For deepening the understanding of media, communication and perception of Europe in the context of the migration process, MIRROR includes social science studies. The core of the developed theoretical framework for the fieldwork is a new Migration-Communication-Model. Field work itself was performed in 6 countries (186 migrant interviews, 42 expert interviews). The analysis of these interviews has resulted in important insight, for example, on the smart phone use by migrants and refugees, on their perceptions of the EU and its migration policy, and on gender differences in the role of mobile media use and EU Perceptions. Also for improving the domain understanding, we have developed a set of Migration-Related Semantic concepts.
MIRROR technology has been driven by the collected requirements and resulted in the overall MIRROR system architecture, information model and system integration approach. The MIRROR system enables perception analysis via rich media annotation, search and filtering, the definition/sharing of observed situations, and -via a Dashboard – the analysis of events, trends, etc For the system, innovative AI-based methods have been developed for text, audio-visual content as well as social media analysis considering the multilinguality of the sources. For text, e.g. methods for the detection of computer generated content and for pseudonymization have been developed. We also analysed the bias in media coverage for European topics in different countries. For image/video annotation methods for visual concept detection, image sentiment analysis, video captioning, summarization, and the detection of migration-related semantic concepts have been developed. For social media analysis, results include methods for bot detection and content summarization. The MIRROR system has been intensively validated in pilots in three countries with practitioners from the field. Overall, the satisfaction of the application partners with the MIRROR system was very high.
During the full duration of the project special attention was given to sound ethical practices and respect to data protection and other fundamental rights. The resulting MIRROR Ethical principles have been considered in method and technology development. Furthermore, a Human Rights check list has been developed and a data protection impact analysis has been performed for the project. Insights from this work have been summarized in a set of policy recommendations. In addition, factors that may influence the societal acceptance of systems such as the MIRROR system have been identified. This is crucial for preparing the operative use.
The goal of Perception-driven risks and threats in MIRROR was to find concrete ways in which existing threat analysis and risk assessment methodologies could be improved, especially by developing a more human-centric approach including the risks to migrants, not only those to EU Member States. Based on existing approaches, we have developed a new methodology, which supports the identification of perception-based threats from online content (perception model). For this, we linked key risk-relevant behaviours (identified together with the application partners) to attitudes (a.k.a. perceptions). The model has been operationalized by integrating it into the MIRROR system.
Insights from all areas of the project were collected in a Toolkit of Actionable Insights, which makes those insights available for interested stakeholders.
The project activities have been accompanied by dissemination activities for increasing visibility. As part of this, MIRROR organized two larger with different types of stakeholders. The events had more than 110 key target audience participants in total.
The MIRROR project has resulted in a number of exploitable assets. The most important ones for joint exploitation are the MIRROR System and the Toolkit of Actionable Insights. For the exploitation of the MIRROR system a multi-step strategy has been defined including further operation of the system, a business canvas for joint exploitation, marketing and networking activities, as well as the identification of funding opportunities for turning the MIRROR prototype into a product.
For easing exploitation of the Toolkit, it was decided to create an extensible online version. The mid-term exploitation perspective is to make the Toolkit a project-independent valuable resource and to continuously extend it in collaboration with other projects (e.g. CRiTERIA).
