BEGIN:VCALENDAR
VERSION:2.0
PRODID:OpenCms 20.0.18
BEGIN:VTIMEZONE
TZID:Europe/Berlin
X-LIC-LOCATION:Europe/Berlin
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700329T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701025T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
END:STANDARD
END:VTIMEZONE				
BEGIN:VEVENT
DTSTAMP:20231129T065942
UID:7d237aae-8e7c-11ee-9d9c-000e0c3db68b
SUMMARY:Democratically healthy news recommendation: aligning NLP with society, theory, and evaluation
DESCRIPTION:Democratically healthy news recommendation: aligning NLP with society, theory, and evaluationNews recommender systems provide news article recommendations based on a user's interests and clicks. This personalization could harm democratic societies: citizens may be unaware of information beyond their own interests, leading to filter bubbles and a lack of shared public debate. A healthy collective information environment may require more diversity in recommendations, such as recommending different viewpoints on societal debates. Natural Language Processing could play a role in providing more diverse recommendations, but solving such a complex societal problem requires several key ingredients. These include interdisciplinary collaboration with experts on democracy, careful reflection on the suitability of existing NLP tasks and models, and data that is connected to relevant social science theory. Additionally, evaluation is essential: do we measure what we intend to measure, and is our NLP model actually improving recommendations in terms of our democratic values? I will discuss my PhD projects and findings - which include the lack of cross-topic robustness of stance detection models, and evaluation of models beyond predictive validity - and relate these to how we can tackle the problem of non-diverse news recommendation, and similar complex societal research questions.BioMyrthe Reuver is a 4th year PhD candidate in the Computational Linguistics and Text Mining Lab (CLTL) at the Vrije Universiteit Amsterdam, with advisors prof.dr. Antske Fokkens (CLTL) and prof.dr. Suzan Verberne (Leiden University). Myrthe’s PhD is on analyzing diversity in news recommender systems, and her research has focussed on computational argumentation, precise evaluation, and interdisciplinary collaboration with social scientists and philosophers. Aside from her research papers, Myrthe has also applied NLP research during a summer internship at LinkedIn in Dublin, and has been featured in Dutch national newspapers about responsible use of AI.
DTSTART;TZID=Europe/Berlin:20231204T140000
DTEND;TZID=Europe/Berlin:20231204T160000
LOCATION: , V5.01 on the ground floor, Pfaffenwaldring 5 b, 70569 Stuttgart 
URL;VALUE=URI:https://www.f05.uni-stuttgart.de/fakultaet/aktuelles/veranstaltung/Democratically-healthy-news-recommendation-aligning-NLP-with-society-theory-and-evaluation/
END:VEVENT
END:VCALENDAR