Arbeitsbereich

Bevölkerungsdynamik und Nachhaltiges Wohlbefinden

Auf einen Blick Projekte Publikationen Team

Projekt

Migration, Culture, and Inequalities in Algorithmically Mediated Societies (Dissertation)

Carolina Coimbra Vieira, Emilio Zagheni, Krishna Gummadi (Max Planck Institute for Software Systems, Saarbrücken, Deutschland); in Zusammenarbeit mit Diego Alburez-Gutierrez, Marília R. Nepomuceno, Tom Theile, Asli Ebru Sanlitürk (alle: MPIDR), Abhisek Dash, Sepehr Mousavi (beide: Max Planck Institute for Software Systems, Saarbrücken, Deutschland), Sophie Lohmann (MPIDR), Savvas Zannettou (Delft University of Technology, Niederlande), Oshrat Ayalon (University of Haifa, Israel), Fabrício Benevenuto, Marisa Vasconcelos, Pedro Olmo Stancioli Vaz de Melo (alle: Federal University of Minas Gerais, Belo Horizonte, Brasilien), Filipe Nunes Ribeiro (Federal University of Ouro Preto, Minas Gerais, Brasilien)

Ausführliche Beschreibung

The widespread use of online platforms has transformed how people interact, express opinions, and access information. As a byproduct of these interactions, users generate large-scale digital trace data, creating new opportunities for social science research. This project uses digital trace data from Facebook, Twitter (now X), TikTok, and Wikipedia to address questions about cultural similarity, human migration patterns, gender and social inequalities, vulnerable populations, and online behavior on algorithmically mediated platforms.

The studies collectively address four thematic areas: (i) culture, by developing measures of cultural similarity from Facebook data; (ii) migration, by applying cultural similarity measures to study international mobility and examining information-seeking behavior on Wikipedia during crises; (iii) inequalities and vulnerable populations, through analyses of gender gaps in STEM-related interests across countries and within Brazil, as well as the characterization of missing children in Guatemala; and (iv) online behavior, by investigating user engagement on TikTok.

The first study proposes cultural similarity measures derived from passively collected Facebook traces, using food and drink preferences as cultural markers. A follow-up study links these measures to international migration, showing that Facebook-based measures of cultural similarity improve the predictive power of traditional gravity models and have predictive capacity comparable to that of established predictors, such as shared language and history.

The third study examines information-seeking on Wikipedia during forced migration in the context of the 2022 Ukrainian refugee crisis. Using daily Wikipedia article views across language editions, we find that readership patterns correlate with the distribution of Ukrainian refugees in Poland and Germany and align temporally with border crossings. The analyses reveal that refugee arrivals precede increases in Wikipedia views, thus highlighting the reactive nature of information-seeking behavior during forced migration.

Two studies use Facebook data to assess global gender gaps in STEM across countries, and to examine disparities across age and education groups in Brazil. Despite Facebook’s bias toward women, female users dominate non-STEM interests, while STEM interests remain largely male-dominated. Within STEM, however, distinct patterns emerge: Women are more represented in life sciences and mathematics, whereas engineering and technology are strongly male-dominated.

An additional study uses Twitter to measure missing children in Guatemala between 2018 and 2020. By complementing official police records, the data provide detailed information on disappearances and enable the first systematic description of missing children by age, sex, and geographic distribution, thus demonstrating the value of digital data in contexts where official statistics are limited.

As platform APIs become increasingly restricted, the seventh study highlights data donation as a promising alternative. Using user-donated TikTok data, we examine user engagement with short-form video platforms. We find that video metadata, especially on video duration, are the strongest predictors of whether users watch a video until the end, while demographic characteristics add little explanatory power. Short-form videos are also harder to recommend than traditional items like movies, underscoring the challenges of modeling engagement in this context.

Overall, the project demonstrates how digital trace data can complement traditional data sources and expand research beyond the Global North, while also reflecting on evolving modes of data access, such as data donation, in light of increasingly restricted social media APIs.

Publikationen

Coimbra Vieira, C.; Mousavi, S.; Ayalon, O.; Dash, A.; Gummadi, K.; Zannettou, S.:

Exploring the limits of predicting user watching behavior with short-form videos on TikTok

In: WebSci Companion '26: companion publication of the 18th ACM Web Science Conference, TU Braunschweig, Germany, 26-29 May 2026, 142–148. New York: Association for Computing Machinery (ACM). (2026)    

Coimbra Vieira, C.; Şanlitürk, A. E.; Zagheni, E.:

Forced migration and information seeking behavior on Wikipedia: insights from the Ukrainian refugee crisis

In: Proceedings of the 20th International AAAI Conference on Web and Social Media (ICWSM-26): Los Angeles, CA, USA, May 27–29, 2026, 2358–2381. Washington, DC: AAAI Press. (2026)    

Coimbra Vieira, C.; Vasconcelos, M.:

Characterizing global gender gaps in STEM using Facebook data

In: 20th International Conference on Scientometrics and Informetrics, ISSI 2025: June 23-27, 2025, Yerevan, Armenia; proceedings: volume 1, 417–431. Yerevan: ISSI. (2025)    

Coimbra Vieira, C.; Lohmann, S.; Zagheni, E.:

The value of cultural similarity for predicting migration: evidence from food and drink interests in digital trace data

Population and Development Review 50:1, 149–176. (2024)    

Coimbra Vieira, C.; Alburez-Gutierrez, D.; Nepomuceno, M. R.; Theile, T.:

Desaparecidxs: characterizing the population of missing children using Twitter

In: WebSci '22: proceedings of the 14th ACM Web Science Conference, Barcelona, Spain, 26-29 June 2022, 185–190. New York: Association for Computing Machinery (ACM). (2022)    

Coimbra Vieira, C.; Lohmann, S.; Zagheni, E.; Vaz de Melo, P. O. S.; Benevenuto, F.; Ribeiro, F. N.:

The interplay of migration and cultural similarity between countries: evidence from Facebook data on food and drink interests

PLOS One 17:2, e0262947–e0262947. (2022)    

Coimbra Vieira, C.; Vasconcelos, M.:

Using Facebook ads data to assess gender balance in STEM: evidence from Brazil

In: WWW'21: companion proceedings of the Web Conference 2021, 145–153. New York: Association for Computing Machinery (ACM). (2021)