Call for Interest — ChildSafeAds @ NLLP 2026
11 views
Skip to first unread message
Jerry Spanakis
Jul 13, 2026, 4:20:55 AM (5 days ago) Jul 13
to Jurix Foundation for Legal Knowledge Based Systems
We are organising ChildSafeAds, a shared task on detecting and classifying commercial content in child-facing YouTube videos, co-located with the NLLP workshop at EMNLP 2026.
Expression of interest via this form: https://forms.office.com/e/q0JSkSuGzD
Influencer marketing directed at children is widespread and largely non-compliant with EU advertising disclosure law. This task is the first NLP benchmark to address detection, categorisation, and compliance assessment together, using a large-scale dataset of community-verified commercial segments from SponsorBlock. Submissions will be handled through CodaBench. Selected submissions will be invited to write and submit a system description paper, which, upon review, will be published with the workshop proceedings in the ACL Anthology.
Three sub-tasks:ST1 — Commercial Type: classify the legal nature of the promotion (physical goods, digital content, services, digital services, nothing)ST2 — Product Category: multi-label classification across toys, food, apps, fashion, health, education, financial products, and moreST3 — Compliance Risk Flags: retrieval-augmented classification over a legal reference corpus (EU directives, CJEU case law, platform guidelines); systems identify one or more flags (such as undisclosed advertising, direct exhortation, prohibited product category, misleading claim) grounded in retrieved legal provisions
Tentative TimelineTraining and development data release: 20 July 2026Evaluation phase opens: 03 August 2026System submission deadline: 11 August 2026System description paper deadline (upon invitation): 1 September 2026Notification (for system description papers): 15 September 2026Camera-ready (for system description papers): 22 September 2026NLLP @ EMNLP 2026: 28 October 2026
The full call for participation with data details and evaluation guidelines will follow shortly.
For any questions or clarifications, contact the lead organiser: Thales Bertaglia (Utrecht University): t.f.costa...@uu.nl.
Expression of interest via this form: https://forms.office.com/e/q0JSkSuGzD
Influencer marketing directed at children is widespread and largely non-compliant with EU advertising disclosure law. This task is the first NLP benchmark to address detection, categorisation, and compliance assessment together, using a large-scale dataset of community-verified commercial segments from SponsorBlock. Submissions will be handled through CodaBench. Selected submissions will be invited to write and submit a system description paper, which, upon review, will be published with the workshop proceedings in the ACL Anthology.
Three sub-tasks:ST1 — Commercial Type: classify the legal nature of the promotion (physical goods, digital content, services, digital services, nothing)ST2 — Product Category: multi-label classification across toys, food, apps, fashion, health, education, financial products, and moreST3 — Compliance Risk Flags: retrieval-augmented classification over a legal reference corpus (EU directives, CJEU case law, platform guidelines); systems identify one or more flags (such as undisclosed advertising, direct exhortation, prohibited product category, misleading claim) grounded in retrieved legal provisions
Tentative TimelineTraining and development data release: 20 July 2026Evaluation phase opens: 03 August 2026System submission deadline: 11 August 2026System description paper deadline (upon invitation): 1 September 2026Notification (for system description papers): 15 September 2026Camera-ready (for system description papers): 22 September 2026NLLP @ EMNLP 2026: 28 October 2026
The full call for participation with data details and evaluation guidelines will follow shortly.
For any questions or clarifications, contact the lead organiser: Thales Bertaglia (Utrecht University): t.f.costa...@uu.nl.
Reply all
Reply to author
Forward
0 new messages