More on automated sentiment analysis of Danish politicians on Wikipedia

Previously today I put up sentiment analysis of Danish politician Ellen Trane Nørby on the text of the Danish Wikipedia.

Unfortunately, I could not resist the temptation of spending a bit of time on also running the analysis for some other Danish Politicians. I did it for Prime Minister Helle Thorning-Schmidt, former Prime Minister Lars Løkke Rasmussen, former Foreign Minister Lene Espersen and former Minister Ole Sohn.

For Rasmussens article we see a neutral factual bibliographic article until 2008, though with a slight increase in the end of 2007 when he became Minister of Finance. Then in May 2008 we see a drop in sentiment with the introduction of a paragraph mentioning an “issue” related to his use of county funds for private purposes. Since then the article has been extended and now generally positive. There are some spikes in the plots. These spikes are typically vandalism that persist for a few minutes until reverted.

For Helle Thorning-Schmidt we see a gradual drop up towards the election she wins and after that her article gains considerable positivity. I haven’t check up much on this in the history, but I believe it is related to the tax issue her and her husbond, movie star Stephen Kinnock, had and a number of other issues. As I remember there was concern or discussion on the Danish Wikipedia on whether these “issues” should fill up so large a portion of the article and on the 3 December 2011 a user moved the content to another page.

I believe I am one of the major perpetrators behind both the Lene Espersen and Ole Sohn articles. Both of the articles have large sections which describe negative issues (I really must work on my positivity, these politicians are not that bad). However, the sentiment analysis shows the Ole Sohn article as more positive. Maybe this is due to the “controversy” section described that he paid “tribute” to East Germany and that his party received “support” from Moscow, i.e., my simple sentiment analysis does not understand the controversial aspect of support from communist Moscow and just think that “support” is positive.

Writing politicians article on Wikipedia I find it somewhat difficult to identify good positive articles that can be used as sources. The sources used for the encyclopedic articles usually comes from news articles and these have often a negative bias with a focus on “issues” (political problems). Writing the Lene Espersen article I found that even the book “Bare kald mig Lene”, which I have used a source, has a negativity bias. If I remember correctly Espersen did not want to participate in the development of the book, presumably because she already had the notion that the writers would focus on the problematic “issues” in her career.

(2013-01-10: spell correction)