of false news stories,” according to 29-year-old founder Katsuhiro Yoneshige.
JX Press’s secret, it turns out, is a combination of social media and artificial intelligence. Yoneshige and his team have developed a tool, using machine learning, for finding breaking news in social media posts and writing it up as news reports. Essentially, it’s a newsroom staffed by engineers.
I’ve always wondered why there wasn’t any way to search news stores posted locally from within a small set of nearby IPs — spoofing would ruin that quickly, I suppose. It looks like the eggheads of the world have been focused instead on the worldwide harvesting of news.
According to the Bloomberg article,
“Fast Alert scours social media postings, analyzing text, photos and even exclamation marks, to find breaking news in Japan, in areas such as fires, traffic accidents and other disasters. It also monitors overseas media and Twitter accounts that it considers trustworthy, seeking to be the first to report major international developments. Once it’s found news, its algorithms write the stories.”
Like I’ve been saying — news starts with rumors, moves to itty bitty blogs and livestreams, then moves to bigger blogs before being picked up by the “real” news two days later. This process appears to be an automation of that process. But I’m sure we’ll still find a way to exploit it to our advantage, won’t we?