Sentiment Analysis of Emotional Words in a Classical Text WEB Corpus

Abstract

This study explores the emotional words based on sentiment analysis. The term 'emotional words' refers to feeling-related words that explain an individual's emotions. This study employed data from a corpus site, namely the Malay Concordance Project, a compilation of 165 texts and 5.8 million words, including 140, 000 versions of text. An analysis sentiment model was employed through collostructure. This model combined a number of pertinent elements, including the basic corpus approach, basic word approach and basic dictionary approach. Three emotional words were chosen based on their high degree of representativeness in the corpus context. These were anger-related words for the emotion of anger, amok/rage-related words for the emotion of rage, and revenge-related words for the emotion of revenge. The results from the testing of the words anger and rage showed negative sentiments; meanwhile, revenge-based words exhibited positive sentiments. This was because the context of the collostructure facilitates sentiment data reading and can be carefully filtered.