Jeremy Ward et Senmiao Yang : Les controverses vaccinales dans les médias français

Jeremy Ward et Senmiao Yang : Les controverses vaccinales dans les médias français

🎙 Jeremy Ward, Senmiao Yang 👥 2K 📅 May 21, 2026 ⏱ 37 min 👁 32 📄 expert opinion 🧭 2026-08-15
Available in: English (current) Français

Keywords

vaccinationmedia coveragecontroversyFranceNLP

Summary

This seminar presentation by Jeremy Ward and Senmiao Yang, part of the INA le lab series, examines the emergence and evolution of vaccine controversies in French general information media. The researchers use a large corpus of transcriptions from INA archives spanning 15 years, focusing on segments mentioning ‘vaccin’. They employ a natural language inference approach, translating French segments to English and using a pre-trained model fine-tuned with active learning. The analysis reveals that media coverage is event-driven, with peaks during the H1N1 and COVID-19 pandemics, and dominated by continuous news channels like CNews, BFM TV, and Sud Radio. The study specifically analyzes the themes of vaccine safety and efficacy, finding that safety discussions are more prevalent in certain media, particularly Sud Radio, and that coverage patterns vary over time. The presentation highlights methodological challenges, such as model performance and the need for further qualitative analysis. Overall, the work provides a systematic, longitudinal perspective on how vaccine controversies are framed in French media, contributing to the sociology of media and vaccination.

170 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information lies in its systematic, longitudinal analysis of a large media corpus, offering empirical insights into the dynamics of vaccine controversies in France. The argumentation is solid, grounded in established sociological theories and supported by quantitative data. The speakers clearly explain their methodology and acknowledge limitations, such as model performance and the need for further qualitative analysis. The presentation effectively demonstrates how media coverage is shaped by event-driven factors and media-specific characteristics, providing a nuanced understanding of the topic.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high, with a transparent methodology and reliance on a comprehensive dataset from INA. The sources cited are primarily the INA archives and the project’s own analyses, with references to academic works such as those by Dominique Marchetti and Claire Sécail. The title accurately reflects the content, which focuses on vaccine controversies in French media. The presentation is well-structured and the speakers are credible researchers in the field.

169 words

Title / Content Match

The title accurately reflects the content, which focuses on vaccine controversies in French media.

Quality & Reliability

7/10

The presentation is based on a systematic analysis of a large corpus of French media transcripts, using a validated NLP classification method. The speakers are researchers from Inserm and CREST, providing expert insight. However, the results are preliminary and the method has limitations acknowledged by the authors.

Key Moments

Cited Sources

  • INAthèque — The archive of French audiovisual media used for the corpus.
  • INA le lab hypotheses — The seminar series where this presentation was given.

Concurring Sources

  • INAthèque — The archive used for the corpus, providing reliable data.

Contribution & Novelties

This presentation offers a novel systematic analysis of vaccine controversies in French media over 15 years, using advanced NLP methods. It provides empirical evidence on the role of media type and political orientation in shaping coverage, particularly highlighting the persistent focus on vaccine safety in Sud Radio. The methodological approach, combining translation and natural language inference, is innovative for French media analysis.

Pour aller plus loin :

104 words

Radar Profile

The radar profile shows balanced scores across all dimensions, with slightly higher scores in quantity of information and technical level, reflecting the comprehensive dataset and methodological sophistication. The lower score in quality of information suggests some limitations in the preliminary results.

Reliability 7/10