Research on Humor Recognition

摘要

Humor, as an important part of human communication, plays an important role in the field of Natural Language Processing. It is worth pondering how to establish an effective model of humor recognition. In this paper, we explore the differences between humorous examples and nonhumorous examples from three perspectives which include Emotion Dimensions, Conflict and Emotion Outburst. Based on these perspectives, we design a series of effective features which weaken the influence of domain difference and clustering on humor recognition. Experimental results show that our features are effective. And our method provides a new idea for future researches.