Germa
A comprehensive corpus of untrustworthy German news
Bibliographic Data
| ID | 21227101 |
|---|---|
| Authors | Fabio Carrella (0000-0003-4918-3875, University of Bristol, corresponding author), Alessandro Miani (0000-0001-6610-3510, University of Bristol) |
| Year | 2025 |
| Publication date | 2025-02-10 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Linguistics Vanguard (JOURNAL) |
| Journal identifiers | ISSN: 2199-174X • E-ISSN: 2199-174X |
| Publisher | Walter de Gruyter GmbH (PUBLISHER • DE) |
| DOI | 10.1515/lingvan-2024-0064 |
| OpenAlex | W4407289338 |
| Language | EN |
| References cited | 37 |
The proliferation of online misinformation undermines societal cohesion and democratic principles. Effectively combating this issue relies on developing automatic classifiers, which require training data to achieve high classification accuracy. However, while English-language resources are abundant, other languages are often neglected, creating a critical gap in our ability to address misinformation globally. Furthermore, this lack of data in languages other than English hinders progress in social sciences such as psychology and linguistics. In response, we present GERMA, a corpus comprising over 230,000 German news articles (more than 130 million tokens) gathered from 30 websites classified as “untrustworthy” by professional fact-checkers. GERMA serves as an openly accessible repository, providing a wealth of text- and website-level data for testing hypotheses and developing automated detection algorithms. Beyond articles, GERMA includes supplementary data such as titles, publication dates, and semantic measures like keywords, topics, and lexical features. Moreover, GERMA offers domain-specific metadata, such as website quality evaluation based on factors like bias, factuality, credibility, and transparency. Higher-level metadata incorporates various metrics related to website traffic, offering a valuable tool into the analysis of online user behavior. GERMA represents a comprehensive resource for research in untrustworthy news detection, supporting qualitative and quantitative investigations in the German language
Credibility · Data science · German · Information retrieval · Linguistics · Metadata · Misinformation · Political science · World Wide Web · Computer Science · Hate Speech and Cyberbullying Detection · Misinformation and Its Impacts · Spam and Phishing Detection
Concreteness ratings for 40 thousand generally known English word lemmas
Quanteda
Age-of-acquisition ratings for 30,000 English words
Topicmodels
“Liwc auf Deutsch”
Loose and Tight
Where ‘fake news’ flourishes
A topic models analysis of the news coverage of the Omicron variant in the United Kingdom press
Digital Technologies and Selective Exposure
The language of conspiracy
The ‘Alice in Wonderland’ mechanics of the rejection of (climate) science
Representations of gender in conspiracy theories
From alternative conceptions of honesty to alternative facts in communications by US politicians
The Psychological Meaning of Words
| Citation velocity | historical |
|---|---|
| Highly cited | No |