Textual affect detection in human computer interaction

Abstract

In this paper we focus on the affect detection of the short text pervasively used in human computer interaction. The research intends to render the interaction more emotionally expressive. In order to estimate the affect in the short text, we construct an affect lexicon firstly. Then a set of extraction rules (ERs) is built to extract the semantic representation of each word. Finally, the affect state of the short text is represented with PAD values and computed through the manually made affect generation rules (AGRs). The evaluation of the results corresponds with the human subjective appraisal. © 2013 Springer-Verlag.