150 Years Of British History From Over 100 Local Newspapers: What Did AI Researchers Find?
A project of Bristol University is making news with a team of its Artificial Intelligence researchers sourcing colossal information from more than 100 British regional newspapers spanning 150 years in defining and pinpointing important turning points in the country's history and culture. The results are amazing with many interesting revelations.
What makes the project remarkable is the methodology. Already, it is known to all that history is well-etched in books. However, using newspapers as the primary source to pinpoint major cultural and historical shifts has been extraordinary.
Thanks to the computer scientists from the university, an artificial intelligence software was used in analyzing the British newspapers between 1800 and 1950 to bring out significant milestones in history and culture.
"The key aim of the study was to demonstrate an approach to understanding continuity and change in history, based on the distant reading of a vast body of news, which complements what is traditionally done by historians," said Nello Cristianini, Artificial Intelligence Professor of at the University's Department of Engineering Mathematics and the lead author of the study.
The study was published in PNAS with observations on the shift in attitudes and values in the 150-year time span.
Changes In Technology
The study succeeded in precisely unveiling the people's approach to the change in technology as seen in the transition from horse to train as transport and the shift from steam to electricity as an energy source.
That "power change" to electricity happened in 1898 and that was palpable from the media as mentions about electricity were going up over those of tired horses.
Media's obsession with celebrity news had its beginning around 1901, with singers started getting more mention in the newspapers than politicians.
The study says, by the middle of 20th century, actors, singers, and dancers gained an edge over politicians in the matter of media attention.
Extension Of Book Project
In fact, the project was an extension of what was already done in 2011 when five million books spanning 200 years were analyzed to draw insights from words.
But the project was slammed as mere word crunching and ignoring the historical context.
In the ThinkBIG project, that shortcoming was addressed. The researchers deployed an AI software to analyze 35 million articles from the British newspapers of 150 years.
Areas Of Investigation
The automated analysis of millions of articles has unraveled big events and showed changes in gender bias over the decades showed trends in adoption of new technologies and political ideas.
"The research team showed that changes and continuities detected in newspaper content can reflect culture, biases in representation or actual real-world events. More detailed studies on the same data will be performed," added Cristianini.
The study used the content analysis to pinpoint key events such as epidemics, wars, coronations. By applying the AI tools the research went beyond word crunching and extrapolated the references on individuals, companies, and places.
Among the stark trends, under-representation of women, and a steady increase of their mention in the news was highlighted, mainly in the 20th century.
Change In Gender Bias
Regarding gender bias, the research found men being more prominent in mention than females during the period of study, albeit a marginal increase of women power after 1900. However, that uptick was hard to be linked to any time factor.
The gender bias in news was not much different from what it is today, according to the study. One expert hailed the efficacy of the study and its methodology in unraveling relationships between textual content and the historical time.
"We have demonstrated that computational approaches can establish meaningful relationships between a given signal in large-scale textual corpora and verifiable historical moments", said Tom Lansdall-Welfare, who handled the computational side of the study and is a Research Associate with the Computer Science department.
There is hope that data-led approaches like this would supplement the traditional way of close reading for picking trends of change and continuity.