Information Refinement Technologies for Crisis Informatics

The investigation of crisis communication, management and technology has become more common in domains such as Computer Supported Cooperative Work (CSCW), Human Computer Interaction (HCI), Computer Science (CS) and Information Systems (IS) as the work of citizens, emergency services, and other organizations is increasingly mediated by computer technology, notably by the use of mobile technologies and social media. In academics, social media is often defined as a group of internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of user-generated content. Whereas Web 2.0 was conceptualized as an architecture for participation with new possibilities for social interaction, user-generated content refers to the various forms of media content that are publicly available and created by end-users.

This chapter introduces the application field and research domain of this thesis. First, it will define and discuss different foundations and terms, such as incidents, emergencies, crises, disasters, and catastrophes, with regard to crisis management. Thereafter, it will introduce crisis informatics, focusing on social media technologies and mobile crisis apps. A further emphasis is set on social media analytics, which be discussed from the perspectives of algorithms and data, interfaces and systems, and human interaction. Finally, the chapter will establish a basic definition and understanding for information refinement in crisis informatics and elaborate open research gaps and potentials that are addressed in this thesis.

This chapter presents the overall research design of the dissertation. Based on the research field and epistemological foundations, I argue for the research approach of extended design case studies, which were applied in three research projects using a multitude of methods for literature reviews, empirical inquiry as well as design and evaluation of artefacts.

Social media has been established in many larger emergencies and crises. This process has not started just a few years ago, but already 15 years ago in 2001 after the terrorist attacks of 9/11. In the following years, especially in the last 10, sometimes summarized under the term crisis informatics, a variety of studies focusing on the use of ICT and social media before, during or after nearly every crisis and emergency has arisen. This chapter aims to recapitulate 15 years of social media in emergencies and its research with a special emphasis on use patterns, role patterns and perception patterns that can be found across different cases in order to point out what has been achieved so far, and what future potentials exist.

Security and conflict research mainly focuses on issues of vulnerability, reinforcement of resilience and preservation or recovery of critical infrastructures. Additionally, the importance of social media and crisis apps is being increasingly acknowledged. To what degree is German civil society effectively prepared for the case a crisis occurs? Which kinds of ICT are being used in everyday life and in a potential infrastructure failure? Our contribution presents the results of a representative survey with 1,024 participants in Germany, which demonstrate that threat awareness, preparation, effective crisis management and distribution of crisis apps are still relatively low in Germany, while traditional communication and information channels as well as informal information networks are preferred. The results additionally point out the significant potential for support which ICT have for increasing risk awareness, facilitating information transfer and improving communication between civil society, critical infrastructure operators and authorities.

The value of social media in crises, disasters, and emergencies across different events (e.g. floods, storms, terroristic attacks), countries, and for heterogeneous participants (e.g. citizens, emergency services) is now well-attested. Existing work has examined the potentials and weaknesses of its use during specific events. Fewer studies, however, have focused on citizens’ perceptions of social media in emergencies, and none have deployed a representative sample to examine this. We present the results of the first representative study on citizens’ perception of social media in emergencies that we have conducted in Germany. Our study highlights, for example, that around half (45%) of people have used social media during an emergency to share and/or look for information. In contrast, false rumours on social media (74%) are perceived as a threat. Moreover, only a minority of people have downloaded a smartphone app for emergencies (16%), with the most popular ones’ weather and first aid apps.

Although media and ICT play an increasingly large role in crisis response and management, in-depth studies on crisis apps and similar technology in the context of an emergency have been missing. Based on responses by 1,024 participants in Germany, we examine the diffusion, usage, perception and adoption of mobile crisis apps as well as required functions and improvements. We conclude that crisis apps are still a little-known form of disaster ICT, but have potential for enhancing communication, keeping users up to date and providing a more effective crisis management as supplement to other media channels dependent on different underlying infrastructures. However, they should be adaptable to user characteristics, consider privacy, allow communication and offer valuable information to raise awareness of potential disasters without creating an overload. Also, the familiarity with and trust in crisis apps should be addressed to maximize their beneficial impact on crisis communication and management. We discuss further implications as well as directions for future research with larger target groups and specific usage scenarios.

Social media have been established in many natural disasters or human-induced crises and emergencies. Nowadays, authorities, such as emergency services, and citizens engage with social media in different phases of the emergency management cycle. However, as research in crisis informatics highlights, one remaining issue constitutes the chaotic use of social media by citizens during emergencies, which has the potential to increase the complexity of tasks, uncertainty, and pressure for emergency services. To counter these risks, besides implementing supportive technology, social media guidelines may help putting artefact and theoretical contributions into practical use for authorities and citizens. This chapter presents the design and evaluation (with 1,024 participants) of citizens’ guidelines for using social media before, during, and after emergencies.

The research field of crisis informatics examines, amongst others, the potentials and barriers of social media use during conflicts and crises. Social media allow emergency services to reach the public easily in the context of crisis communication and receive valuable information (e.g. pictures) from social media data. However, the vast amount of data generated during large-scale incidents can lead to issues of information overload and quality. To mitigate these issues, this chapter proposes the semi-automatic creation of alerts including keyword, relevance and information quality filters based on cross-platform social media data. We conducted empirical studies and workshops with emergency services across Europe to raise requirements, then iteratively designed and implemented an approach to support emergency services, and performed multiple evaluations, including live demonstrations and field trials, to research the potentials of social media-based alerts. Finally, we present the findings and implications based on semi-structured interviews with emergency services, highlighting the need for usable configurability and white-box algorithm representation.

The research field of crisis informatics examines, amongst others, the potentials and barriers of social media use during disasters and emergencies. Social media allow emergency services to receive valuable information (e.g., eyewitness reports, pictures, or videos) from social media. However, the vast amount of data generated during large-scale incidents can lead to issue of information overload. Research indicates that supervised machine learning techniques are suitable for identifying relevant messages and filter out irrelevant messages, thus mitigating information overload. However, they require a considerable amount of labelled data, clear criteria for relevance classification, a usable interface to facilitate the labelling process and a mechanism to rapidly deploy retrained classifiers. To overcome these issues, we present (1) a system for social media monitoring, analysis and relevance classification, (2) abstract and precise criteria for relevance classification in social media during disasters and emergencies, (3) the evaluation of a well-performing Random Forest algorithm for relevance classification incorporating metadata from social media into a batch learning approach (e.g., 91.28%/89.19% accuracy, 98.3%/89.6% precision and 80.4%/87.5% recall with a fast training time with feature subset selection on the European floods/BASF SE incident datasets), as well as (4) an approach and preliminary evaluation for relevance classification including active, incremental and online learning to reduce the amount of required labelled data and to correct misclassifications of the algorithm by feedback classification. Using the latter approach, we achieved a well-performing classifier based on the European floods dataset by only requiring a quarter of labelled data compared to the traditional batch learning approach. Despite a lesser effect on the BASF SE incident dataset, still a substantial improvement could be determined.

Emergencies threaten human lives and overall societal continuity, whether or not the crises and disasters are induced by nature, such as earthquakes, floods and hurricanes, or by human beings, such as accidents, terror attacks and uprisings. In such situations, not only do citizens demand information about the damage and safe behaviour, but emergency services also require high quality information to improve situational awareness. For this purpose, there are currently two kinds of apps available: General-purpose apps, such as Facebook Safety Check or Twitter Alerts, already integrate safety features. Specific crisis apps, such as Katwarn in Germany or FEMA in the US, provide information on how to behave before, during and after emergencies, and capabilities for reporting incidents or receiving disaster warnings. In this chapter, we analyse authorities’ and citizens’ information demands and features of crisis apps. Moreover, we present the concept, implementation and evaluation of a crisis app for incident reporting and bidirectional communication between authorities and citizens. Using the app, citizens may (1) report incidents by providing a category, description, location and multimedia files and (2) receive broadcasts and responses from authorities. Finally, we outline features, requirements and contextual factors for incident reporting and bidirectional communication via mobile app.

In part I of the thesis, information refinement was defined as the process of applied practices and used technologies to refine obtained information according to contextual, i.e. event-based, human, organizational, societal, and technological, boundary conditions in order to improve crisis management, response and, as a desired consequence, social wellbeing. Based on the findings of the parts II and III, this chapter proposes an information refinement framework for emergency services. In the following, it will discuss the event-based, organisational, societal, and technological perspective of the established information refinement framework.

Almost two decades of mobile technology and social media usage in crises led to the emergence and steadily foundation of crisis informatics, which is a research field combining computing and social science knowledge of disasters to improve ICT-supported crisis management and response. In order to utilise these sources of social big data for the situational awareness, crisis communication, and decision making of emergency services, available information must be refined according to event-based factors, organizational requirements, societal boundary conditions, and technical feasibility. This dissertation combined the methodological framework of design case studies with principles of design science research to research technologies for information refinement focusing mobile crisis apps and social media. These extended design case studies comprised the four phases of theoretical reviews, empirical pre-studies, design of supportive ICT, and evaluation of ICT appropriation. This conclusion will outline main findings of the dissertation by answering the initial research questions. Furthermore, it will delimit and discuss empirical and theoretical contributions followed by design and practical contributions using the HCI research contribution typology of Wobbrock and Kientz, which differentiates empirical, artefact, methodological, theoretical, dataset, survey, and opinion contributions. The conlusion finishes with the discussion of limitations and future work.