|
Salemi, H., Senarath, Y., & Purohit, H. (2023). A Comparative Study of Pre-trained Language Models to Filter Informative Code-mixed Data on Social Media during Disasters. In Jaziar Radianti, Ioannis Dokas, Nicolas Lalone, & Deepak Khazanchi (Eds.), Proceedings of the 20th International ISCRAM Conference (pp. 920–932). Omaha, USA: University of Nebraska at Omaha.
Abstract: Social media can inform response agencies during disasters to help affected people. However, filtering informative messages from social media content is challenging due to the ungrammatical text, out-of-vocabulary words, etc., that limit the context interpretation of messages. Further, there has been limited exploration of the challenge of code-mixing (using words from another language in a given text of one language) in user-generated content during disasters. Hence, we proposed a new code-mixed dataset of tweets related to the 2017 Iran-Iraq Earthquake and annotated them based on their informativeness characteristics. Additionally, we have evaluated the performance of state-of-the-art pre-trained language models: mBERT, RoBERTa, and XLM-R, on the proposed dataset. The results show that mBERT (with F1 score of 72%) overweighs the other models in classifying informative code-mixed messages. Moreover, we analyzed some patterns of exploiting code-mixing by users, which can help future works in developing these models.
|
|
|
Andrea Bellucci, Alessio Malizia, Paloma Díaz, & Ignacio Aedo. (2010). Framing the design space for novel crisis-related mashups: The eStoryS example. In C. Zobel B. T. S. French (Ed.), ISCRAM 2010 – 7th International Conference on Information Systems for Crisis Response and Management: Defining Crisis Management 3.0, Proceedings. Seattle, WA: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: Web 2.0 can be viewed as a platform where users can develop their own web applications. It is also characterized by a vast amount of user-generated contents presenting spatial and temporal components, by means of associated metadata. These metadata has been successfully exploited to generate map-based mashups (web applications gathering data from different sources) facing different kind of crisis situations, ranging from natural disasters (earthquakes, wildfires, floods...) to human-made disasters (terrorist attacks, school shootings, conflicts...). The social and collaborative dimensions of the Web 2.0 can be also exploited for managing crisis-related information. We present here a survey of current crisis-related mashups we employed to extract design dimensions and provide a conceptual framework that can be used: A) to understand current systems and; b) to design next generation of crisis-related mashups. We propose the eStoryS system as an example of application developed following the design principles presented in this paper. On the basis of our analysis, we believe that the design dimensions posited here provide useful insights for the design of novel web mashups in the emergency management domain.
|
|
|
Thomas Heverin, & Lisl Zach. (2010). Microblogging for crisis communication: Examination of twitter use in response to a 2009 violent crisis in the Seattle-Tacoma, Washington area. In C. Zobel B. T. S. French (Ed.), ISCRAM 2010 – 7th International Conference on Information Systems for Crisis Response and Management: Defining Crisis Management 3.0, Proceedings. Seattle, WA: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: This research-in-progress paper reports on the use of microblogging as a communication and information sharing resource during a recent violent crisis. The goal of the larger research effort is to investigate the role that microblogging plays in crisis communication during violent events. The shooting of four police officers and the subsequent 48-hour search for the suspect that took place in the Seattle-Tacoma area of Washington in late November 2009 is used as a case study. A stream of over 6,000 publically available messages on Twitter, a popular microblogging site, was collected and individual messages were categorized as information, opinion, technology, emotion, and action-related. The coding and statistical analyses of the messages suggest that citizens use microblogging as one method to organize and disseminate crisis-related information. Additional research is in progress to analyze the types of information transmitted, the sources of the information, and the temporal trends of information shared.
|
|
|
Amanda L. Hughes, & Leysia Palen. (2009). Twitter adoption and use in mass convergence and emergency events. In S. J. J. Landgren (Ed.), ISCRAM 2009 – 6th International Conference on Information Systems for Crisis Response and Management: Boundary Spanning Initiatives and New Perspectives. Gothenburg: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: This paper offers a descriptive account of Twitter (a micro-blogging service) across four high profile, mass convergence events-two emergency and two national security. We statistically examine how Twitter is being used surrounding these events, and compare and contrast how that behavior is different from more general Twitter use. Our findings suggest that Twitter messages sent during these types of events contain more displays of information broadcasting and brokerage, and we observe that general Twitter use seems to have evolved over time to offer more of an information-sharing purpose. We also provide preliminary evidence that Twitter users who join during and in apparent relation to a mass convergence or emergency event are more likely to become long-term adopters of the technology.
|
|
|
Sophia B. Liu. (2010). The rise of curated crisis content. In C. Zobel B. T. S. French (Ed.), ISCRAM 2010 – 7th International Conference on Information Systems for Crisis Response and Management: Defining Crisis Management 3.0, Proceedings. Seattle, WA: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: In a networked world, we are increasingly inundated with information from online data streams especially from the social web. Curation has increasingly become the buzzword for managing this problem of information overload in the digital age. However, the applications and interpretations of curation by social web users are varied and often stray away from traditional curator roles. I present seven curatorial activities (i.e. collecting, organizing, preserving, filtering, crafting a story, displaying, and facilitating discussions) based on the analysis of 100 web artifacts. I introduce the concept, socially-distributed curation, to emphasize the distributed nature of this curatorial process emerging from the social web. Lastly, I present seven case studies to illustrate preliminary examples of curated crisis content for four crises. These findings are to inform future designs and developments of crisis management tools that could benefit from curated crisis content.
|
|
|
Kate Starbird, & Leysia Palen. (2010). Pass it on?: Retweeting in mass emergency. In C. Zobel B. T. S. French (Ed.), ISCRAM 2010 – 7th International Conference on Information Systems for Crisis Response and Management: Defining Crisis Management 3.0, Proceedings. Seattle, WA: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: We examine microblogged information generated during two different co-occurring natural hazards events in Spring 2009. Due to its rapid and widespread adoption, microblogging in emergency response is a place for serious consideration and experimentation for future application. Because microblogging is comprised of a set of practices shaped by a number of forces, it is important to measure and describe the diffuse, multiparty information exchange behaviors to anticipate how emergency governance might best play a role. Here we direct consideration toward information propagation properties in the Twitterverse, describing features of information redistribution related to the retweet (RT ) convention. Our analysis shows that during an emergency, for tweets authored by local users and tweets that contain emergency-related search terms, retweets are more likely than non-retweets to be about the event. We note that users are more likely to retweet information originally distributed through Twitter accounts run by media, especially the local media, and traditional service organizations. Comparing local users to the broader audience, we also find that tweet-based information redistribution is different for those who are local to an emergency event.
|
|
|
Kate Starbird, & Jeannie Stamberger. (2010). Tweak the tweet: Leveraging microblogging proliferation with a prescriptive syntax to support citizen reporting. In C. Zobel B. T. S. French (Ed.), ISCRAM 2010 – 7th International Conference on Information Systems for Crisis Response and Management: Defining Crisis Management 3.0, Proceedings. Seattle, WA: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: In this paper, we propose a low-tech solution for use by microbloggers that could enhance their ability to rapidly produce parsable, crisis-relevant information in mass emergencies. We build upon existing research on the use of social media during mass emergencies and disasters. Our proposed intervention aims to leverage the affordances of mobile microblogging and the drive to support citizen reporting within current behavioral Twitter-based microblogging practice. We introduce a prescriptive, tweet-based syntax that could increase the utility of information generated during emergencies by gently reshaping current behavioral practice. This offering is grounded in an understanding of current trends in norm evolution of Twitter use, an evolution that has progressed quickly but appears to be stabilizing around specific textual conventions.
|
|
|
Grace, R., Montarnal, A., Petitdemange, E., Rutter, J., Rodriguez, G. R., & Potts, M. (2023). Collaborative Information Seeking during a 911 Call Surge: A Case Study. In Jaziar Radianti, Ioannis Dokas, Nicolas Lalone, & Deepak Khazanchi (Eds.), Proceedings of the 20th International ISCRAM Conference (pp. 649–662). Omaha, USA: University of Nebraska at Omaha.
Abstract: This case study examines collaborative information seeking in a public-safety answering point during a 911 call surge that occurred when a man fired an assault rifle at police officers and evaded capture for nearly an hour in March 2020. Overwhelmed by questionable and imprecise reports from 911 callers, telecommunicators and on scene responders began working together to conduct broad and deep searches for the shooter. Whereas broad searches improved the scope of information gathering by identifying multiple, albeit questionable and imprecise, reports of the suspect’s location, deep searches improved the quality of information gathering by investigating 911 callers’ reports using drone, helicopter, and patrol units. These findings suggest requirements for collaborative information seeking in public-safety answering points, including capabilities to conduct broad and deep searches using next-generation 911 technologies, and command and control requirements for triaging these search tasks within inter-organizational emergency response systems.
|
|
|
Robert Soden, Leysia Palen, Claire Chase, Derya Deniz, Erin Arneson, Leah Sprain, et al. (2015). The Polyvocality of Resilience: Discovering a Research Agenda through Interdisciplinary Investigation & Community Engagement. In L. Palen, M. Buscher, T. Comes, & A. Hughes (Eds.), ISCRAM 2015 Conference Proceedings ? 12th International Conference on Information Systems for Crisis Response and Management. Kristiansand, Norway: University of Agder (UiA).
Abstract: This paper presents findings from an interdisciplinary research effort studying community resilience in Boulder, Colorado. Boulder is a progressive region with a history of environmental leadership. The area is currently in the process of recovering from major flooding and has launched several new initiatives related to building long-term resilience to natural disasters and other stressors. In our research, we consider the stakeholders involved in building local resilience as well as the different and often contradictory framings of the concept. This study takes a phenomenological and inductive approach to understanding resilience. In contrast to more reductionist frameworks that are frequently offered, we argue that this allows for greater understanding of the polyvocal and emergent qualities of resilience.
|
|
|
Joanne I. White, & Leysia Palen. (2015). Participatory Mapping for Disaster Preparedness: The Development & Standardization of Animal Evacuation Maps. In L. Palen, M. Buscher, T. Comes, & A. Hughes (Eds.), ISCRAM 2015 Conference Proceedings ? 12th International Conference on Information Systems for Crisis Response and Management. Kristiansand, Norway: University of Agder (UiA).
Abstract: People who own animals are faced with complex decision making in evacuations. In the US, the Emergency Operations Center is often inundated with calls from animal owners who are aware they are under pre- or mandatory evacuation, but are unsure of what to do about evacuating their animals. Often animal evacuation is a highly improvised activity for owners and responders, though there is a now a general push toward streamlining procedures because of the high impact the matter of animals has on society?s welfare during times of emergency. This paper reports on the use of participatory design methods in a mapping project to support the range of people involved in animal evacuation during mass displacement events. The work provides insight into both procedures and standards for creating evacuation maps that communicate clearly with the public and across the range of emergency responders.
|
|
|
Daniel Auferbauer, Roman Ganhör, & Hilda Tellioglu. (2015). Moving Towards Crowd Tasking for Disaster Mitigation. In L. Palen, M. Buscher, T. Comes, & A. Hughes (Eds.), ISCRAM 2015 Conference Proceedings ? 12th International Conference on Information Systems for Crisis Response and Management. Kristiansand, Norway: University of Agder (UiA).
Abstract: Advancements in information and communication technology (ICT) offer new possibilities when dealing with crisis situations. In this paper we present the design for a crowd tasking tool (CTT) that is currently under development. We describe how the tool can assist disaster relief coordinators during a crisis by selectively distributing tasks to a crowd of volunteers. We also compare the CTT with an already existing ICT based solution for supporting volunteerism during crisis. The differences between these two tools are addressed and the implications for volunteerism are discussed. The paper concludes with an outlook on future work emphasizing a form of volunteer involvement that offers potential for gathering information that is more relevant and easier to digest for decision-making than information provided solely by self-organised volunteers through social media.
|
|
|
Daniel Auferbauer, Roman Ganhör, Hilda Tellioglu, & Jasmin Pielorz. (2016). Crowdtasking: Field Study on a Crowdsourcing Alternative. In A. Tapia, P. Antunes, V.A. Bañuls, K. Moore, & J. Porto (Eds.), ISCRAM 2016 Conference Proceedings ? 13th International Conference on Information Systems for Crisis Response and Management. Rio de Janeiro, Brasil: Federal University of Rio de Janeiro.
Abstract: In this paper we elaborate on the concept of crowdtasking as a form of crowdsourcing. The paper describes the setup and boundaries of a first controlled live field test of a prototypical implementation of a possible crowdtasking workflow. The implemented workflow allows crisis managers rapid intelligence gathering due to direct and tailored task distribution. Practitioners of Crisis and Disaster Management and volunteer managers who were present during the field test made favourable comments on the approach and its implementation. The analysis of the records and the conducted interviews give new insights and ideas for further development.
|
|
|
Wang, D. (2023). Public Cognition and Perception on Social Media in Crisis. In Jaziar Radianti, Ioannis Dokas, Nicolas Lalone, & Deepak Khazanchi (Eds.), Proceedings of the 20th International ISCRAM Conference (pp. 1081–1082). Omaha, USA: University of Nebraska at Omaha.
Abstract: Microblogging platforms have been increasingly used in crisis, facilitating more participatory communication between official response channels and affected communities. Despite the potential benefits, research has shown that disaster response organizations could not effectively utilize social media data due to data deluge (Knox 2022). To better understand the information needed for disaster response, we turn to the National Incident Management System Guidance for public information officers (PIOs) (NIMS Basic Guidance for PIOs 2020), the primary spokesperson for emergency management organizations. The guidance indicates that PIOs use social media for two major purposes, supporting their operational needs and gauging public perception of risk and incident response. To support the operational needs, the crisis informatics literature has heavily focused on information types supporting situational awareness, including serviceable, eyewitness or actionable information. However, the information representing public perception, such as people’s cognitive and perceptual processes in response to incidents, has been less addressed at scale. To bridge the gap between quantitative study in crisis informatics and information representing cognitive and perceptual processes and better support the task of PIOs, I focus on the study of people’s cognitive and perceptual processes on social media for my research. Cognitive and perceptual processes refer to the way that people pay attention to or process environmental inputs, including the mental activities of acquisition, processing or evaluation of environmental cues, social cues, and warnings. These processes reveal people’s perception of- and decision-making in response to potential threats. With this focus, I seek to answer the following research question: How could people’s cognitive and perceptual processes be inferred from their social media activities in crisis to benefit stakeholders in incident response? My interest in tracing this overall theme through a varied range of sub-tasks produces three more specific research questions: RQ1. How can information exposure and attention be operationalized to highlight cognitive and perceptual processes? RQ2. How do people’s perception of risk communications from stakeholders vary in crisis? RQ3. How could a principled and scalable pipeline be designed to identify people’s cognitive and affective perceptions on Twitter? I took cues from the Protective Action Decision Model (Lindell and Perry 2012) and leveraged baselines in the literature to address these research questions. To address the first research question, I proposed a metric that conceptualized and operationalized the predecision process. The proposed metric was incorporated into a pipeline and applied to two real-word events to recommend messages that represent the shift of collective attention of those locally affected with a specialized focus on cognitive and perceptual processes. To address the second research question, I went beyond the perception of risks to include perceptions of risk communications by stakeholders. I performed an empirical study of the relation between risk communications by stakeholders and different kinds of public perceptions (Lindell and Perry 2012). To address the third research question, I proposed a future work to provide benchmark coding schemes, datasets and models to quantitatively identify information representing cognitive and perceptual processes. I will leverage existing benchmark datasets in the literature (Olteanu et al. 2014; Imran et al. 2016; Alam et al. 2018; Zahra et al. 2020; Rudra et al. 2017; Mazloom et al. 2018; Purohit et al. 2018) and coding schemes in qualitative studies (Trumbo et al. 2016; Demuth et al. 2018) and create benchmark classification models.
|
|
|
Jeannette N. Sutton, Leysia Palen, & Irina Shklovski. (2008). Backchannels on the front lines: Emergent uses of social media in the 2007 Southern California Wildfires. In B. V. de W. F. Fiedrich (Ed.), Proceedings of ISCRAM 2008 – 5th International Conference on Information Systems for Crisis Response and Management (pp. 624–631). Washington, DC: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: Opportunities for participation by members of the public are expanding the information arena of disaster. Social media supports “backchannel” communications, allowing for wide-scale interaction that can be collectively resourceful, self-policing, and generative of information that is otherwise hard to obtain. Results from our study of information practices by members of the public during the October 2007 Southern California Wildfires suggest that community information resources and other backchannel communications activity enabled by social media are gaining prominence in the disaster arena, despite concern by officials about the legitimacy of information shared through such means. We argue that these emergent uses of social media are pre-cursors of broader future changes to the institutional and organizational arrangements of disaster response.
|
|
|
Sophia B. Liu, & Leysia Palen. (2009). Spatiotemporal mashups: A survey of current tools to inform next generation crisis support. In S. J. J. Landgren (Ed.), ISCRAM 2009 – 6th International Conference on Information Systems for Crisis Response and Management: Boundary Spanning Initiatives and New Perspectives. Gothenburg: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: Developments in information and communication technology (ICT) have adjusted the opportunities for spatial and temporal representations of data, possibly permitting the simultaneous visualization of how different regions and populations are affected during large-scale emergencies and crises. We surveyed 13 crisis-related mashups to derive some high-level design directions to guide the design and testing of next generation crisis support tools. The current web mashups offer a new way of looking at how crises are spatiotemporally ordered. However, since all technology is constrained by limitations of design choice, examining the limits and possibilities of what current design choices afford can inform attributes of what next generation crisis support tools would require.
|
|
|
Robert Soden, Nama Budhathoki, & Leysia Palen. (2014). Resilience-building and the crisis informatics agenda: Lessons learned from open cities Kathmandu. In and P.C. Shih. L. Plotnick M. S. P. S.R. Hiltz (Ed.), ISCRAM 2014 Conference Proceedings – 11th International Conference on Information Systems for Crisis Response and Management (pp. 339–348). University Park, PA: The Pennsylvania State University.
Abstract: Information systems that support crisis responders and disaster risk management efforts are complex sociotechnical phenomena comprised of human capacities and relationships, data and software tools. Research in crisis informatics has highlighted the ways in which emergent groups of digital volunteers, or volunteer technical communities, have mobilized during disaster events to support information management efforts. This paper describes an action research project to support the creation of an ex ante volunteer technical community from among the potentially affected population in Kathmandu, Nepal, one of the most seismically at-risk cities in the world. In exploring this case, we argue that projects that attempt to create local open data ecosystems can be valuable but require investment in their design, execution and on-going maintenance.
|
|
|
Amanda Hughes, Keri Stephens, Steve Peterson, Hemant Purohit, Anastazja G. Harris, Yasas Senarath, et al. (2022). Human-AI Teaming for COVID-19 Response: A Practice & Research Collaboration Case Study. In Rob Grace, & Hossein Baharmand (Eds.), ISCRAM 2022 Conference Proceedings – 19th International Conference on Information Systems for Crisis Response and Management (pp. 1048–1057). Tarbes, France.
Abstract: Practice and research collaborations in the disaster domain have the potential to improve emergency management practices while also advancing disaster science theory. However, they also pose challenges as practitioners and researchers each have their own culture, history, values, incentives, and processes that do not always facilitate collaboration. In this paper, we reflect on a 6-month practice and research collaboration, where researchers and practitioners worked together to craft a social media monitoring system for emergency managers in response to the COVID-19 pandemic. The challenges we encountered in this project fall into two broad categories, job-related and timescale challenges. Using prior research on team science as a guide, we discuss several challenges we encountered in these two categories and show how our team sought to overcome them. We conclude with a set of best practices for improving practice and research collaborations.
|
|
|
Amanda L. Hughes, Leysia Palen, Jeannette N. Sutton, Sophia B. Liu, & Sara Vieweg. (2008). Site-seeing in disaster: An examination of on-line social convergence. In B. V. de W. F. Fiedrich (Ed.), Proceedings of ISCRAM 2008 – 5th International Conference on Information Systems for Crisis Response and Management (pp. 324–333). Washington, DC: Information Systems for Crisis Response and Management, ISCRAM.
Abstract: On-line websites and applications are increasingly playing a role in disaster response and recovery. Yet with the wide variety of on-line grassroots activities that occur in such situations, it can be difficult to make sense of them. In this paper, we describe on-line behavior as socially convergent activity, interpreting it within existing sociological understandings of behavior in disaster events. We discuss seven types of convergent behavior and give examples of on-line activities for each type. By seeing these activities as an essential part of the disaster social arena, we can begin to think about how to support socially convergent phenomena in new and creative ways.
|
|
|
Iva Seto, David Johnstone, & Jennifer Campbell-Meier. (2018). Experts' sensemaking during the 2003 SARS crisis. In Kristin Stock, & Deborah Bunker (Eds.), Proceedings of ISCRAM Asia Pacific 2018: Innovating for Resilience – 1st International Conference on Information Systems for Crisis Response and Management Asia Pacific. (pp. 44–55). Albany, Auckland, New Zealand: Massey Univeristy.
Abstract: This paper depicts the real-time sensemaking of experts as they worked to combat the first emerging disease of the 21st century: Severe Acute Respiratory Syndrome (SARS). Newspaper data was analysed from the 2003 SARS crisis, with a Canadian perspective, to follow the process of solving the puzzle of this emerging disease. Retrospective sensemaking is a process that is triggered by the unexpected, which leads to actors gathering information (taking action) in order to consider possible interpretations for the unexpected event. Disease outbreaks serve as sensemaking triggers, and actors engage in retrospective sensemaking to find out the factors involved in how the outbreak happened. Prospective sensemaking (future-oriented) is employed when actors work together to plan how to combat the disease. The newspaper data demonstrate that retrospective and prospective sensemaking are tethered: to make plans to combat a disease, actors first require a collectively agreed upon understanding from which they can generate possibilities for a crisis response. This paper contributes to the field by providing concepts for long-duration crisis sensemaking, as the bulk of organisational research focuses on acute crises such as wildfires, or earthquakes.
|
|
|
Soudip Roy Chowdhury, Muhammad Imran, Muhammad Rizwan Asghar, Amer-Yahia, S., & Carlos Castillo. (2013). Tweet4act: Using incident-specific profiles for classifying crisis-related messages. In J. Geldermann and T. Müller S. Fortier F. F. T. Comes (Ed.), ISCRAM 2013 Conference Proceedings – 10th International Conference on Information Systems for Crisis Response and Management (pp. 834–839). KIT; Baden-Baden: Karlsruher Institut fur Technologie.
Abstract: We present Tweet4act, a system to detect and classify crisis-related messages communicated over a microblogging platform. Our system relies on extracting content features from each message. These features and the use of an incident-specific dictionary allow us to determine the period type of an incident that each message belongs to. The period types are: Pre-incident (messages talking about prevention, mitigation, and preparedness), during-incident (messages sent while the incident is taking place), and post-incident (messages related to the response, recovery, and reconstruction). We show that our detection method can effectively identify incident-related messages with high precision and recall, and that our incident-period classification method outperforms standard machine learning classification methods.
|
|
|
Lise Ann St. Denis, Amanda L. Hughes, & Leysia Palen. (2012). Trial by fire: The deployment of trusted digital volunteers in the 2011 shadow lake fire. In Z.Franco J. R. L. Rothkrantz (Ed.), ISCRAM 2012 Conference Proceedings – 9th International Conference on Information Systems for Crisis Response and Management. Vancouver, BC: Simon Fraser University.
Abstract: We report on the use of a team of trusted digital volunteers during the 2011 Shadow Lake Fire that occurred in the US Pacific Northwest to extend the social media capacity of a Type I incident management team. In this case study, we outline the tools and processes used by this virtual team to coordinate their activities, monitor social media communication and to establish communications with the public around the event. Finally, we discuss the potential merits and limitations of implementing a team of trusted volunteers and explore how this idea could be incorporated into emergency management organizations. © 2012 ISCRAM.
|
|
|
Kate Starbird, Grace Muzny, & Leysia Palen. (2012). Learning from the crowd: Collaborative filtering techniques for identifying on-the-ground Twitterers during mass disruptions. In Z.Franco J. R. L. Rothkrantz (Ed.), ISCRAM 2012 Conference Proceedings – 9th International Conference on Information Systems for Crisis Response and Management. Vancouver, BC: Simon Fraser University.
Abstract: Social media tools, including the microblogging platform Twitter, have been appropriated during mass disruption events by those affected as well as the digitally-convergent crowd. Though tweets sent by those local to an event could be a resource both for responders and those affected, most Twitter activity during mass disruption events is generated by the remote crowd. Tweets from the remote crowd can be seen as noise that must be filtered, but another perspective considers crowd activity as a filtering and recommendation mechanism. This paper tests the hypothesis that crowd behavior can serve as a collaborative filter for identifying people tweeting from the ground during a mass disruption event. We test two models for classifying on-the-ground Twitterers, finding that machine learning techniques using a Support Vector Machine with asymmetric soft margins can be effective in identifying those likely to be on the ground during a mass disruption event. © 2012 ISCRAM.
|
|
|
Cody Buntain, Richard Mccreadie, & Ian Soboroff. (2022). Incident Streams 2021 Off the Deep End: Deeper Annotations and Evaluations in Twitter. In Rob Grace, & Hossein Baharmand (Eds.), ISCRAM 2022 Conference Proceedings – 19th International Conference on Information Systems for Crisis Response and Management (pp. 584–604). Tarbes, France.
Abstract: This paper summarizes the final year of the four-year Text REtrieval Conference Incident Streams track (TREC-IS), which has produced a large dataset comprising 136,263 annotated tweets, spanning 98 crisis events. Goals of this final year were twofold: 1) to add new categories for assessing messages, with a focus on characterizing the audience, author, and images associated with these messages, and 2) to enlarge the TREC-IS dataset with new events, with an emphasis of deeper pools for sampling. Beyond these two goals, TREC-IS has nearly doubled the number of annotated messages per event for the 26 crises introduced in 2021 and has released a new parallel dataset of 312,546 images associated with crisis content – with 7,297 tweets having annotations about their embedded images. Our analyses of this new crisis data yields new insights about the context of a tweet; e.g., messages intended for a local audience and those that contain images of weather forecasts and infographics have higher than average assessments of priority but are relatively rare. Tweets containing images, however, have higher perceived priorities than tweets without images. Moving to deeper pools, while tending to lower classification performance, also does not generally impact performance rankings or alter distributions of information-types. We end this paper with a discussion of these datasets, analyses, their implications, and how they contribute both new data and insights to the broader crisis informatics community.
|
|
|
Jens Kersten, Jan Bongard, & Friederike Klan. (2022). Gaussian Processes for One-class and Binary Classification of Crisis-related Tweets. In Rob Grace, & Hossein Baharmand (Eds.), ISCRAM 2022 Conference Proceedings – 19th International Conference on Information Systems for Crisis Response and Management (pp. 664–673). Tarbes, France.
Abstract: Overload reduction is essential to exploit Twitter text data for crisis management. Often used pre-trained machine learning models require training data for both, crisis-related and off-topic content. However, this task can also be formulated as a one-class classification problem in which labeled off-topic samples are not required. Gaussian processes (GPs) have great potential in both, binary and one-class settings and are therefore investigated in this work. Deep kernel learning combines the representative power of text embeddings with the Bayesian formalism of GPs. Motivated by this, we investigate the potential of deep kernel models for the task of classifying crisis-related tweet texts with special emphasis on cross-event applications. Compared to standard binary neural networks, first experiments with one-class GP models reveal a great potential for realistic scenarios, offering a fast and flexible approach for interactive model training without requiring off-topic training samples and comprehensive expert knowledge (only two model parameters involved).
|
|
|
Shivam Sharma, & Cody Buntain. (2022). Bang for your Buck: Performance Impact Across Choices in Learning Architectures for Crisis Informatics. In Rob Grace, & Hossein Baharmand (Eds.), ISCRAM 2022 Conference Proceedings – 19th International Conference on Information Systems for Crisis Response and Management (pp. 719–736). Tarbes, France.
Abstract: Over the years, with the increase in social media engagement, there has been an in increase in various pipelines to analyze, classify and prioritize crisis-related data on various social media platforms. These pipelines utilize various data augmentation methods to counter imbalanced crisis data, sophisticated and off-the-shelf models for training. However, there is a lack of comprehensive study which compares these methods for the various sections of a pipeline. In this study, we split a general crisis-related pipeline into 3 major sections, namely, data augmentation, model selection, and training methodology. We compare various methods for each of these sections and then present a comprehensive evaluation of which section to prioritize based on the results from various pipelines. We compare our results against two separate tasks, information classification and priority scoring for crisis-related tweets. Our results suggest that data augmentation, in general,improves the performance. However, sophisticated, state-of-the-art language models like DeBERTa only show performance gain in information classification tasks, and models like RoBERTa tend to show a consistent performance increase over our presented baseline consisting of BERT. We also show that, though training two separate task-specific BERT models does show better performance than one BERT model with multi-task learning methodology over an imbalanced dataset, multi-task learning does improve performance for more sophisticated model like DeBERTa with a much more balanced dataset after augmentation.
|
|