On the difficulty of automatically detecting irony: beyond a simple case of negation

Abstract

Rosso, P. (2014Abstract. It is well-known that irony is one of the most subtle devices used to, in a refined way and without a negation marker, deny what is literally said. As such, its automatic detection would represent valuable knowledge regarding tasks as diverse as sentiment analysis, information extraction, or decision making. The research described in this article is focused on identifying key values of components to represent underlying characteristics of this linguistic phenomenon. In absence of a negation marker, we focus on representing the core of irony by means of three conceptual layers. These layers involve 8 different textual features. By representing four available data sets with these features, we try to find hints about how to deal with this unexplored task from a computational point of view. Our findings are assessed by human annotators in two strata: isolated sentences and entire documents. The results show how complex and subjective the task of automatically detecting irony could be.

Analyzing and annotating for sentiment analysis the socio-political debate on #labuonascuola

“…Finally, we annotated irony, whose recognition is a very challenging task for the automatic detection of sentiment because the inferring process goes beyond syntax or semantics (Reyes et al, 2013;Reyes and Rosso, 2014;Maynard and Greenwood, 2014;Ghosh et al, 2015). As in Sentipolc (Basile et al, 2014), we were interested in annotating manually the polarity of the ironic tweets, where the presence of ironic devices can work as an unexpected "polarity reverser" (e.g.…”

A Proposal for New Evaluation Metrics and Result Visualization Technique for Sentiment Analysis Tasks

“…Finally, it is also worth mentioning that both tasks should take into consideration the presence of irony. Few works have dealt with the effect of irony when analyzing polarity [14,15], but its correct analysis should increase the performance of SA and RA approaches. Our intuition is that this phenomenon is more common in RA texts and can explain, to some extent, the remarkable differences in the results.…”

Single and Cross-domain Polarity Classification using String Kernels

“…Although polarity classification tasks can be tackled with text classification methods, it has been proven to be a more challenging task (Pang et al, 2002): sentiment may be expressed more subtly (Reyes and Rosso, 2013) than categories generally recognised with keywords alone. In addition, the cross-domain variant has the additional difficulty of using a different vocabulary among domains.…”