Opinion corpus of Slovene web commentaries KKS

Version 1.001, 2017-04-24

The corpus of web commentaries with sentiment categorizations was developed as a part of BSc Thesis (Kadunc, 2016) and served for evaluation of Slovene Sentiment Lexicon KSS. It contains web commentaries about different topics (business, politics, sport, and other) from 4 Slovene web portals (RtvSlo, 24ur, Finance, Reporter). The corpus is in XML format and available in two forms:

original corpus, containing 4,777 commentaries, 898 positive, 3,291 negative and 588 neutral commentaries.
balanced corpus, containing 1,740 commentaries, 580 of each type of sentiment (positive, negative and neutral).

Both files are available from Clarin.si repository.

References:

Klemen Kadunc (2016). Določanje sentimenta slovenskim spletnim komentarjem s pomočjo strojnega učenja. Diplomsko delo. Univerza v Ljubljani, Fakulteta za računalništvo in informatiko (in Slovene). metainfo
Klemen Kadunc, Marko Robnik-Šikonja (2016). Analiza mnenj s pomočjo strojnega učenja in slovenskega leksikona sentimenta. Conference on Language Technologies & Digital Humanities, Ljubljana (in Slovene), slides, proceedings

Back to repository of research resources