Banner
    If An Artificial Intelligence Read 2.5 Million News Articles, What Would It Learn?
    By News Staff | November 26th 2012 01:00 PM | 3 comments | Print | E-mail | Track Comments

    Artificial Intelligence (AI) algorithms recently parsed 2.5 million articles from 498 different English-language (online) news outlets over a period of ten months and created data about what was contained.  Could AI qualitatively give people more interesting news?

    The results showed what you likely knew - online tabloid newspapers are more readable than broadsheets and use more sentimental language. Among 15 U.S. and U.K. newspapers, The Sun was the 'easiest' to read - comparable to the BBC's children's news program, Newsround - while they found the The Guardian was the most difficult to read.

    'Sport' and 'Arts' topics were the most readable topics while 'Politics' and 'Environment' were the least and sports showed a predictable male bias.  Sports news mentioned men eight times more often than women while 'Fashion' and 'Arts' were the least biased - 'Fashion' articles mentioned equal proportions of men and women. 

    The Sun was also the most likely to use sentimental adjectives while the Wall Street Journal used the fewest emotional adjectives. 

    The most appealing topics to online readers were 'Disasters', 'Crime', and the 'Environment' while the least appealing topics were 'Fashion', 'Markets' and 'Prices'. The researchers also found that the popular articles tend to be more readable and more linguistically subjective.


    Obviously not all content is put online so that skewed the data set.  Online papers put the most popular stuff on the WWW.

    Nello Cristianini, Professor of Artificial Intelligence at the University of Bristol, said about the research, "The automation of many tasks in news content analysis will not replace the human judgement needed for fine-grained, qualitative forms of analysis, but it allows researchers to focus their attention on a scale far beyond the sample sizes of traditional forms of content analysis."

    Professor Justin Lewis, Head of the School of Journalism, Media and Cultural Studies at Cardiff, found the results a way to invoke gender bias - "Even some of the more predictable findings give us pause for thought. The extent to which news is male dominated shows how far we are from gender equity across most areas of public life. The fact that articles about politics are the least readable might also explain widespread public disengagement."

    The results were published online in Digital Journalism.


    Comments

    An interesting, somewhat objective evaluation of on-line news stories. Of course on-line pornography beats news every time when it comes to hits. Sentimentality and readability ought to be a real draw to the hoi polloi. When hungry chimps were shown pictures of food and chimp troop leaders, they preferred looking at chimp celebrities more - and we didn't even need an AI analysis to suggest human celebrity on-line articles would be more popular than news (more readable and sentimental too;)

    Watson had read 200 MILLION pages last count ! It outperforms humans on quizzes and is being used to give better medical diagnosis as it now devours all medical journals on line. Although presently housed in ibm supercomputers, you should soon be able to flip your cellphone to read any area you wish or just cloud-access it and give an expert opinion.
    After that we'll link them to machines that carrry out tasks too complex for humans.

    Gerhard Adam
    ... you should soon be able to flip your cellphone to read any area you wish or just cloud-access it and give an expert opinion.
    Actually that would be nothing of the sort.  You'd simply be parroting back something that you assumed was an expert opinion.  More to the point, if it were a cellphone app, then why would I ask someone else anyway?  Watson simply becomes the new Oracle at Delphi, and the faithful will follow his advice.
    Mundus vult decipi