F-Word
Fake News Detecting Hub
While
brainstorming on the subject of fake news, we realised that it is something
which is growing rapidly, popularized by its consequences in the US
Presidential Elections. We realized that two major reasons for the birth of
fake news are:
1. Easy Money: A huge percentage of the
Pro-Trump and Anti-Clinton posts were made from countries like Macedonia as
stated on a Buzz feed post. When the author further investigated into the
matter, he came to know that most of these people were doing it for easy money.
2. Bias: It may be personal/emotional.
It might as well be triggered by political/organisational factors. It’s a psychological
thing. Defamation was one motive behind fake news articles.
We took our
research further when we asked for the opinion of the common people. We
conducted short surveys in which people were asked questions about fake news
and its consequences and how to tackle it. Most people were much aware about
the facts and thought verifying the sources or only trusting reliable sources
is the solution. There were other people who thought that internet users rating
a piece of news as they read it can change things dramatically.
Statistical Survey
We made a
two part Google form which dealt with data part and other with statistical part
therefore graphs and Pie Charts.
1) In the first form we asked people
about the general phenomenon of fake news and whether it affects them on
personal basis or not?
What we derived from our survey was that majority of people
are highly annoyed with the advent of fake news taking its toll on news media
2) Another interesting aspect of our
survey was that, people came across fake news more alarmingly more than we
expected.
3) After reading the responses we asked
people whether they would be interested in an application that would help to
uncover fake news?
4) After getting the response we decided
to take this data and implement them into our product via stats, which will
tell us about the data points that we need to incorporate into our product.
We found
that more 64.3 % people think that personal agendas are behind for the advent
of fake news
5) We found that people mostly go to
Google for their News or for checking their sources
6) Since this was expected majority depend on
news media as their trustworthy source
We the
decided to take this amazing data and pool our ideas to create the design of an
application which will solve these data points and help us to make better
decisions without being biased towards one opinion or other
Our approach
was then focused on these above mentioned viewpoints: Bringing
reliable/verified sources under one platform and involving user ratings.
Thus the
idea of an application that acts like a Hub for determining fake news was
started. Our approach to handle fake news was is a unique one as we don’t
decide which news is fake or not, we let our users decide with the help of
feedback algorithms to come to a conclusion regarding the authenticity of the
news
How does our Product Work
We have
developed a simple and elegant process by which we allow our User to be in
complete dominance over the aspect of determining whether the news is fake or
not.
For this we
have set up a clean UI and implemented a feedback algorithm which helps to
solidify the user in deciding the authenticity of the news
How does our Algorithm Work
The algorithm for determining fake news takes place in
two ways
1.
Client Side -
The user is the most important aspect for determining fake news. The judegment lies in the hand of
user. If a user encounters a news on our
application which they feel might be fake, then they need to provide credible
sources or links to validate their opinion. This will be followed by a necessary opinions section in which the
user must provide his/her reasons for which they feel that the news might be
fake.
After an extensive polling is done a bar is generated which displays the probability of a news that might be fake
2.
App side – The aspect of
determining whether the news is fake or not is done via Source Ranking which
takes inspiration from Page Ranking. After
Source Ranking is done the process of
Pattern Recognition will start , which will calculate the percentage of
the authenticity of the news.
Source Ranking
Whenever a news is posted on our aplication, we first check the sources of the news. There are usually multiple sources which report the news. If a news has listed 3 sources then we assign a weight (value) onto that sources. If another user has posted a link by another news publisher on that news and it has same sources then the value assigned to the weights increase.
This is a neutral
approach as we don’t know which news is actually true. More the weights
assigned to a particular source , more is the probability of that source to
attain a higher rank under the domain of news (fake or real) it falls.
If a user then clicks
on Real or Fake button then the sources
given by user will be matched with the sources. If the source is of higher
probability along with the arguments given then the weightage of that user will
be taken into account. By this way, we remain neutral and let the statistics
reveal the actual truth.
Pattern Matching
After some consensus has been drawn and users have submitted their
arguments we can take help of pattern matching to differ both the pieces of
texts and make a better understanding of how fake news works and calculate the
percentage of news that was fake.
After a story has been labelled as truth we can match it with the
highest source rank of fake domain news. We can use pattern matching algorithm
or any method which tells the difference between two pieces of text. If there
are any pieces which resemble the text present in fake news we can highlight it
and exclude it from the original piece. Thus, if we calculate the remaining
part we can get the percentage of fake news present in the media.
Thus our product was made and we designed the product
with utmost enthusiasm and passion.
Thank you @Design360 for this amazing experience and
opportunity .
Comments
Post a Comment