Twitter has been analyzed in the 2008 US presidential
elections. Topsy, which a real time social network analytics
provider, in a joint work with Twitter, built the “Twitter
Political Index” which presented a daily analysis of user’s
sentiments for various topics related to the elections. Topsy
analyzed tweets about both the presidential candidates
(Obama and Romney) and calculated their popularity score
based on the sentiment present in the users’ tweets [10] [11].
A big data analysis project for a Stanford course adopted a
related approach for the 2012 US presidential elections [7].
They found that support vector machines are better predictive
models than naïve Bayes, and used them to correctly predict
Obama as the election winner. Notwithstanding the result, the
authors also highlight that Twitter cannot be considered
completely self-sufficient in predicting elections. Also, in
[16], the authors employ sentiment analysis on tweets to
predict the voting percentage, i.e., the percentage of votes,
which each presidential candidate will receive in the
Singapore Presidential Election 2011. They create sentiment
divisions and employ re-weighting techniques on these
divisions to predict voting percentage. The results didn’t
predict the correct winner, and the authors indicate a change in
voting trend and fake Twitter sentiment for these
discrepancies.
In [12], the authors analyzed about 100,000 tweets
containing the name or mention to a political party or
politician. This study was done to understand if there was any
linkage or similarity between the online political sentiment
and offline sentiment about the German Federal Elections.
They used Twitter messages to predict the popularity of
political parties, candidates and coalitions in the real world.
This work also uses sentiment analysis techniques to measure
the popularity of the candidates in question. Twitter opinion
mining has also been applied to the Belgian elections of 2010.
The experiment was conducted on 7600 tweets about Belgian
politicians, and positive and negative sentiments were
calculated using sentiment analysis techniques. Different
analytics were extracted from the study and are elaborated in
[13]. Notwithstanding these applications, it has been
significantly highlighted in [17] that the predictive accuracy of
Twitter in predicting the election winner is quite exaggerated.
The authors prove that Twitter analysis often provides a
myopic view of the election results, but it definitely cannot
replace the traditional voting method.
Twitter has been analyzed in the 2008 US presidential
elections. Topsy, which a real time social network analytics
provider, in a joint work with Twitter, built the “Twitter
Political Index” which presented a daily analysis of user’s
sentiments for various topics related to the elections. Topsy
analyzed tweets about both the presidential candidates
(Obama and Romney) and calculated their popularity score
based on the sentiment present in the users’ tweets [10] [11].
A big data analysis project for a Stanford course adopted a
related approach for the 2012 US presidential elections [7].
They found that support vector machines are better predictive
models than naïve Bayes, and used them to correctly predict
Obama as the election winner. Notwithstanding the result, the
authors also highlight that Twitter cannot be considered
completely self-sufficient in predicting elections. Also, in
[16], the authors employ sentiment analysis on tweets to
predict the voting percentage, i.e., the percentage of votes,
which each presidential candidate will receive in the
Singapore Presidential Election 2011. They create sentiment
divisions and employ re-weighting techniques on these
divisions to predict voting percentage. The results didn’t
predict the correct winner, and the authors indicate a change in
voting trend and fake Twitter sentiment for these
discrepancies.
In [12], the authors analyzed about 100,000 tweets
containing the name or mention to a political party or
politician. This study was done to understand if there was any
linkage or similarity between the online political sentiment
and offline sentiment about the German Federal Elections.
They used Twitter messages to predict the popularity of
political parties, candidates and coalitions in the real world.
This work also uses sentiment analysis techniques to measure
the popularity of the candidates in question. Twitter opinion
mining has also been applied to the Belgian elections of 2010.
The experiment was conducted on 7600 tweets about Belgian
politicians, and positive and negative sentiments were
calculated using sentiment analysis techniques. Different
analytics were extracted from the study and are elaborated in
[13]. Notwithstanding these applications, it has been
significantly highlighted in [17] that the predictive accuracy of
Twitter in predicting the election winner is quite exaggerated.
The authors prove that Twitter analysis often provides a
myopic view of the election results, but it definitely cannot
replace the traditional voting method.
การแปล กรุณารอสักครู่..
