SemEval-2023 Task 9: Multilingual Tweet Intimacy Analysis

Multilingual Tweet Intimacy Analysis is a task to predict the intimacy of tweets in different languages. The task focuses on on perceived intimacy by asking annotators to give their subjective judgment of tweet intimacy.  Each annotator is asked to answer “How intimate do you think the given tweet is?” using a 1-5 likert scale. Tweets are annotated in 10 languages. The training data contains labeled intimacy for six languages: English, French, Spanish, Italian, ortuguese, and Chinese. To encourage new studies on understanding intimacy in language four other languages are included without training data (Dutch, Hindi, Korean, and Arabic). The participants are asked to build models that can predict tweet intimacy from 1 (not intimate at all) to 5 (very intimate).