Tuesday, 21 May 2013

Irony and Machines

“I'd kill for a Nobel Peace Prize.” ―Steven Wright 

Could a machine learn how to detect irony in written text, as in the above quote for example? I’ve been working on this topic for my thesis so I thought I’d share a small introduction as a blog post here. First we must ask the question; what is irony and what is it used for? As Veale (2010) states: “Irony is an effective but challenging mode of communication that allows a speaker to express sentiment-rich viewpoints with concision, sharpness and humour”.
Irony can also be used to give different messages to a single group of people, one message to those who can understand the irony and another message to those who don’t, thus dividing the group with the same message (Veale 2004). Irony can often be confused with sarcasm so irony is generally held to be less offensive as it is usually directed towards us in a softer manner (Reyes et al. 2012).

There are many more factors in language that have to be taken into account but so far irony has been operationalized in a few different ways to see if computers can detect irony with varying success. One approach used by (Leigh, 1994) is to codify the standard definition of irony and let a computer organise statements into ironic and non-ironic. This approach has been built upon and altered in other studies to improve the definition of irony but none of these methods have been very effective because computers do not have an effective structure for language understanding. Maybe the work being done in Natural Language Processing (NLP) can help this. Reyes et al. (2012) is a good example of this progress, in this paper they were able to detect irony in twitter posts based on several textual features.



Leigh, J. H. (1994). The use of figures of speech in print ad headlines. Journal of Advertising, 17-33.
Reyes, A., Rosso, P., & Veale, T. (2012). A multidimensional approach for detecting irony in twitter. Language Resources and Evaluation, 1-30.
Veale, T. (2004). The challenge of creative information retrieval. In Computational Linguistics and Intelligent Text Processing (pp. 457-467). Springer Berlin Heidelberg.
Veale, T., Hao, Y. (2010). An ironic fist in a velvet glove: Creative mis-representation in the construction of ironic similes. Minds and Machines, 20(4), 635-650

1 comment: