Fake news warriors now get a 'Fake-O-Meter'
The easy to use online tool developed by
scientists at the IIIT-Hyderabad helps determine degree of truth in news
content on the Internet and social media platforms.
In
a fillip to the war on fake news, information scientists at the Information
Retrieval and Extraction Lab (iREL), of the International Institute of
Information Technology - Hyderabad (IIIT-H), have developed tools to help
publishers and readers alike to easily check the degree of truth in news content
on the Internet and social media platforms.
Dubbed the ‘Fake-O-Meter’, the tools are based on computing
models that use deep learning methods and natural language processing (NLP),
frameworks and are programmed to inherently and continuously learn to identify
fake news in real time.
The Fake-O-Meter enables volume processing of content across
genres and languages by using past data to ‘teach’ itself to cross check for
fake content and also apply the learning to future news from the same domain,
Professor Vasudeva Varma, Dean, (Research & Development) IIIT-H, told NewsPlus.
The model can be expanded to detect fake news from domains
ranging from politics to finance, public health, celebrity news and practically
anything else, added J Ramachandran, Entrepreneur-in-Residence at the iREL and
Founder CEO of data science company Gramener.
The Fake-O-Meter gains significance given that the volume of
fake news is expected to go up in 2019 particularly in countries like with
India, Indonesia and sundry other Africa which are going to the polls this
year. In fact, some experts say 2019 will be a test year for all the fake news
detection tools that were developed in 2018.
Not surprisingly, governments have doubled efforts to strengthen
legal frameworks to curb proliferation of fake news. For instance, in India,
the Ministry of Electronics and Information Technology, has invited comments
and suggestions by January 15, on the Draft of “The Information Technology
[Intermediary Guidelines (Amendment) Rules] 2018, aimed at reining in fake news
on social media platforms.
There are 11 iterations of the Fake-O-Meter on offer from the
IIIT-H to detect fake news on politics, health, technology, Chinese on Twitter
and the Spanish language apart from Hindi and English. A Hate Meter to
determine the extent of hate language used in online content is also available.
As such the Fake-O-Meter provides the degree or possibility of
'fakeness' in a news story through a colour coded display, explained
Ramachandran, who has spearheaded the effort to productise the research done by
Professor Varma and team and make it amenable to specific use cases.
“Some domains like politics, celebrity news and health news are
particularly vulnerable to fake news and as such any news in these domains can be
cross checked as long as past references are available in the database,” said
Professor Varma.
While it is anybody’s guess as to how much fake news is out
there on the Internet, some estimate it at over 60%, a more immediate problem
for countries like India in 2019 will be the use of bots to create and
proliferate hyper-partisan and patently false content on peer-to-peer platforms
like Facebook, WhatsApp and Instagram.
It is possible to adapt the Fake-O-Meter to cross check content
shared on such peer-to-peer platforms by providing the framework through an
external layer either on an App or even in a WhatsApp group where people submit
links, he added.
Disclosure
Professor Vasudeva Varma is co-founder NewsPlus (Veooz Labs)
which uses cutting-edge ML and NLP technology to ensure its readers get only
Real News. NewsPlus uses a mix of propagation detection and source validation
algorithms to ensure this.
(This article first appeared on NewsPlus in January 2019)
Post a Comment