Contextualized Sarcasm Detection on Twitter

By David Bamman, Noah A. Smith. International Conference on Web and Social Media, 2015

409 citations recorded by the source corpus.

Sarcasm requires some shared knowledge between speaker and audience; it is a profoundly contextual phenomenon. Most computational approaches to sarcasm detection, however, treat it as a purely linguistic matter, using information such as lexical cues and their corresponding sentiment as predictive features. We show that by including extra-linguistic information from the context of an utterance on Twitter — such as properties of the author, the audience and the immediate communicative environment — we are able to achieve gains in accuracy compared to purely linguistic features in the detection of this complex phenomenon, while also shedding light on features of interpersonal interaction that enable sarcasm in conversation.

Subfields: Artificial Intelligence, Experimental and Cognitive Psychology

View Paper: https://www.semanticscholar.org/paper/0f9023736824d996414bac85a1388225d3c92283

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Experimental and Cognitive Psychology

Citations in the corpus

Beneath the Tip of the Iceberg: Current Challenges and New Directions in Sentiment Analysis Research (2020, background): “While convenient, obtaining labeled data from hashtags has been found to introduce both noise (incorrectly-labeled examples) and bias (only certain forms of sarcasm are likely to be tagged, Davidov et al. [31], and predominantly by certain types of Twitter users, Bamman & Smith [5]).”
Returning the N to NLP: Towards Contextually Personalized Classification Models (2020, background): “for sarcasm (Bamman and Smith, 2015), stance prediction (Sasaki et al.”
Towards Multimodal Sarcasm Detection (An _Obviously_ Perfect Paper) (2019, background): “Other research leverages context to acquire shared knowledge between the speaker and the audience (Wal-lace et al., 2014; Bamman and Smith, 2015).”
Are You a Racist or Am I Seeing Things? Annotator Influence on Hate Speech Detection on Twitter (2016, methodology): “Author Historical Salient Terms Given the promising results obtained for sarcasm detection (Bamman and Smith, 2015), we calculate the Au-thor Historical Salient Terms (AHST).”
That’s So Annoying!!!: A Lexical and Frame-Semantic Embedding Based Data Augmentation Approach to Automatic Categorization of Annoying Behaviors using #petpeeve Tweets (2015, background): “To the best of our knowledge, even though there have been studies on using Twitter hashtags to study language-related behaviors (GonzálezIbánez et al., 2011; Bamman and Smith, 2015),”
Computational Sociolinguistics: A Survey (2016, background): “While some features tend to be effective across domains, others are domain-specific (Nguyen, Smith, and Rosé 2011).”
SemEval-2018 Task 3: Irony Detection in English Tweets (2018, background): “Also beneﬁcial for this task are combinations of different feature types (e.g. Van Hee et al., 2016b), author information (e.g. Bamman and Smith, 2015), features based on (semantic or factual) opposi-tions (e.g Karoui et al., 2015; Gupta and Yang, 2017; Van Hee, 2017) and even eye-movement patterns…”