apologies for cross-posting
We are pleased to announce the *GermEval Shared Task on Candy Speech Detection („Flausch-Erkennung“)*
This is the second call to participate in the shared task on candy speech detection („Flausch-Erkennung“). We invite everyone from academia and industry to participate in the shared task. The workshop discussing the results of this shared task is planned to be held in conjunction with the Conference on Natural Language Processing (KONVENS) in September 2025.
*Introduction* Numerous methods have been developed for detecting and censoring negative speech (e.g., hate speech or offensive or harmful language) on social media platforms. However, there is much less focus on identifying and promoting positive supportive discourse in online communities. Our shared task aims to address this gap and encourage researchers to focus on such positive expressions. The task is to identify expressions of candy speech (Flausch) in online posts (YouTube comments). We define candy speech as expression of positive attitudes on social media toward individuals or their output (videos, comments, etc.). The purpose of candy speech is to encourage, cheer up, support and empower others. It can be viewed as the counterpart to hate speech, as it also aims to influence the self-image of the target person or group, but in a positive way.
*Data* We will provide the participants with annotated training (and development) and unlabeled test datasets containing complete written, German language comment threads under YouTube videos posted by different content creators. The content creators and communities vary in topic, style, age group, etc. The training and test datasets do not overlap in terms of YouTube videos. Furthermore, the test dataset mostly contains (comments on) videos from content creators that are different from those in the training dataset. The communities commenting on these videos can therefore be expected to differ.
*Task Details* Candy speech detection is the task of identifying the presence of candy speech (at the span level) in a given YouTube comment and classifying each expression in one of the predefined categories. This shared task focuses on German speaking YouTube communities. Participants will be provided with a dataset of YouTube comments manually annotated for different types of candy speech expressions.
We offer the following two subtasks. Participants in this year's shared task may choose to participate in either subtask:
Subtask 1: Coarse-Grained Classification The goal of this subtask is to identify whether the given comment contains candy speech ("Flausch") or not. The dataset is manually annotated for the presence of candy speech.
Subtask 2: Fine-Grained Classification The goal of this subtask is to identify the span of each candy speech expression in a given text and classify it in one of the predefined categories. The dataset is manually annotated for 10 different types of candy speech expressions, such as “positive feedback”, “compliment”, “group membership” etc.
More details on the subtasks (including examples) can be found at the website of the shared task (see link below).
*Important dates* Trial data available: February 15, 2025 Training data available: March 3, 2025 Test data available: May 17, 2025 Evaluation start: June 16, 2025 Evaluation end: June 27, 2025 Paper submission due: July 11, 2025 Camera ready due: August 15, 2025 GermEval workshop: September 8 or 12, 2025 (co-located with KONVENS)
*Website* https://yuliacl.github.io/GermEval2025-Flausch-Erkennung/
*GermEval* GermEval is a series of shared task evaluation campaigns that focus on Natural Language Processing for the German language. GermEval has been conducted regularly since 2014 in co-location with KONVENS/GSCL conferences: https://germeval.github.io/tasks/
*contact email* Please send any enquiry to the following email address: germeval-2025-candy-speech@ruhr-uni-bochum.de
Best regards,
Yulia Clausen, Ruhr-Universität Bochum, Germany Tatjana Scheffler, Ruhr-Universität Bochum, Germany Michael Wiegand, Universität Wien, Austria