Sentiment annotation of The Leipzig Corpora Collection.
Manual sentiment annotation by Finn Årup Nielsen of Danish sentences.
The data is directly derived from The Leipzig Corpora Collection which is downloadable from http://corpora2.informatik.uni-leipzig.de/download.html. The format in the data files is changed slightly from tab-separated to comma-separated and an column header is added. An extra column is introduced called 'valence'. This contains the sentiment scoring. The scoring follows the pattern from AFINN with integer scoring from -5 to +5, though for most sentence, if any, the full range of the score is not used.
An example in Python with the Pandas for reading the data with the read_csv function is shown in the Jupyter Notebook.
The citation to the unannotated Leipzig Corpora Collection is
Quasthoff, U.; M. Richter; C. Biemann: Corpus Portal for Search in Monolingual Corpora, Proceedings of the fifth international conference on Language Resources and Evaluation, LREC 2006, Genoa, pp. 1799-1802.
There is presently no citation to a scientific paper for the annotation.
https://github.com/fnielsen/europarl-da-sentiment