Irony detection based on semantic similarity

The goal of this work is to propose an algorithm to automatically determine whether the word used ironically or literally using semantic similarity. Previous approaches to this problem are reviewed. The term “irony” is defined. Two sets are collected, the first of them contains statements with words used ironically, the second contains statements with the same words used literally. Different methods of semantic similarity measuring are studied. A semantic similarity based algorithm to automatically determine whether the word used ironically or literally is proposed.