The OpiNews project consists in building an algorithm capable of automatically evaluating to what extent a French-language press article can be considered subjective, or how much its writing was influenced by the opinion of the journalist who wrote it.

Our goal is to produce results that are as efficient and transparent as possible, so that we can provide truthful explanations for the decisions made by the algorithm. To achieve this, we seek to combine deep learning approaches based on powerful language models trained on large amounts of data, with state-of-the-art methods from traditional linguistics and natural language processing.

  • Timeschedule : from October 2021 to October 2025
  • Supervisors : Antonin Descampe & Cédrick Fairon
  • Researcher : Louis Escouflaire