[Asis-l] new emotional database of a half million hours of television news

[kalev leetaru]

Apologies for cross-posting.  I thought many of you would find of interest
the latest GDELT Project (http://www.gdeltproject.org/) release.  All
540,000 hours of English-language American television broadcasts held by
the Internet Archive's Television News Archive were processed to extract
the complete list of people, organizations, disambiguated locations, and
more than 2,238 emotions and themes from each broadcast.  The complete
dataset (only the extracted metadata, not the broadcasts themselves) is
available for download, along with an interactive viewer timeline that
allows you to visualize a particular emotion or theme over the last four
years of American television news:

http://blog.gdeltproject.org/visualizing-the-emotions-of-american-television-news/
http://analysis.gdeltproject.org/cgi-bin/iatvemotions/iatvemotions

As an example, here is the plot of the intensity of "Anxiety" over time -
notice the huge surge around the government shutdown last October and in
June 2012 with the massive derecho storm:

http://analysis.gdeltproject.org/cgi-bin/iatvemotions/iatvemotions?var=c8.3&network=

While the extracted metadata is currently only at the level of a entire
broadcast, it offers an enormous range of possibilities in terms of a first
exploration of the emotions and themes of American broadcast news.
Everything from mapping the geography of television (
http://blog.archive.org/2013/12/13/mapping-400000-hours-of-u-s-tv-news/) to
building emotionally-weighted cooccurance networks is possible with this
dataset.  The ability to infuse cooccurance graphs with emotional,
geographic, and thematic context layers opens some particularly interesting
doors.

A second dataset of possible interest applies the same coding to English
translations of 101 speeches given by Syrian President Bashar al-Assad from
2000 to present and a comparison dataset of 21,295 presidential documents
from The American Presidency Project for Presidents Clinton, Bush, and
Obama:

http://blog.gdeltproject.org/leader-speeches-assad-vs-american-presidents/

In the next few days we will be debuting this emotional/thematic coding for
all global news media monitored by the GDELT Project, meaning that you will
shortly be able to assess these dimensions live over the world's news media:

http://blog.gdeltproject.org/introducing-the-global-content-analysis-measures-gcam/


~Kalev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.asis.org/pipermail/asis-l/attachments/20141015/136be399/attachment.html>