Scholarlink

This paper is the condensed overview of Touché: the sixth edition of the lab on argumentation systems that was held at CLEF 2025. With the goal to foster the development of support-technologies for decision-making and opinion-forming, we organized four shared tasks: (1) Retrieval-Augmented Debating (RAD), in which participants submit generative retrieval systems that argue against their users and evaluate such systems (new task); (2) Ideology and Power Identification in Parliamentary Debates, in which participants identify from a speech the political leaning of the speaker’s party and whether it was governing at the time of the speech (2nd edition); (3) Image Retrieval/Generation for Arguments, in which participants find images to convey a written argument (4th edition, joint task with ImageCLEF); and (4) Advertisement in Retrieval-Augmented Generation, in which participants generate responses to queries with ads inserted and detect such inserted ads (new task). In this paper, we describe these tasks, their setup, and participating approaches in detail.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Overview of Touché 2025: Argumentation Systems

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Johannes Kiesel,
  • Çağrı Çöltekin,
  • Marcel Gohsen,
  • Sebastian Heineking,
  • Maximilian Heinrich,
  • Maik Fröbe,
  • Tim Hagen,
  • Mohammad Aliannejadi,
  • Sharat Anand,
  • Tomaž Erjavec,
  • Matthias Hagen,
  • Matyáš Kopp,
  • Nikola Ljubešić,
  • Katja Meden,
  • Nailia Mirzakhmedova,
  • Vaidas Morkevičius,
  • Harrisen Scells,
  • Moritz Wolter,
  • Ines Zelch,
  • Martin Potthast,
  • Benno Stein

摘要

This paper is the condensed overview of Touché: the sixth edition of the lab on argumentation systems that was held at CLEF 2025. With the goal to foster the development of support-technologies for decision-making and opinion-forming, we organized four shared tasks: (1) Retrieval-Augmented Debating (RAD), in which participants submit generative retrieval systems that argue against their users and evaluate such systems (new task); (2) Ideology and Power Identification in Parliamentary Debates, in which participants identify from a speech the political leaning of the speaker’s party and whether it was governing at the time of the speech (2nd edition); (3) Image Retrieval/Generation for Arguments, in which participants find images to convey a written argument (4th edition, joint task with ImageCLEF); and (4) Advertisement in Retrieval-Augmented Generation, in which participants generate responses to queries with ads inserted and detect such inserted ads (new task). In this paper, we describe these tasks, their setup, and participating approaches in detail.