Introduction to Cofacts @ Google News Lab APAC summit 2017

Introduction to Cofacts @ Google News Lab APAC summit 2017


Everyone, um, my name is Johnson I am a web developer that
works for an internet company but in my free time, I participate in g0v, it’s an online community in Taiwan that promotes public participation In demo..democratic societies with the aid of open-source softwares. So… by the way in this session photos are welcomed, and sharing to social media is more than welcomed. Yeah, you see Ronny’s now photoshop…photographing me. So in previous sessions like yesterday, and the day before yesterday, we focuses on news reports and newsrooms. However, in our project, Cofacts we approach internet hoax and rumors from consumers’ perspective. So what is consumers’ perspective? My family members starts a group or a chatroom on LINE to share information. It is really common in Taiwan, and the same would happen among coworkers in a company, or friends in group tour trips and so on. Basically everyone would have
dozens of group chatrooms in their mobile phone. Everyday, messages like this may pop out in one of those chatrooms. This message… this message claims that raw eggs on the windshield will block your vision if you activate the washer and wiper, and gangsters would use this technique to
force you to pull over and harm you. This is proved to be a false information many years ago. Trust me, when someone want to make a scramble egg
on your windshield, you can totally trust the wiper to get rid of it. However, we are still receiving this hoax today. If we want to know if this is correct,
or even more than that, we want to stop rumors from spreading what do we do? Well, we may spend minutes or hours googling for the fact-checking reports. Then, we rewrite the information we found into chat messages that is consumable
in the context of instant messaging. Lastly, we send it back to the
chatroom and hope that we are not hurting their feelings. and we need to repeat that every time. Since our effort is only
perceivable in one chatroom, we will need to repeat that in the other chatrooms. How tedious is that! That’s why I started the Cofacts project, a chatbot with a collaborative fact-checking database. Currently we are experimenting on LINE messenger, but I think the experience could apply
to other social platforms as well. Whenever a friend sends you a message
and you wanna check it you can just forward that information to Cofacts The chatbot looks the message up in our reply database, then returns the found replies. It also lowers the barrier for senior citizens
to find answers on their own. In 2016, more than 90% of our senior citizens in Taiwan use LINE every week For them, LINE could be their first step
into the world of Internet. We could be familiar with this process of googling and finding information on our own. But I found that for my…
at least for my family members this process is actually quite difficult. It turns out forwarding messages in LINE is much much easier for them,
and they do it every day. So with chatbots we can empower
everyone, old and young, to find information more easily,
without even leaving chat applications. So that’s how the chatbot would serve the users. But how do we get the database? The database stores two kinds of data. The reported internet message from LINE and how editors could reply to the article. When the user sends us a message
that is not found in our database, the chatbot asks if they want to
submit the message to the database. Then the submitted message is
stored and displayed in a website, cofacts.g0v.tw, where… It is where the editors would
come and submit their replies. An editor can take a look at the message, and choose from “contains information” or… “contains misinformation”, “contains correct information”
or “contains misinformation”, then provide a short reply text for the chatbot
and the references that the reply text is based on. Later, if someone asks the chatbot
the same message again it answers the user with both the reference
and the reply text provided by the editors. Furthermore, if two editors have different opinions to one message they can also submit their replies respectively, so it is possible for one message to have two or more replies. The chatbot will display multiple replies at once to notify the user that there are different
versions or different opinions to reply. or to choose from. So we embrace and appreciate diversity. Lastly, we allow everyone…
anyone to become editors. Everyone, as long as they have a
Twitter, or Facebook or Github account, can register at the out website and become
an editor to start writing replies right away. It is like Wikipedia that links fact-checking
reports and internet messages together. The linkage… the linkage between them are the crowd-sourced replies from the editors. So this is how Cofacts work,
there are two groups of people involved, accessing a collaborated database
of messages and their replies. Now let’s take a look at some numbers. At the end of June, there are more than 18,000 people adding Cofacts as their friend in LINE. More than 7.5k LINE messages were
forwarded to our database. About 150 new messages were
reported to the database every week from about 100 weekly active users. As for the editing side,
by the end of June, there are 160 editors registered in our website. Every week, there are 7 unique editors
that would write up to 90 replies as a whole. As you may… as you may notice, that it’s actually slower than
how the reported messages grow. So to make our editors feel supported, we’ve held offline editors’ meetup inviting the online editors to work together in a same place. This photo is taken from our first editors’ meetup. Within 100 minutes,
we have more than 10 editors writing replies to success…
successfully process 112 messages. The meetup was a real fun. It’s nothing like sitting in front of a laptop
to do fact-checking alone. The collaborators get a chance to sit down
and work face-to-face, and they’ve talked a lot
and laughed a lot. We are having our second meetup, uh second editors’ meetup in mid August. Let’s hope we can motivate
more people to become editors and start contributing to the database. Last but not least, the source code of the chatbot, the website and the API server are available at Github, so anyone can start hosting their very own Cofacts today. Our working folder, including all our design notes and meeting notes are
open in public as well, so that you can get to know what we…
how we made our design decisions. We really hope that others can duplicate,
duplicate Cofacts and bring it to a different level. For example, a group of doctors can
build their own LINE bot that specifically targets health-
related messages,and so on. As we go all-open, all professions can leverage our current effort into their own fields and
finally benefits the general public. To sum up, Cofacts is a chatbot that searches for
replies right inside a chat app. For those who don’t know how to google, Cofacts lowers the entry barrier and empowers
them to find information on their own. By collecting real-world internet hoax and inviting more people to become an editor, we hope that Cofact can facilitate the
spreading of fact-checking reports, and help users to gain information
to think independently, and finally improves the
media literacy of all citizens. This is the end of my presentation.
And the link is… of the slide is up there. bit.ly/cofacts-slides So thanks for your listening.

Leave a Reply

Your email address will not be published. Required fields are marked *