Sentiment Lexicon from paired Comparisions (with source code)
Here you can find additional material to the following publication:
- C. Dalitz, K.E. Bednarek: Sentiment Lexica from Paired Comparisons. International Conference on Data Maining Workshops (ICDMW), pp. 924-930 (2016)
A sentiment lexicon is a dictionary that assigns each term a polarity score representing the strength of the positive or negative affect associated with the term. We have devised a new method for estimating polarity scores from preference judgements of test persons. Here we provide both the source code for computing the polarity scores and an initial set of 200 German words with their polarities. Moreover, we provide a GUi for extending this lexicon with arbitrary words.
Available releases (see the file README for usage instructions):
- sentire-data-1.1.tgz: 200 words sentiment lexicon with all-pair comparisons and R-scripts for computing the scores (version 1.1 from 2017/06/02)
- wordrating-1.4.tgz: GUI for adding arbitrary words with R-script for computing the new scores.
Comment: The sentiment lexicon with 200 words described in the above paper had inherited from SentiWS a double word in two different spellings ("phantasievoll" versus "fantasievoll"). Both words should be replaced by a single word ("fantasievoll") with a score value equal to the average score of the differently spelled words. The GUI wordrating takes care of this since version 1.4.
Authors and license
The source code is copyrighted by Christoph Dalitz and Jens Wilberg, Institute for Pattern Recognition, Niederrhein University of Applied Sciences, Krefeld, Germany. It may be used freely under the terms of the GNU General Public License.
The sentiment lexicon is copyrighted by C. Dalitz and K.E. Bednarek. It may be freely copied and used, provided the paper listed at the top of this page is given as a reference.