This is a very basic classifier that gives a polarity (positive, negative, neutral) given a text and a polarity lexicon. It uses Tree-Tagger by Helmut Schmid for Bulgarian language. Please, pay attention to the private license of Tree-Tagger which only allows you to use the system for research purposes.
- Perl interpreter (in Linux)
echo брат ми е много щастлив | ./classif.sh
The system consists of two modules:
- the PoS tagger Tree-Tagger for Bulgarian
- a basic lexicon-based classifier in Perl that just counts positive and negative lemmas by looking up a polarity lexicon
The two modules are connected by means of the following pipeline:
./cmd/tree-tagger-bulgarian |./classif.perl ./resources/dico-bl.csv
The system makes uses of a polarity lexicon, dico-bl.csv, with more than 2,000 lemmas, built by Velyana Patsova.