A structural method based on fuzzy approach at producing emotional data

Human-machine interaction includes various challenges. Emotion recognition is one of the most important issues, which researchers become interested in recent years. Text content is the core of emotional data. The text of emotional data plays significant role in the emotional states. Textual content is important due to the fuzzy behavior, variegation in the meaning of words, and the effect of meaning on emotional states. Accordingly, meaning independent text which its polarity is neutral should be applied on the emotional data. This type of text induces advantages such as: multiple meanings are not generated, multiple emotional states are not perceived from one sentence and the phonotactics has no effect on the emotional states. This paper investigates the related challenges and proposes a structural method for designing the text of emotional data. The results illustrate emotional data which their text has been designed via proposed method eventuate higher performance in the emotion recognition. The proposed method was applied in the Persian language and the Persian Meaning Independent Text Dataset was born. The proposed method is not unique to the Persian language and it has been potential to apply to all languages.