| 摘要 |
Disclosed are an apparatus and a method for predicting a pleasant-unpleasant index of a word by using relative emotion similarity. The apparatus for predicting a pleasant-unpleasant index of a word includes: a calculation unit which outputs emotion relation between the word and at least one comparison word, outputs emotion relation between each of multiple reference words, included in a reference word group, and each comparison word, outputs multiple first absolute emotion similarities which are absolute emotion similarities for the word and each reference word, and also outputs at least one second absolute emotion similarity for one reference word and another reference word with respect to total multiple reference words included in the reference word group; and a prediction unit which predicts the pleasant-unpleasant index of the word by using the first absolute emotion similarities, the at least one second absolute emotion similarity, and a predetermined pleasant-unpleasant index of the reference words. |
