Predicting and organizing migration flows is an enormous challenge for the European Union and the stakeholders involved, e.g., in municipalities. The EU-funded project ITFLOWS therefore aims to generate new insights into migration flows. It is intended to enable decision-makers to take adequate organizational measures. In particular, the project focuses on the stages of reception and resettlement, but also on the settlement and integration of refugees. Precise models are supposed to provide deeper insights into future migration patterns and lay the foundation for the EUMigraTool (EMT). The EMT has two main functions: predicting migration flows and detecting risks of tensions related to migration.
The project consortium, coordinated by the University of Barcelona, brings together 14 partners from a wide range of disciplines to ensure that this complex topic is researched and handled comprehensively. Since September 2020, FIZ Karlsruhe with its research units Information Service Engineering (ISE) and Intellectual Property Rights (IGR) has been participating in the EU project ITFLOWS. The project will run until August 2023.
Research by FIZ Karlsruhe
Social media have become one of the most widely used channels for sharing opinions about social events, and migration is no exception. Researchers from the ISE research unit examined more than 200,000 migration-related tweets from 11 European countries (the United Kingdom, Germany, Spain, Poland, France, Sweden, Austria, Hungary, Switzerland, the Netherlands, and Italy) over a nine-year period (January 2013 –July 2021). The data generated in this process was analyzed to detect hate speech and extremist expressions using deep learning technologies on distributed platforms. They are used to identify possible tensions and causes of flight, as well as conflicts between refugees and residents at the destinations, and to make them visible.
As a first result of this analysis, a Migration Knowledge Base (MGKB) was created. MGKB statistics and other content analysis, including diagrams, are available at https://migrationskb.github.io/MGKB/. MGKB also serves as a basis for a migration monitoring and prediction tool that is currently being developed by other ITFLOWS partners.
Twitter data processing, and natural language processing in general, is a valuable source of information. However, it also raises complex and largely unexplored legal and ethical questions regarding data processing. In this context, FIZ Karlsruhe‘s IGR research unit is developing approaches to bring the processing of personal data in line with the legal framework (especially the GDPR). In doing so, the interests of researchers and those of the data subjects, for example Twitter users, must be thoroughly balanced on a regular basis. To facilitate this, a special platform for managing legal requirements (https://compliance-monitor.eu/cmp) is used, among other things. Important issues here include the scope of data collected, the usability of metadata, legal implications of entity linking (e.g., BLINK), and the use of specific language models.
Information Service Engineering Unit
Prof. Dr. Harald Sack, harald.sack [at] fiz-karlsruhe.de (harald[dot]sack[at]fiz-karlsruhe[dot]de)
Dr. Mehwish Alam, mehwish.alam [at] fiz-karlsruhe.de (mehwish[dot]alam[at]fiz-karlsruhe[dot]de)
Yiyi Chen, yiyi.chen [at] fiz-karlsruhe.de (yiyi[dot]chen[at]fiz-karlsruhe[dot]de)
Intellectual Property Rights Unit
Prof. Dr. Franziska Boehm, franziska.boehm [at] fiz-karlsruhe.de (franziska[dot]boehm[at]fiz-karlsruhe[dot]de)
Thilo Gottschalk, thilo.gottschalk [at] fiz-karlsruhe.de (thilo[dot]gottschalk[at]fiz-karlsruhe[dot]de)
Francesca Pichierri, francesca.pichierri [at] fiz-karlsruhe.de (francesca[dot]pichierri[at]fiz-karlsruhe[dot]de)