Stance Detection in German News Articles

Abstract

The widespread use of the Internet and the rapid dissemination of information poses the challenge of identifying the veracity of its content. Stance detection, which is the task of predicting the position of a text in regard to a specific target (e.g. claim or debate question), has been used to determine the veracity of information in tasks such as rumor classification and fake news detection (Küçük and Can, 2020). While most of the work and available datasets for stance detection address short texts snippets extracted from textual dialogues, social media platforms, or news headlines with a strong focus on the English language, there is a lack of resources targeting long texts in other languages. Our contribution in this paper is twofold. First, we present a German dataset of debate questions and news articles that is manually annotated for stance and emotion detection. Second, we leverage the dataset to tackle the supervised task of classifying the stance of a news article with regards to a debate question and provide baseline models as a reference for future work on stance detection in German news articles.

Pitch Proposal: Recommenders with a Mission - Assessing Diversity in News Recommendations

“…For the recommended items, only the top 8 are considered. 3 As the goal of the neural recommenders is to predict which items have been clicked, the category distributions for the neural recommenders often resembles the distribution in the 'clicked' column.…”

“…Automatic stance-or viewpoint detection based on article fulltext, which could be retrieved by following the url to the MSN News website, may be a direction for future research [7]. For example, Mascarell et al [3] published a detailed annotation of different stances and emotions present in German news articles. They do, however, lack the scale and user interactions that MIND has.…”

As Little as Possible, as Much as Necessary: Detecting Over- and Undertranslations with Contrastive Conditioning

“…Below we analyze the data overlap between pre-training and downstream evaluation: Note that the German sentences and Romansh sentences never appear together in the pre-training corpus, making it unlikely that overlap gives SwissBERT an advantage in the alignment task. We also make sure to exclude articles that occur in the CHeeSE dataset [43] to facilitate future evaluation on this dataset.…”

Towards Analyzing the Bias of News Recommender Systems Using Sentiment and Stance Detection

“…In [27] the authors create Cheese, a new dataset containing 3,693 pairs of debate questions and the associated Swiss news articles in German, annotated with stances and emotions. The news articles were published between 2004 and 2020, and cover 24 topics, including science, environment, politics, religion, society, etc.…”

“…Training Data Selection. There are two datasets which are introduced for German language, i.e., Cheese [27] which specifically focuses on German language, and the German part of x-stance [42].…”