Japanese media teams compete to combat misinformation

David Cortez
Editors Lab Impact
Published in
9 min readApr 30, 2018
The busy Shibuya crossing in Tokyo, Japan.

In Japan, social media has both helped and hurt the process of accurately informing citizens. During the tragic Tokoku Earthquake in 2011, the nation learned the hard way how the spread of misinformation on social media can negatively effect the population. One scholarly paper wrote about the role Twitter played during the disaster stating that —

…numerous false rumors were spread, probably due to the sources of information being unclear. To prepare for future disasters, we must analyze the diffusion of information through social media as soon as possible.

In consideration of these revelations and in light of the current scandal embroiling the Abe administration, media critics have been ramping up calls to improve Japan’s fact-checking abilities. Recently, the Fact-check Initiative Japan (FIJ) has been one of the major voices discussing what it sees as a low-level of awareness in the Japanese media regarding the spread of misinformation. Their aim is to help foster the creation of new fact-checking activities in the Japanese media landscape in order to improve the dissemination of information in the digital age.

Keiko Kanai speaking at FCCJ

In a press conference at the Foreign Correspondence Club of Japan on February 8, 2018, FIJ board member Keiko Kanai explained —

We believe fact-checking is a way to counter the sort of incidents which involve false facts. Fact-checking is something that has not been actively carried out here in Japan, so in order to make up for this, we created this initiative.

Speaking on the same panel, FIJ founding member Yoichiro Tateiwa explained that quality fact-checking requires hard evidence and rational behind every claim, and it is this very thing that the Japanese media struggles to provide. He stated —

You will notice in almost all cases here in Japan, the newspapers and television do not offer any type of evidence at all.

In this media climate, it is hard for average Japanese citizens to know if they are getting accurate information. With this issue clearly in need of addressing, the Global Editors Network (GEN) with support from COPILOT held its second Editors Lab competition in Tokyo to encourage Japanese media members to develop new and effective ways of fighting misinformation.

Tokyo Editors Lab

On April 21-22 , 2018, eleven teams from across Japan — made up of one journalist, one developer, and one designer — were asked to create media prototypes to counter mis/disinformation. They spent the first few hours of the competition listening to expert lectures meant to inspire the direction that their prototypes would eventually take.

Researcher Ben Decker speaking at Tokyo Editors Lab on the role social media plays in the spread of misinformation.

BEN DECKER

The first lecturer, Harvard Kennedy School research fellow, Ben Decker, wanted to impress upon the teams that no true solution to misinformation could be created without first understanding the role of social media in the production and diffusion of misinformation.

Decker explained that the spread of misinformation online is very similar to a pathogen seeking to spread from host to host. Misinformation starts in ‘closed networks’ such as 4chan, Reddit, and the dark web, and eventually is pushed onto ‘open networks’ such as Facebook, Twitter, and YouTube where it can then go viral and be absorbed into the population.

One of the more exciting things for computational scientists is to look at how to track a piece of content from closed networks to open networks,” stated Decker, a clear nudge to the programmers in the audience to create prototypes that can be useful in this regard.

HITOFUMI YANAI

Hitofumi Yani from GOHOO speaking at Tokyo Editors Lab

The second speaker, Hitofumi Yanai, is the founder of Japan News Coverage Verification Organization (GOHOO) as well as a FIJ board member.

Yanai stressed that Japan is lagging far behind Western nations in the creation of effective misinformation fighting technology and that he would like for the teams to seriously consider emulating current automated fact-checking software such as FactPopUp and ClaimBuster to create a simple interface that everyday citizens can make use of.

Yanai stated, “In the United States the fact checking technology has come very far. Japan needs to have something similar. We don’t want to be swayed by the wrong information, and fact-checking is the right way to make sure that the information we have is accurate. We want to focus only on the facts, regardless of political opinions.”

In closing, Yanai asked the teams to notice the gap between fact-checking organizations and the average news consumer in Japan, hoping to inspire them to create prototypes that could bridge this gap

IRENE JAY LIU

Irene Jay Liu from Google News Lab speaking at Tokyo Editors Lab

The last speaker, Irene Jay Liu, the Google News Lab Lead in APAC, spoke about the large variety of tools already at a newsroom’s disposal to help combat misinformation in their reporting. She discussed simple Google search tricks to how to check meta data on an image, using Google Earth for verifying locations, and taking advantage of Google’s new A.I. image analysis tool — Google Cloud Vision API — which evaluates images for authenticity and contextual data.

Prototype creation

After the lectures, teams from NHK, Asahi Shinbun, Ryukyu Shimpo, Tokyo Shimbun, NewsPicks, YAHOO! Japan, FujiTV, Gunosy News, and Nippon Broadcasting System all went to work creating their prototypes to be judged the next day by a jury of news critics and scholars.

Programmer for NHK’s team diligently coding before the prototype submission deadline.

The Prototypes

1. Consider the impact of social media on misinformation

  • Fake News Vaccinator: a browser plugin that allows readers to ‘vaccinate’ dis/misinformation by injecting URLs with hypothetical vaccines. Users who who discover fake news click a small syringe icon in their plugin and the URL gets tagged as fake and then posted to the “Fake News Vaccinator Twitter” account. The idea here is help dilute the amount of unverified news on the Twitter platform by flagging what is known to be fake. (built by NHK)
  • Disruption Finder: a tool that seeks to prevent echo chambers on social media by collecting information about the demographics of the readers specific content across major news webpages. The aim is to take expose such information to users in order to help them better who is more likely to be reading what kind of news. (built by Gunosy News)
Image of Disruption Finder aggregating demographics.

2. Bridge the gap between reader and fact-checkers

  • FactCheckers: a mobile game that trains users to have greater media literacy by asking them if what they are reading is correct, partially correct, or false. The idea behind this is to motivate people to ask questions about the news they read or hear on the television. (built by Asahi Shimbun)
Promotion for the FactCheckers mobile game
  • Sesshu: a browser plug-in that dims the text of dubious sentences in articles being read online — as depicted below — by using working journalists to rank the quality of statements based on their knowledge of the facts. This prototype received a special mention from the jury thanks to its user friendliness. (built by FujiTV)
Image of Sesshu highlighting and dimming articles on your browser
  • Press Dash Button: a website that collects verified stories from rural communities and helps these communities stay informed. The site also has a feature that allows users to directly contact a journalist if they wish to bring attention to matters in their region. (built by Tokyo Shimbun)
  • DOKUMI: a mobile app with a ranking system of news sources enabling users to choose trusted newsrooms when seeking information. The app scores news outlets from -2 to 3 for trustworthiness, with 0 being neutral. (built by YAHOO! Japan team B)
Image of DOKUMI’s user interface, rankings are shown at the bottom of each article.
  • FactChecks: a tool that collects comments on articles and news from social media in order to track what is trending and what needs to be fact checked. This helps fact-check organizations focus on looking for counter information, choosing which articles to review, and delivering the fact-check article to the right audiences. (built by NewsPicks team A)
A description of how FactChecks uses social media comments to quickly locate misinformation.

3. Improve fact checking

  • Courtroom: a plug-in that uses the google API to extract keywords to assess polarization. The tool aims to inform readers of potential bias. (built by YAHOO! Japan team A)
  • PICK++: a webpage that groups news in related topics using keyword ranking software. This allows users to easily view various news sources that are running stories about a particular topic in order to absorb the fuller range of opinion — be it reliable news or something dubiously published.(built by NewsPicks team B)
Image of the Pick++ webpage grouping related news
http://community.globaleditorsnetwork.org/projects_by_city_editorial/3577

The Winner

Jury member Daisuke Furuta announcing the winning team

Nippon Broadcasting System, a radio station, created a plug-in called CATALANA that takes quotations cited in articles and locates the original audio source of the utterance using text matching and speech to text analysis.

The team proceeded to show an example of a politician that was being slammed in the mainstream media for comments made on a radio station that seemed inappropriate. The team ran the quotations of his alleged comments through the plug in and located the audio source, which allowed them to see the whole context of the original utterance. It turned out that much of the original context had been lost to the ‘quote-mining’ that took place after the original radio interview. Having to power to check nuance, context, and quickly find an audio source is a remarkable achievement of design given the time restriction for the Editors Lab event.

Nippon Broadcasting System (left — Hirofumi Hiranuma, middle — Shingo Sawada, right- Tensei Kanasugi)

CATALANA best exemplified the Editors Lab innovation criteria in that what they built has extremely practical real-world application.

Jury member Daisuke Furuta explained it as such:

The winner, Nippon Broadcasting System, the editorial quality of their product is very good. It is useful not only for users, but also for fact-checkers because they can link some quote to the original audio source. So, because it allows anyone to check whether something is truthful, it is useful for all sides. It is not about print media versus audio media. Most news content is made from a spoken comment from a politician or VIP, and we really should check the audio to see if other media is reporting something correctly.”

The team leader, Shingo Sawada, was quoted as saying:

Because we are from a radio station, we are very familiar with the issue of someone taking parts of audio out of context and attempting to make a fake news with the segments. This kind of practice creates a lot of misinformation out there, and this was what we wanted to solve. Regarding our medium, there is a lot of news that is also considered as entertainment for listeners, not only as hard news. So, the speakers are sometimes a little candid in our forum. However, if someone takes only part of their speech and they publish it publicly elsewhere, they can make guests views seem extreme. Because our medium deals with entertainment rather than hard news — such as politics and business — we have become especially aware of this issue.

The Nippon Broadcasting System team will be traveling to Lisbon, Portugal for the GEN SUMMIT on May 30, 2018 to bring CATALANA to compete with prototypes from other 2018 Editors Lab finalists. Best of luck to them.

--

--