Using Tweets for Assigning Sentiments to Regions

Erik Tjong Kim Sang

Using Tweets for Assigning Sentiments to Regions

Erik Tjong Kim Sang

Meertens Institute

Research output: Contribution to conference › Paper › Scientific › peer-review

324 Downloads (Pure)

Abstract

We derive a sentiment lexicon for Dutch tweets and apply the lexicon for classifying Dutch tweets as positive, negative or neutral. The
classifier enables us to test what regions in the Netherlands and Flanders express more positive sentiment on Twitter than others. The
results reveal sentiment differences between Flemish and Dutch provinces, and expose municipalities which are a lot more negative than
their neighborhood. The results of this study can be used for finding areas with local issues that might be expressed in tweets.

Original language	English
Pages	64-67
Number of pages	4
Publication status	Published - 26 May 2014

Access to Document

lrec2014bFinal published version, 666 KB

http://ifarm.nl/erikt/papers/lrec2014b.pdf

Cite this

@conference{f4ec7f0569ee419e96dbf318ddde3c8f,

title = "Using Tweets for Assigning Sentiments to Regions",

abstract = "We derive a sentiment lexicon for Dutch tweets and apply the lexicon for classifying Dutch tweets as positive, negative or neutral. Theclassifier enables us to test what regions in the Netherlands and Flanders express more positive sentiment on Twitter than others. Theresults reveal sentiment differences between Flemish and Dutch provinces, and expose municipalities which are a lot more negative thantheir neighborhood. The results of this study can be used for finding areas with local issues that might be expressed in tweets.",

author = "{Tjong Kim Sang}, Erik",

note = "in: Proceedings of the 5th International Workshop on Emotion, Social Signals, Sentiment and Linked Open Data at LREC2014, Reykjavik, Iceland, 2014",

year = "2014",

month = may,

day = "26",

language = "English",

pages = "64--67",

}

TY - CONF

T1 - Using Tweets for Assigning Sentiments to Regions

AU - Tjong Kim Sang, Erik

N1 - in: Proceedings of the 5th International Workshop on Emotion, Social Signals, Sentiment and Linked Open Data at LREC2014, Reykjavik, Iceland, 2014

PY - 2014/5/26

Y1 - 2014/5/26

N2 - We derive a sentiment lexicon for Dutch tweets and apply the lexicon for classifying Dutch tweets as positive, negative or neutral. Theclassifier enables us to test what regions in the Netherlands and Flanders express more positive sentiment on Twitter than others. Theresults reveal sentiment differences between Flemish and Dutch provinces, and expose municipalities which are a lot more negative thantheir neighborhood. The results of this study can be used for finding areas with local issues that might be expressed in tweets.

AB - We derive a sentiment lexicon for Dutch tweets and apply the lexicon for classifying Dutch tweets as positive, negative or neutral. Theclassifier enables us to test what regions in the Netherlands and Flanders express more positive sentiment on Twitter than others. Theresults reveal sentiment differences between Flemish and Dutch provinces, and expose municipalities which are a lot more negative thantheir neighborhood. The results of this study can be used for finding areas with local issues that might be expressed in tweets.

M3 - Paper

SP - 64

EP - 67

ER -