Sentiment mining showed a sharp change in tone around Egypt ahead of
President Mubarak's ousting
Feeding a
supercomputer with news stories could help predict major world
events, according to US research.
A study, based on millions of articles, charted deteriorating
national sentiment ahead of the recent revolutions in Libya and
Egypt.
While the analysis was carried out retrospectively, scientists
say the same processes could be used to anticipate upcoming
conflict.
The system also picked up early clues about Osama Bin Laden's
location.
Kalev Leetaru, from the University of Illinois' Institute for
Computing in the Humanities, Arts and Social Science,
presented his findings in the journal First Monday.
Mood and location
The study's information was taken from a range of sources
including the US government-run
Open Source Centre and
BBC Monitoring, both of which monitor local media output around
the world.
News outlets which published online versions were also analysed,
as was the New York Times'
archive, going back to 1945.
In total, Mr Leetaru gathered more than 100 million articles.
Reports were analysed for two main types of information: mood -
whether the article represented good news or bad news, and location
- where events were happening and the location of other participants
in the story.
The Nautilus SGI supercomputer
crunched the 100 million articles
Mood detection, or "automated sentiment mining" searched for
words such as "terrible", "horrific" or "nice".
Location, or "geocoding" took mentions of specific places, such
as "Cairo" and converted them in to coordinates that could be
plotted on a map.
Analysis of story elements was used to create an interconnected
web of 100 trillion relationships.
Predicting trouble
Data was fed into an
SGI Altix supercomputer, known as Nautilus, based at the
University of Tennessee.
The machine's 1024 Intel Nehalem cores have a total processing
power of 8.2 teraflops (trillion floating point operations per
second).
Based on specific queries, Nautilus generated graphs for
different countries which experienced the "Arab Spring".
In each case, the aggregated results of thousands of news stories
showed a notable dip in sentiment ahead of time - both inside the
country, and as reported from outside.
Media "sentiment" around Egypt fell
dramatically in early 2011, just before the resignation of
President Mubarak.
For Egypt, the tone of media coverage in the month before
President Hosni Mubarak's resignation had fallen to a low only seen
twice before in the preceding 30 years.
Previous dips coincided with the 1991 US aerial bombardment of
Iraqi troops in Kuwait and the 2003 US invasion of Iraq.
Mr Leetaru said that his system appeared to generate better
intelligence than the US government was working with at the time.
Continue reading the main story
“Start Quote
If you look at this tonal curve it
would tell you the world is darkening so fast and so
strongly against him that it doesn't seem possible he could
survive.”
Kalev Leetaru
University of Illinois
"The mere fact that the US President stood
in support of Mubarak suggests very strongly that that even the
highest level analysis suggested that Mubarak was going to stay
there," he told BBC News.
"That is likely because you have these area experts who have been
studying Egypt for 30 years, and in 30 years nothing has happened to
Mubarak.
The Egypt graph, said Mr Leetaru, suggested that something
unprecedented was happening this time.
"If you look at this tonal curve it would tell you the world is
darkening so fast and so strongly against him that it doesn't seem
possible he could survive."
Similar drops were seen ahead of the revolution in Libya and the
Balkans conflicts of the 1990s.
Saudi Arabia, which has thus far resisted a popular uprising, had
experienced fluctuations, but not to the same extent as some other
states where leaders were eventually overthrown.
Mapping Bin Laden
In his report, Mr Leetaru suggests that analysis of global media
reports about Osama Bin Laden would have yielded important clues
about his location.
Media reports mentioning Osama Bin
Laden may have helped narrow down his location
While many believed the al-Qaeda leader to be hiding in
Afghanistan, geographic information extracted from media reports
consistently identified him with Northern Pakistan.
Only one report mentioned the town of Abbottabad prior to Bin
Laden's discovery by US forces in April 2011.
However, the geo-analysis narrowed him down to within 200km, said
Mr Leetaru.
Real time analysis
The computer event analysis model appears to give forewarning of
major events, based on deteriorating sentiment.
However, in the case of this study, its analysis is applied to
things that have already happened.
According to Kalev Leetaru, such a system could easily be adapted
to work in real time, giving an element of foresight.
"That's the next stage," said Mr Leetaru, who is already working
on developing the technology.
"It looks like a stock ticker in many regards and you know what
direction it has been heading the last few minutes and you want to
know where it is heading in the next few.
"It is very similar to what economic forecasting algorithms do."
Mr Leetaru said he also hoped to improve the resolution of
analysis, especially in relation to geographic location.
"The next iteration is going to city level and beyond and looking
at individual groups and how they interact.
"I liken it to weather forecasting. It's never perfect, but we do
better than random guessing."
BBC © 2011
The BBC is not
responsible for the content of external sites. Read more.