The workshop will host two shared tasks on metaphor and sarcasm detection.

Metaphor detection shared task

Following a successful shared task on metaphor detection in news, editorial, conversations, and academic writing sampled from the BNC during the Workshop on Figurative Language Processing in 2018, we will expand to a new domain and conduct a shared task on detection of metaphors in persuasive essays written by non-native speakers of English. Beigman Klebanov, Leong, and Flor (NAACL 2018) showed that usage of metaphorical language that is related to the writer’s arguments is correlated with the human holistic scores of essay quality. We will use the annotated dataset from the Beigman Klebanov et al (2018) study to conduct the shared task, and use their results as a baseline. We also intend to publish the features to help participants directly build on the prior work. In addition, we will run a second round of competition on the BNC data, to help track improvements on this benchmark since the last shared task.

For more information about the shared task and to participate visit our CodaLab website.

Important dates:

• January 12, 2020: CodaLab competition is open; training data and auxiliary scripts can be downloaded
• February 12, 2020: Test data can be downloaded and results submitted; performance will be tracked on CodaLab dashboard
• April 11, 2020: Last day for submitting predictions on test data
• April 23, 2020: Papers describing the systems are due
• May 15, 2020: Notification of acceptance
• May 25, 2020: Camera-ready papers due
• July 9, 2020: Workshop

Contact:

• Beata Beigman Klebanov
• Ben Leong

Sarcasm detection shared task

The second shared task will be on sarcasm detection. Sarcasm detection has received considerable attention in the NLP community in recent years (Joshi, 2016). This current shared task aims to study the role of conversation context for sarcasm detection (Ghosh et al., 2018). We will be using two different datasets: Twitter conversations and conversation threads from Reddit. For both datasets, we will provide the immediate context (i.e., only the previous dialogue turn) as well as the full dialogue thread, when available. The goal is to understand how much conversation context is needed or helpful for sarcasm detection.

For more information about the shared task and to participate visit our CodaLab site.

Important dates:

• January 19, 2020: CodaLab competition is open; training data can be downloaded
• February 19, 2020: Test data can be downloaded, and results submitted; performance will be tracked on CodaLab dashboard
• March 22, 2020: Last day for submitting predictions on test data
• April 23, 2020: Papers describing the systems are due
• May 15, 2020: Notification of acceptance
• May 25, 2020: Camera-ready papers due
• July 9, 2020: Workshop

Contact

• Debanjan Ghosh
• Smaranda Muresan