Exploiting Syntactic Structures for Humor Recognition

Proceedings of the 27th International Conference on Computational Linguistics, pages 1875–1883 Santa Fe, New Mexico, USA, August 20-26, 2018. 1875 Exploiting Syntactic Structures for Humor Recognition Lizhen Liu Information Engineering Capital Normal University Beijing, China liz liu7480@cnu.edu.cn Donghai Zhang Information Engineering Capital Normal University Beijing, China dhzhang@cnu.edu.cn Wei Song∗ Information Engineering Capital Normal University Beijing, China wsong@cnu.edu.cn Abstract Humor recognition is an interesting and challenging task in natural language processing. This paper proposes to exploit syntactic structure features to enhance humor recognition. Our method achieves significant improvements compared with humor theory driven baselines. We found that some syntactic structure features consistently correlate with humor, which indicate interesting linguistic phenomena. Both the experimental results and the analysis demonstrate that humor can be viewed as a kind of style and content independent syntactic structures can help identify humor and have good interpretability. 1 Introduction Humor, as a human-specific attribute, plays an important role in human communication. In addition to the tremendous help for human social success, humor also has positive and far-reaching effects on human psychology and physical health (Martineau, 1972; Anderson and Arnoult, 1989; Lefcourt and Martin, 1986). In recent years, the development of artificial intelligence has reinforced the human requirement for the intelligence of machines. As one of the qualities that embodies human wisdom, humor has attracted wide interests and attention (Mihalcea and Strapparava, 2005; Friedland and Allan, 2008; Zhang and Liu, 2014; Yang et al., 2015). The establishment of humor understanding mechanism promotes the development of language intelligence. With the encouragement of exploring the essence of humor, great progress has been made in the research of humor theories in recent decades. Well recognized theories including superiority theory (Gruner, 1997), relief theory (Rutter, 1997) and incongruity theory (Suls, 1972) have been successively put forward, which explain the origin and essence of humorous feelings. Inspired by these theories, many computational methods are designed to model humor and make some achievements (Mihalcea and Strapparava, 2005; Friedland and Allan, 2008; Zhang and Liu, 2014; Yang et al., 2015). Although many linguistic cue features have been studied, one aspect is often ignored that humor is a kind of style as well. An interesting question is whether we can explain humor from the perspective of styles. In this paper, we attempt to answer this question by exploring the syntactic structures, which are content independent and style related, for recognizing humor in text. The main contributions can be summarized as below: (1) We extract production rules, dependency relations and statistics of structural elements as features to model humor. The syntactic structures significantly improve the performance of humor recognition. (2) We demonstrate that some syntactic structures, which are differently distributed in humorous and non-humorous texts, indicate some interesting linguistic phenomena about humor. 2 Humor Recognition The main goal of humor recognition is to judge whether the given text expresses humor (Mihalcea and Strapparava, 2005). It can be viewed as a typical classification task. ∗corresponding author This work is licenced under a Creative Commons Attribution 4.0 International Licence. …

Content truncated to 500 words