UPB at SemEval-2020 Task 11: Propaganda Detection with Domain-Specific Trained BERT
This addresses the issue of manipulative news influencing public opinion, but it is incremental as it applies an existing method to a specific domain.
The paper tackled the problem of detecting propaganda techniques in news articles by specializing a pre-trained BERT model on propagandistic and hyperpartisan data, achieving an F1-score of 46.060% in span identification (ranking 5th) and 54.302% in technique classification (ranking 19th).
Manipulative and misleading news have become a commodity for some online news outlets and these news have gained a significant impact on the global mindset of people. Propaganda is a frequently employed manipulation method having as goal to influence readers by spreading ideas meant to distort or manipulate their opinions. This paper describes our participation in the SemEval-2020, Task 11: Detection of Propaganda Techniques in News Articles competition. Our approach considers specializing a pre-trained BERT model on propagandistic and hyperpartisan news articles, enabling it to create more adequate representations for the two subtasks, namely propaganda Span Identification (SI) and propaganda Technique Classification (TC). Our proposed system achieved a F1-score of 46.060% in subtask SI, ranking 5th in the leaderboard from 36 teams and a micro-averaged F1 score of 54.302% for subtask TC, ranking 19th from 32 teams.