Overview of the EVALITA 2018 Task on Irony Detection in Italian Tweets (IronITA)

Abstract

IronITA is a new shared task in the EVALITA 2018 evaluation campaign, focused on the automatic classification of irony in Italian texts from Twitter. It includes two tasks: 1) irony detection and 2) detection of different types of irony, with a special focus on sarcasm identification. We received 17 submissions for the first task and 7 submissions for the second task from 7 teams.

Detecting Hate Speech for Italian Language in Social Media

“…The team resham further investigated the sentiment polarity retrieved from SentiWordNet . Differently, the team SB exploited a manually modeled lexicon for the misogyny detection task proposed in (Frenda et al, 2018a). The HurtLex lexicon has been used by the team 14-exlab also for the English task.…”

“…• ABSITA -Aspect-based Sentiment Analysis. The task is organized as a cascade of two subtasks consisting in automatically annotating sentences from hotel reviews with respect to the identified aspects (Aspect Category Detection (ACD) subtask) and the polarity associated to each one of them (Aspect Category Polarity (ACP) subtask) (Basile et al, 2018a);…”

“…The intersection includes tweets related to three social groups deemed as potential target for hate speech online: immigrants, Muslims and Roma. Also, the sentiment corpora with multi-layer annotations developed in last years by the EVALITA community, which included also morpho-syntactic and entity linking annotations, were exploited by some ABSITA (Basile et al, 2018a) and IronITA participants to address the finer-grained sentiment related tasks proposed this year under the Affect, Creativity and Style track.…”

Multilingual Irony Detection with Dependency Syntax and Neural Models

“…• (RQ-3) Are results obtained using syntactic features stable across different languages? -We experiment with datasets made available from previous shared tasks on irony detection in four languages: French (DEFT 2017 (Benamara et al, 2017)), English (SemEval-2018Task 3 (Van Hee et al, 2018), Spanish (IroSvA (Ortega-Bueno et al, 2019)) and Italian (IronITA (Cignarella et al, 2018)). The obtained results overcome the competitive baselines and are also favorably comparable with the best results reported in shared tasks on irony for the four studied languages.…”

“…Being irony a pragmatic device inherently related to culture and language, the task captured the attention of various researchers, dealing with different languages and mostly focusing on social media texts. The presence of irony for Italian has been investigated -among other things-in the SENTIPOLC task of the EVALITA 2014 (Basile et al, 2014) and 2016 (Barbieri et al, 2016) and at IronITA (Cignarella et al, 2018). It was addressed for French within the context of DEFT 2017 (Benamara et al, 2017), for English at Semeval 2018 (Task 3: Irony Detection in English Tweets) (Van Hee et al, 2018) as well as for Spanish at the IroSVA (González et al, 2019).…”