Irony Sentence Detection Techniques Using Fuzzy Historical Classifier
Adhitia Erfina* and Yeffry Handoko Putra
Department of Magister Information System, Faculty of Post Graduate, Universitas Komputer Indonesia, Bandung, Indonesia
Abstract
This study presents a new approach to the extraction of the meaning of sentences that are irony, that is by way of classifying someone based on their utterances in the past. The history of ones utterances influences the assessment of a sentence having an irony tendency or not, for example when someone often speaks negatively, suddenly gives a positive opinion on a topic while other people give negative opinions on the topic. The fuzzy logic method needs to be used to assess the historical tendency of ones utterances when the values of positive and negative sentiments are almost balanced so that the value of the majority of sentiments is unclear. The results show that the greater the level of difference in sentiment between a topic and the higher the level of the historical tendency of a persons utterance, the higher the value of the potential irony of the utterance.
Keywords: Sentiment Analysis; Fuzzy Logic; Irony Sentence; Sarcasm Detection
Topic: Informatic and Information System