Recent years, have seen many advancements in computerized
document ranking and classification systems.
In this paper, we compare one recently developed algorithm known as
Wordscores to several industry standard
document ranking, classification, and summarization algorithms.
We point out several of the weaknesses in Wordscores, illustrating
that it performs well at partisan identification in specific situations,
as well as other tasks, but show that it
fails in other instances. Particularly, Wordscores identifies that
in his interview with Dan Rather in February 2003,
Saddam Hussein gave less support to terrorism than any Supreme Court Justice, in
recent rulings.
Quoting Ian Brett Yohai <yohai(a)fas.harvard.edu>du>:
The last sentence seems to be a bit unclear: did you
mean stances on
terrorism taken by Justices - or something about terrorism cases in their
opinions? Likewise, did you really mean Saddam Hussein's stance on
terrorism?
Best,
Ian
On Mon, 8 May 2006 ghumphr(a)fas.harvard.edu wrote:
Recent years, have seen many advancements in
computerized
document ranking and classification systems.
In this paper, we compare one recently developed algorithm known as
Wordscores to several industry standard
document ranking, classification, and summarization algorithms.
We point out several of the weaknesses in Wordscores, illustrating
that it performs well at partisan identification in specific instances,
but show that it fails critically at other tasks, such as identifying
stances on
terrorism taken by Supreme Court Justices and
Saddam Hussein.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l