The MIRROR project is broken down into work packages to structure work to be done during the runtime and to enable multiple teams to simultaneously work on different components in the project. It is divided into 12 Work Packages that contain a group of related tasks. A description of each package is displayed below.
WP1: Project Management & Coordination (LUH)
WP1 concerns all aspects of project management, coordination, quality assurance, and risk management. It ensures the effective planning, implementation, coordination and achievement of the project activities and success and provides project structure and support to assist decision making, internal and external communications, encourage greater accountability and control, minimise risk, identify and address and exploit project related opportunities. Furthermore, the WP coordinates the quality control for project output and support the consortium to achieve their project objectives.
The project management encompasses contractual issues, technical, administrative, finance, communication and knowledge management inside the project, and external relationships between the project and the EC. The Project’s coordinator will lead the work package but will be assisted by members of the MIRROR management team including the Research Manager, the Integration Manager, the Community Liaison Manager and the work package leaders. This WP has the following objectives:
- Ensure effective planning, implementation, coordination, and achievement of the project activities, including timely production of deliverables and successful completion of the tasks.
- Provide project structure and support to assist decision making, internal and external communications, encourage greater accountability and control, minimize risk, identify and address and exploit project-related opportunities.
- Coordinate quality control for project output and help the consortium achieve their project objectives.
The work package is broken down into tasks. These include:
T1.1 Planning, Scheduling and Risk Management [M1-M36] Leader: LUH, Participants: CERTH, EURIX, SAIL, all
Detailed project planning and scheduling, including tasks, responsibilities and timescales will be performed in collaboration with the WP Leaders. Planning activities also includes regular revisiting of the project, which have to be considered for the project planning. The process for risk management will be detailed and documented in the MIRROR project handbook. Furthermore, a data and knowledge management plan will be created in close collaboration with WP3.
T1.2 General Project Management Activities [M1-M36] Leader: LUH
Project communications channels and protocols will be established and internal and external communication strategies will be managed. In collaboration with the WP leaders the work plan will be monitored, and – if required – reviewed and adapted in line with monitoring feedback. Up-to-date financial project-wide records will be maintained.
T1.3 Quality Plan and Progress Monitoring [M1-M36] Leader: CERTH, Participants: LUH, WP Leaders
An internal quality management will be agreed upon and documented in the early phase of the project.
T1.4 Reporting [M2-M36] Leader: LUH, Participants: all
Reporting structures will be established; partner reports will be collected and consolidated.
LUH (project coordinator) will lead WP1 and carry out overall project management. CERTH as the Research Manager will supervise T1.3. All partners will be involved in planning and reporting tasks.
WP2: Demand Analysis & Forecasting for Border control Units and Policymakers (Lead: LUH)
It is a goal of this WP to develop an improved and more detailed understanding of the demands and expectations of the different target groups towards the planned MIRROR functionality and solutions. This embraces the technical solutions as well as the actionable insights to be produced by the project.
The results of this WP provide the basis for the activities in the technical WPs involved in media analysis methods WP4-WP6, for WP7 working on the system architecture as well as for WP8, WP9, and WP10. Furthermore, a close interaction with WP3 is planned to already consider the legal, ethical and societal framework (early results) in this early phase of the project.
Description of work
T2.1 Scenarios and User Stories [M1-M4] Leader: FOI, Part.: Freunde, LUH, NOA, MPF, MoDAT
This task will start from the scenarios outlined in the proposal, will refine those scenarios and add further scenarios.
In a second step these scenarios will be divided into smaller user stories and/or use cases in close collaboration with the target users involved in the project. This activity will already include very early sketches of possible user interfaces, as well as the inspection of user interfaces of existing system for gathering ideas. It will also include the identification of migration-related-semantic-concepts (MRSCs).
T2.2 Requirement Analysis [M2-M6] Leader: LUH, Part.: EURIX, Freunde, NOA, MPF, MoDAT
The user stories defined in T2.1 will be prioritized and translated into requirements for the solutions developed in MIRROR. First of all, requirements towards the MIRROR system will be created in this way, which form the basis for the system functionality, and thus the work in WP4-WP7.Furthermore, the demands of border agencies and policy makers towards actionable insights will also be elaborated from the scenarios and the prioritized user stories
T2.3 Revisiting User Requirements [M17-M20] Leader: LUH, Part.: EURIX, Freunde, NOA, MPF, MoDAT
As a result of the validation activities the user stories and requirements will be revisited. This revisiting task equips the project with a high flexibility and the opportunity to react to evolving needs on the site of the target communities.
The demand analysis activities in WP2 will be driven by the border authorities NOA, MPF and MoDAT. For reducing the administrative overhead for those application partners LUH will lead and coordinate the WP. In its role as mediator between border authorities and system development, FOI will lead the task T2.1 in gathering scenarios and user stories. Freunde will contribute by gathering additional stakeholder requirements and EURIX will be involved for creating the link to the technical and integration activities (WP7).
WP3: Legal Compliance, Societal Acceptance & Ethical Framework WP Leader: RUG
MIRROR is built on strong respect for the rule of law, fundamental rights in particular rights to privacy, data protection, and expression. Moreover, MIRROR will be implemented taking into account the features of a complex phenomenon such as migration processes are and in alignment with social expectations of the involved actors. Therefore, the three main objectives for this WP are:
- To ensure that solutions developed and exploited by the consortium, its partners and stakeholders are compliant with ethical principles, applicable data protection legislation, and other legal frameworks.
- To ensure that the solutions developed are acceptable to the different social groups affected by them and by society at large.
- To ensure that the solutions developed to consider all the societal issues related to the migration processes towards Europe.
Description of work
T3.1 Adhere to Good Ethical Practices [M2-10, M30]: Leader: RUG, Part.: all
Subtask 3.1.1: The partners will compile a list of the titles and contact details (e-mail addresses and websites) of the national ethics committees where they exist in the countries of the partners or the partner institution’s own ethics committee (e.g., that of a university partner). RUG, as the leader of the ethics WP, will form a project ethics advisory board of three, independent, well-regarded ethicists and human rights civil society organizations (CSOs) to review and comment on ethical issues presented by the partners.
Subtask 3.1.2: Identify and assess ethical issues arising from the project: The partners will compile a list of all of the activities to be undertaken by the project and all of the deliverables to be produced and identify and assess any ethical issues that might arise from each of those activities and deliverables. The partners will filter these activities and deliverables by conducting an initial review to set aside those that don’t produce a risk or only a negligible one. The partners will discuss with the WP leader what measures could be taken to address or resolve the ethical issues.
T3.2 Human Rights Implications of Building a Sentiment Analysis Model for Border Security [M1-M12] Leader: RUG, Part.: UM, AgF, K&I, Freunde
This task will carry out a broad human rights analysis of the potential effects from a human rights perspective of using sentiment analysis techniques in border security. This task will not only focus on the rights of privacy, data protection, and association but will explore broader implications such as discrimination, inequality, bias, lack of due process, etc.
Based on this research, this task will prepare a human rights implication check-list that can be used by WP4, 5, 6 and 7.
T3.3 Privacy and Data Protection Impact Assessment [M4-M24] Leader: RUG Participants: UM, AgF, Freunde
This task will conduct a privacy and data protection impact assessment of the work being carried out in WP4, WP5, WP6, WP7, WP8, and WP9.It will, therefore, include an analysis of the purposes of the processing of personal data, the effects of the processing, considerations on consent, data minimization, rights of individuals, reliability of the data being collected, security and responsibilities of data controllers and data processors (not only under the General Data Protection Regulation but also under the Directive for the processing of personal data in the criminal law sector). Following from and together this impact assessment review, this task also explores and provides guidelines for more privacy and data protection friendly approaches to the work carried out in MIRROR.
T3.4 Factors Influencing Societal Acceptance [M4-M24] Leader: K&I; Participants: RUG, UM, Freunde
Any novel technology or organizational measure will need to be accepted by the concerned social groups. Therefore, this task aims to carry out a literature review of the societal aspects of the work being carried out in WP4, WP5, WP6, WP7, WP8, and WP9. This task will also carry out an analysis from the perspective of sociology and social sciences with a particular focus on the perceived and real impact of the use of the proposed solutions to be developed and exploited by the consortium, its partners and stakeholders and options and risks identifiable in future scenarios. Based on the literature review and analysis, this task will proceed to explore the possible societal acceptance of systems combining different proposed solutions.
T3.5 Societal Issues of Migration [M4-M30] Leader: K&I; Participants: UM, Freunde, AgF, RUG
Although there is no doubt that between Europe and countries of emigration there is a big difference in the level of well-being, it is also true that the perception of Europe that is generated abroad often make it possible to under evaluate some problems connected to migrations and therefore to generate some deviations concerning the perception of European reality.
T3.6 Practical Guidance and Policy Recommendations[M22-M35] Leader: RUG Participants: K&I, UM, AgF
Considering the previous tasks, the partners will propose practical guidance and policy recommendations to partners, technology developers, border security authorities, stakeholders and the European Commission with regard to possible ways of resolving legal, ethical and societal issues and risks that may arise in the course of using MIRROR solutions on a daily basis.
The work in this Legal Compliance, Societal Acceptance & Ethical Framework work package is led by RUG based on its strong legal and ethical issues research profile. K&I will lead the work on societal acceptance. They bring a strong societal dimension to the project. AgF brings in the ‘law in practice’ dimension through the civil society experience. UM will complement the legal work with an information policy dimension. Freunde’s experience on the ground with migrant communities will complement the theoretical findings on migration especially in tasks 3.4 and 3.5.
WP4: Text Analysis Methods for Social and Traditional Media; WP Leader: SAIL
This WP will develop and adapt a set of text-analysis technologies for the enrichment of information collected from traditional and social media. It will address the often-informal nature of communication (i.e. the different registers employed by communicating parties) and the target languages and dialects found to be relevant within the scope of the project (informed by the outcomes of WP2, Deliverables 2.1 and 2.2). Enrichment technologies will focus on immigration-related topics and concepts as well as sentiment/affective categories and links to the results of the threat analysis in WP9 will be created.
Technologies to be developed and adapted include (a) language identification: from a variety of texts from traditional and social media, taking into account that documents may contain mixed-text sections, (b) cleaning, tokenization and extraction of relevant sections of text, eliminating superfluous and irrelevant sections, (c) annotation with named entities and concepts including creation and adaptation of migration-related ontologies and annotation of these concepts in input texts (in a multi-lingual manner, in order to be able to annotate/detect them consistently across languages and media), (d) Sentiment/polarity analysis: annotation of emotionally-charged content with particular focus on social media contents (e) Clustering of documents according to content, building upon the above technologies and a model of word-embedding of migration specific terminology.
All technologies and models will be integrated in WP7 (also formats and interfaces will be defined there to allow for cross-media processing from multiple platforms and modalities).
WP4, WP5, and WP6 are working closely together, in order to create a portfolio of analysis methods that are aligned with each other and can create synergies by combining evidences coming from different types of analysis methods from all three WPs. This requires strong mutual interaction, which is fostered by following the same time line in all three work packages.
Description of work
T4.1 Text-extraction and Pre-processing [M1-M34] Leader: SAIL, Participants: LUH, FOI
The goal of this task is to access the different sources of textual content, extract documents, extract from these the relevant sections of text, clean and tokenize the text and perform normalization of content.
T4.2 Named Entity and Concept Detection [M1-M34] Leader: SAIL, Participants: LUH, EURIX
The goal of this task is to enrich text-documents (and textual-meta-data) with migration-related-semantic-concepts (MRSCs). Based upon an existing infrastructure for pattern-based detection of Named-Entities as well as statistical tagging of entities, the mechanisms and models will be extended and adapted to the target domains and use-cases of MIRROR. On the one hand, this will entail adaptation on the morphological level to support the set of languages and dialects targeted; on the other hand, this will entail the creation of new categories of entities having been identified to be of importance in the migration domain (by WP2). In a second phase, these classes and instances of entities will be connected into a migration domain specific ontology allowing to link related concepts in a multi-lingual manner. Results of this task will be used in T4.3, T4.4 and T5.1.
T4.3 Sentiment/Polarity/Stance Detection [M1-M34] Leader: SAIL, Participants: FOI, CERTH
The objective of this task is to create and extend appropriate technologies and models for annotation of sentiment polarity and stance of textual documents.
Further work in this task will study the phenomenon from a psychological background, relating the use of words and expressions with psychological phenomena providing an important window into writers’ intentions and readers’ possible perceptions. Results of this task will be used in T5.2.
T4.4 Early Topic Detection and Topic Evolution [M1-M36] Leader: LUH, Participants: SAIL, CERTH
This task focuses on mining and tracking of migration-related topics discussed in the media. We will examine, extend, and adapt advanced methods for topic detection and temporal topic modelling. We will also develop novel methods for deriving topics from multi-modal, multilingual and heterogeneous data sources as well as for joint modelling of topics and networks/communities.
T4.5 Impact of Editorial Information and Perceived Information Quality [M1-M34] Leader: FOI, Part.: SAIL
Different kinds of news and media inform as well as influence people through editorial material.
In this task the work will focus on developing techniques and methods for analysing what kind of editorial messages related to migration are spread in relevant media sources (as identified in WP2). Perceived information quality (IQ) is known to be a determining factor in accepting and sharing information. The task will investigate how perceived IQ and the choice of editorial material correspond and might be used for manipulation.
SAIL will lead the WP and together with LUH and FOI will do the main method development in the WP. CERTH will be responsible for creating links to related tasks in WP5. K&I will advise the WP team in gender-related issues and aspects of societal acceptance. EURIX will be involved for creating links to WP7 (integration, information model).
WP5: Multimedia analysis methods for Social and Traditional Media (CERTH)
WP5 is responsible for the development of multimedia analysis technologies for the annotation and summarization of social and traditional media using new deep learning architectures. Multimedia items are enriched with migration-related semantic concepts and sentiment/affective categories. Based on the enriched annotations, a novel multimodal media summarization technique is developed for selecting the most appropriate media items and for creating descriptive representations of large media collections.
WP6: Methods for Cross-Media Network Analysis (LUH)
The goal of WP6 is to explore, examine, adapt, and develop analysis methods for migration-related social networks collected from available sources. These methods provide target users (such as border control agencies) the tools and frameworks for exploring, understanding, and inspecting the possible sources of misperceptions that can result in threats. Additionally, this WP provides methods for analysis of the misperception evolution across the geographical areas of interest as well as along the migration paths.
WP7: Methods for Cross-Media Network Analysis (LUH)
WP7 delivers a lightweight, flexible and extensible framework, implementing the reference scenarios and supporting a variety of integration mechanisms, metadata and protocols. The MIRROR framework is designed and implemented in compliance to a reference model for information management, enabling the representation of the different data sources and the combination of the results from the automated media analysis and from empirical studies into a coherent model. The framework is capable of integrating in a central repository the data collected from external sources (e.g. from the Web) or produced by the different technologies developed in the scope of the project enabling data exchange, transformation and persistence with specific APIs and workflows.
WP8: Analysis & Field Studies for EU Perception and Media Impact (UNIVIE)
WP9: Threat Analysis & Actionable Insights (UM)
The aim of this WP9 is (a) to analyse threats related to the misperception of Europe heavily relying on existing threat analysis results, (b) to integrate results created in the MIRROR project into threat analysis and risk assessment methodologies used by border control across Europe, and (c) to provide border control & foreign policy bodies with a set of legal, ethical, technological, organizational etc. actionable insights, best practices and recommendations with respect to the use of social media in addressing security challenges generated by migration flows.
WP10: Pilot, Evaluation and Validation (UM)
WP10 defines and operates pilot studies in Sweden, Austria and Malta to validate the MIRROR approach and the technical solutions developed through the target communities of border agencies and policy makers and to assess to which degree the developed solution meets the expectations of the target communities. The results of each pilot are closely monitored and fed to the technical team, thus allowing the MIRROR approach to be improved from one pilot to the next
WP 11: Dissemination, communicationand exploitation (SAIL)
WP11 fosters the project impact by planning and executing activities regarding the dissemination, communication, and exploitation of the project and its results. The work includes the creation of a project identity, the establishment of an online presence as well as social media channels, the production of promotional and informational material, actual communication activities via diverse channels, and the development of the exploitation plan for the project results for future research and commercialization.
WP12: Ethics requirements (LUH)
WP12 sets out the ‘ethics requirements’ that the project must comply with which performed in close cooperation with RUG (WP3 – Legal Compliance, Societal Acceptance & Ethical Framework). This includes data protection issues and ethical approval of field work as well as the formation of the MIRROR Ethics board.