Using Lexical Resources for Irony and Sarcasm Classification

Authors: Miljana Mladenović, Cvetana Krstev, Jelena Mitrović, Ranka Stanković
Year: 2017
Venue: “8th Balkan Conference of Informatic” 2107, September 21-23, 2017 Skopje, Makedonia
Product of the Action: Yes

Keystone Members Authors:

The paper presents a language dependent model for classification of statements into ironic and non-ironic. The model uses various language resources: morphological dictionaries, sentiment lexicon, lexicon of markers and a WordNet based ontology. This model uses various features: antonymous pairs obtained using the reasoning rules over the Serbian WordNet ontology (R), antonymous pairs in which one member has positive sentiment polarity (PPR), polarity of positive sentiment words (PSP), ordered sequence of sentiment tags (OSA), Part-of-Speech tags of words (POS) and irony markers (M). The proposed model was evaluated on two collection of tweets that had been manually annotated according to irony. These collections of tweets are in the Serbian language (or one of closely related languages Bosnian/Croatian/Montenegrin) as well as the used language resources. The best achieved accuracy of the developed classifier was achieved for irony with the set of 5 features – (PPR, PSP, POS, OSA, M) – acc = 86.1%, while for sarcasm the best results were achieved with the set (R, PSP, POS, OSA, M) – acc = 72.8.