Irony is an effective but challenging mode of communication that allows a speaker to express sentiment-rich viewpoints with concision, sharpness and humour. Irony is especially common in online documents that express subjective and deeply-felt opinions, and thus represents a significant obstacle to the accurate analysis of sentiment in web texts. In this paper we look at one commonly used framing device for linguistic irony – the simile – to show how irony is often marked in ways that make it computationally feasible to detect. We conduct a very large corpus analysis of web-harvested similes to identify the most interesting characteristics of ironic comparisons, and provide an empirical evaluation of a new algorithm for separating ironic from non-ironic similes.

An effective speaker can use metaphor to communicate a wealth of propositions and affective attitudes with a single juxtaposition of ideas [12,8,6,10,7,3,15]. But as such, an effective metaphor requires effective communication, which in turn requires that the speaker has a clear idea of the content to be communicated, and an equally clear understanding of which conceptual vehicles best communicate this content. We present here a concise corpusderived meaning representation for metaphor processing that captures the most widely-used talking points that are evoked in everyday metaphors and similes. We illustrate how these talking points can be acquired by harvesting the web, and further show how comparable but discretely different talking points can be reconciled during metaphor processing. Finally, by replicating the clustering experiments of [1], we show that talking points yield an especially concise representation of concepts in general.