TRACES Bulgarian Twitter Dataset on Covid-19 Annotated with Linguistic Markers of Lies

This dataset is provided under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA) with the additional terms below. After reading the terms, please state that you read and accept the conditions and describe the entity you are applying from (if it exists - e.g. academic institution, company, government agency), and your intended use of the dataset.

The TRACES team members will review your application and you may be granted access or not.

If you have questions, please contact us at: irina.temnikova@gmail.com

Conditions for using the dataset TRACES_Dts9.1.3_socialmedia_AutomaticAnnTwitterCovid_(version).csv:

In order to be allowed access to this dataset, in line with applicable legislation, including but not limited to the General Data Protection Regulation (GDPR), the Artificial Intelligence Act (AI Act, current draft as of 01 November 2022, pending adoption and entry into force), the TRACES Project Data Management Plan, as well as the Twitter requirements, if you want to download or use this dataset, you must agree with and abide with the following terms and conditions:

• The dataset has been created with Twitter Academic Access and thus cannot be used for commercial purposes.

• The dataset is anonymized, and all personal information, revealing the authors is removed. The dataset cannot be used for profiling Twitter users or for any applications, which breach the AI Act’s provisions. The identity reconstruction of the authors of the social media datasets is forbidden.

• Upon request of the authors of the tweets or of the TRACES team, specific tweet IDs and their annotations must be deleted.

• The social media posts, included in this dataset, are annotated with linguistic markers *potentially* signaling lies. The presence of such markers should not indicate that the social media posts contain disinformation, misinformation, lies, untrue facts and/or other inconsistencies with 100% confidence, but with a lower degree of confidence (certain likelihood, but not certainty).

• The linguistic markers are currently being developed and are provided purely and solely for scientific purposes. They cannot be used as conclusive evidence, as arguments on the merits of the dataset, as evidence in judicial or administrative proceedings or in any other way not directly related to Project TRACES.

• No legal action should or could be taken against the authors of the social media posts, included in the dataset, solely based on the presence of linguistic markers, potentially signaling lies.

• The dataset is not suitable to be used and shall not be used for governmental or public authority purposes, including for investigations, government surveillance, intelligence work, analysis, criminal investigation, court or administrative proceedings. 

• The presence of linguistic markers of potential lies in social media posts is a not statement/belief/affirmation of the Project's team members or affiliated institutions. 

• The Project Sponsors (AI4Media, F6S, and the European Commission), the members of the TRACES team, users, or subjects shall not be liable or otherwise responsible for any consequences and/or damages (including pecuniary or moral damages) arising out of or in relation to the Project, the data collected, and the methods used for their analysis and/or the results/outcomes.

• This notice, as well as all the activities of the TRACES Project and of its Project Sponsors, team members, users, or subjects, including any contractual and/or non-contractual liability, are governed exclusively by the European Union laws and by the laws of the Republic of Bulgaria.

• You agree to provide attribution to the TRACES project in the following format:

  • The TRACES project (https://traces.gate-ai.eu/)

  • Dataset name: TRACES_Dts9.1.3_socialmedia_AutomaticAnnTwitterCovid_1

  • Data source: Twitter.

  • Research article to cite: Irina Temnikova, Silvia Gargova, Ruslana Margova, Veneta Kireva, Ivo Dzhumerov, Tsvetelina Stefanova and Hristiana Nikolaeva (2023) New Bulgarian Resources for Detecting Disinformation. 10th Language and Technology Conference: Human Language Technologies as a Challenge for Computer Science and Linguistics (LTC'23). Poznań. Poland.

  • Link to the original dataset: https://zenodo.org/record/7614247