Classification of Emotion Words in Russian and Romanian Languages
Marina V. Sokolova, Victoria Bobicev · Recent Advances in Natural Language Processing · 2009
This paper presents a machine learning study of aective words in Russian and Romanian languages. We tag the word aective meaning by one of the WordNet Aect six labels anger, disgust, fear, joy, sadness, surprise and group into “positive” (joy, surprise) and “negative” (anger, disgust, fear, sadness) classes. We use the word spelling, a word form, to represent words in machine learning experiments to solve the multiclass classification and binary classification problems. The results show that the word form can be a reliable source of learning the aect.