Humor Recognition and Humor Anchor Extraction

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 2367–2376, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Humor Recognition and Humor Anchor Extraction Diyi Yang, Alon Lavie, Chris Dyer, Eduard Hovy Language Technologies Institute, School of Computer Science Carnegie Mellon University. Pittsburgh, PA, 15213, USA {diyiy, alavie, cdyer}@cs.cmu.edu, hovy@cmu.edu Abstract Humor is an essential component in personal communication. How to create computational models to discover the structures behind humor, recognize humor and even extract humor anchors remains a challenge. In this work, we first identify several semantic structures behind humor and design sets of features for each structure, and next employ a com- putational approach to recognize humor. Furthermore, we develop a simple and effective method to extract anchors that enable humor in a sentence. Experiments conducted on two datasets demonstrate that our humor recognizer is effective in automatically distinguishing between humorous and non-humorous texts and our extracted humor anchors correlate quite well with human annotations. 1 Introduction Humor is one of the most interesting and puzzling research areas in the field of natural language understanding. Recently, computers have changed their roles from automatons that can only perform assigned tasks to intelligent agents that dynami- cally interact with people and learn to understand their users. When a computer converses with a human being, if it can figure out the humor in human’s language, it can better understand the true meaning of human language, and thereby make better decisions that improve the user experience. Developing techniques that enable computers to understand humor in human conversations and adapt behavior accordingly deserves particular attention. The task of Humor Recognition refers to determining whether a sentence in a given context expresses a certain degree of humor. Humor recognition is a challenging natural language problem (Attardo, 1994). First, a universal definition of humor is hard to achieve, because different people hold different understandings of even the same sentence. Second, humor is always situated in a broader context that sometimes requires a lot of external knowledge to fully understand it. For example, consider the sentence, “The one who invented the door knocker got a No Bell prize” and “Veni, Vidi, Visa: I came, I saw, I did a little shopping”. One needs a larger cultural context to figure out the subtle humorous meaning expressed in these two sentences. Last but not least, there are different types of humor (Raz, 2012), such as wordplay, irony and sarcasm, but there exist few formal taxonomies of humor characteristics. Thus it is almost impossible to design a general algorithm that can classify all the different types of humor, since even human cannot perfectly classify all of them. Although it is impossible to understand univer- sal humor characteristics, one can still capture the possible latent structures behind humor (Bucaria, 2004; Binsted and Ritchie, 1997). In this work, we uncover several latent semantic structures behind humor, in terms of meaning incongruity, ambigu- ity, phonetic style and personal affect. In addition to humor recognition, identifying anchors, …

Content truncated to 500 words