|
Amanda Langer, Marc-André Kaufhold, Elena Maria Runft, Christian Reuter, Margarita Grinko, & Volkmar Pipek. (2019). Counter Narratives in Social Media: An Empirical Study on Combat and Prevention of Terrorism. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: With the increase of terrorist attacks and spreading extremism worldwide, countermeasures advance as well. Often
social media is used for recruitment and radicalization of susceptible target groups. Counter narratives are trying
to disclose the illusion created by radical and extremist groups through a purposive and educational counter
statement, and to initiate a rethinking in the affected individuals via thought-provoking impulses and advice. This
exploratory study investigates counter narrative campaigns with regard to their fight and prevention against
terrorism in social media. Posts with strong emotions and a personal reference to affected individuals achieved
the highest impact and most reactions from the target group. Furthermore, our results illustrate that the impact of
a counter narrative campaign cannot be measured solely according to the reaction rate to their postings and that
further analysis steps are therefore necessary for the final evaluation of the campaigns.
|
|
|
Anna Kruspe, Jens Kersten, & Friederike Klan. (2019). Detecting event-related tweets by example using few-shot models. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Social media sources can be helpful in crisis situations, but discovering relevant messages is not trivial. Methods
have so far focused on universal detection models for all kinds of crises or for certain crisis types (e.g. floods).
Event-specific models could implement a more focused search area, but collecting data and training new models for
a crisis that is already in progress is costly and may take too much time for a prompt response. As a compromise,
manually collecting a small amount of example messages is feasible. Few-shot models can generalize to unseen
classes with such a small handful of examples, and do not need be trained anew for each event. We show how
these models can be used to detect crisis-relevant tweets during new events with just 10 to 100 examples and
counterexamples. We also propose a new type of few-shot model that does not require counterexamples.
|
|
|
Asmelash Teka Hadgu, Sallam Abualhaija, & Claudia Niederée. (2019). Real-time Adaptive Crawler for Tracking Unfolding Events on Twitter. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: When a major event such as a crisis situation occurs, people post messages on social media sites such as Twitter, in
order to exchange information or to share emotions. These posts can provide useful information to raise situation
awareness and support decision making, e.g., by aid organizations. In this paper, we propose a novel method for
social media crawling, which exploits a Bayesian inference framework to keep track of keyword changes over time
and uses a counter-stream to gauge the inclusion of noise and irrelevant information. In addition, we present a
framework to evaluate real-time adaptive social search algorithms in a reproducible manner, which relies on a
semi-automated approach for ground-truth construction. We show that our method outperforms previous methods
for very large scale events.
|
|
|
Fedor Vitiugin, & Carlos Castillo. (2019). Comparison of Social Media in English and Russian During Emergencies and Mass Convergence Events. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Twitter is used for spreading information during crisis events. In this paper, we first retrieve event-related information
posted in English and Russian during six disasters and sports events that received wide media coverage in both
languages, using an adaptive information filtering method for automating the collection of about 100 000 messages.
We then compare the contents of these messages in terms of 17 informational and linguistic features using a
difference in differences approach. Our results suggest that posts in each language are focused on different types
of information. For instance, almost 50% of the popular people mentioned in these messages appear exclusively
in either the English messages or the Russian messages, but not both. Our results also suggest differences in the
adoption of platform mechanics during crises between Russian-speaking and English-speaking users. This has
important implications for data collection during crises, which is almost always focused on a single language.
|
|
|
Firoj Alam, Ferda Ofli, & Muhammad Imran. (2019). CrisisDPS: Crisis Data Processing Services. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Over the last few years, extensive research has been conducted to develop technologies to support humanitarian aid
tasks. However, many technologies are still limited as they require both manual and automatic approaches, and
more importantly, are not ready to be integrated into the disaster response workflows. To tackle this limitation, we
develop automatic data processing services that are freely and publicly available, and made to be simple, efficient,
and accessible to non-experts. Our services take textual messages (e.g., tweets, Facebook posts, SMS) as input to
determine (i) which disaster type the message belongs to, (ii) whether it is informative or not, and (iii) what type of
humanitarian information it conveys. We built our services upon machine learning classifiers that are obtained from
large-scale comparative experiments utilizing both classical and deep learning algorithms. Our services outperform
state-of-the-art publicly available tools in terms of classification accuracy.
|
|
|
Gabriela C Barrera, & Maria C Yang. (2019). Evaluation of Digital Volunteers using a Design Approach: Motivations and Contributions in Disaster Response. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: With the growth of social media and crowdsourcing in disaster response, further research is needed on the motivations
and contributions of digital volunteers. This study applies a user-centered design approach to understanding how we
might make better tools to support digital volunteers. This user-centered design approach involves stated preference
elicitation methods through an online survey to understand what digital volunteers want in such tools. Through
choice-based conjoint analysis, we contribute to mixed-methods research to gain additional insight into motivations
and user preferences for a set of design features that might be incorporated into an online tool specifically for digital
volunteers. Initial results show preferences for measures of success that were not monetary, which aligned with
directly stated motivations for volunteering. Our findings corroborate with previous research in that feedback to
volunteers is very important, as well as being able to measure the impact of their work.
|
|
|
Guoqin Ma, & Chittayong Surakitbanharn. (2019). Predicting Hurricane Damage Using Social Media Posts Coupled with Physical and Socio-Economic Variables. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: During a natural disaster or emergency event, individual social media posts or hot spots may not necessarily correlate
to the most devastated areas. To better understand the correlation between social media and physical damage, we
compare Tweets, data about the physical environment, and socio-economic factors with insurance claim information
(as a proxy for physical damage) from 2017 Hurricane Irma in the state of Florida. We use machine learning
to identify relevant Tweets, sensitivity analyses to identify socio-economic factors, and statistical regression to
determine the predictive capability of insurance claims as a proxy for damage. We find that Tweets alone result in a
poorly fitted regression model of insurance claims, but the inclusion of physical features (e.g., power outages, wind
level) and socio-economic factors (e.g., population density, education, Internet access) improves the model?s fit.
Such models contribute to the knowledge base that may allow social media to predict damage in real-time.
|
|
|
Humaira Waqas, & Muhammad Imran. (2019). #CampFireMissing: An Analysis of Tweets About Missing and Found People From California Wildfires. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Several research studies have shown the importance of social media data for humanitarian aid. Among others,
the issue of missing and lost people during disasters and emergencies is crucial for disaster managers. This work
analyzes Twitter data from a recent wildfire event to determine its usefulness for the mitigation of the missing and
found people issue. Data analysis performed using various filtering techniques, and trend analysis revealed that
Twitter contains important information potentially useful for emergency managers and volunteers to tackle this
issue. Many tweets were found containing full names, partial names, location information, and other vital clues
which could be useful for finding missing people.
|
|
|
Jens Kersten, Anna Kruspe, Matti Wiegmann, & Friederike Klan. (2019). Robust filtering of crisis-related tweets. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Social media enables fast information exchange and status reporting during crises. Filtering is usually required to
identify the small fraction of social media stream data related to events. Since deep learning has recently shown to
be a reliable approach for filtering and analyzing Twitter messages, a Convolutional Neural Network is examined for
filtering crisis-related tweets in this work. The goal is to understand how to obtain accurate and robust filtering
models and how model accuracies tend to behave in case of new events. In contrast to other works, the application
to real data streams is also investigated. Motivated by the observation that machine learning model accuracies
highly depend on the used data, a new comprehensive and balanced compilation of existing data sets is proposed.
Experimental results with this data set provide valuable insights. Preliminary results from filtering a data stream
recorded during hurricane Florence in September 2018 confirm our results.
|
|
|
Jess Kropczynski, Rob Grace, Shane Halse, Doina Caragea, Cornelia Caragea, & Andrea Tapia. (2019). Refining a Coding Scheme to Identify Actionable Information on Social Media. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: This paper describes the use of a previously established qualitative coding scheme developed through a design workshop with public safety professionals, and applied the schema to social media data collecting during crises. The intention of applying this scheme to existing crisis datasets was to acquire training data for machine learning. Applying the coding scheme to social media data revealed that additional subcategories of the coding scheme are necessary to satisfy information requirements necessary to dispatch first responders to an incident. The coding scheme was refined and adapted into a set of instructions for qualitative coders on Amazon Mechanical Turk. The contribution of this work is a coding scheme that is more directly related to the information needs of public safety professionals. Implications of early results using the refined coding scheme are discussed in terms of proposed automated methods to identify actionable information for dispatch of first responders during emergency incidents.
|
|
|
Liuqing Li, & Edward A. Fox. (2019). Understanding patterns and mood changes through tweets about disasters. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: We analyzed a sample of large tweet collections gathered since 2011, to expand understanding about tweeting
patterns and emotional responses of different types of tweeters regarding disasters. We selected three examples for
each of four disaster types: school shooting, bombing, earthquake, and hurricane. For each collection, we deployed
our novel model TwiRole for user classification, and an existing deep learning model for mood detection. We
found differences in the daily tweet count patterns, between the different types of events. Likewise, there were
different average scores and patterns of moods (fear, sadness, surprise), both between types of events, and between
events of the same type. Further, regarding surprise and fear, there were differences among roles of tweeters. These
results suggest the value of further exploration as well as hypothesis testing with our hundreds of event and trend
related tweet collections, considering indications in those that reflect emotional responses to disasters.
|
|
|
Marion Lara Tan, Raj Prasanna, Kristin Stock, Emma Hudson-Doyle, Graham Leonard, & David Johnston. (2019). Enhancing the usability of a disaster app: exploring the perspective of the public as users. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Limited research has studied how citizens? perspectives as end-users can contribute to improving the usability of disaster apps. This study addresses this gap by exploring end-user insights with the use of a conceptual disaster app in the New Zealand (NZ) context. NZ has multiple public alerting authorities that have various technological options in delivering information to the population?s mobile devices; including social media platforms, apps, as well as the Emergency Mobile Alert system. However, during critical events, the multiplicity of information may become overwhelming. A disaster app, conceptualised in the NZ context, aims to aggregate, organise, and deliver information from official sources to the public. After the initial conceptual design, a usability inquiry was administered by interviewing members of the public. Partial results of the inquiry show that the public?s perspective has value; in the process of understanding the new user?s viewpoint, usability highlights and issues are identified.
|
|
|
Paige Maas, Shankar Iyer, Andreas Gros, Wonhee Park, Laura McGorman, Chaya Nayak, et al. (2019). Facebook Disaster Maps: Aggregate Insights for Crisis Response & Recovery. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: After a natural disaster or other crisis, humanitarian organizations need to know where affected people are located
and what resources they need. While this information is difficult to capture quickly through conventional methods,
aggregate usage patterns of social media apps like Facebook can help fill these information gaps.
In this paper, we describe the data and methodology that power Facebook Disaster Maps. These maps utilize
information about Facebook usage in areas impacted by natural hazards, producing aggregate pictures of how the
population is affected by and responding to the hazard. The maps include insights into evacuations, cell network
connectivity, access to electricity, and long-term displacement.
In addition to descriptions and examples of each map type, we describe the source data used to generate the maps,
and efforts taken to ensure the security and privacy of Facebook users. We also describe limitations of the current
methodologies and opportunities for improvement.
|
|
|
Richard McCreadie, Cody Buntain, & Ian Soboroff. (2019). TREC Incident Streams: Finding Actionable Information on Social Media. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: The Text Retrieval Conference (TREC) Incident Streams track is a new initiative that aims to mature social
media-based emergency response technology. This initiative advances the state of the art in this area through an
evaluation challenge, which attracts researchers and developers from across the globe. The 2018 edition of the track
provides a standardized evaluation methodology, an ontology of emergency-relevant social media information types,
proposes a scale for information criticality, and releases a dataset containing fifteen test events and approximately
20,000 labeled tweets. Analysis of this dataset reveals a significant amount of actionable information on social
media during emergencies (> 10%). While this data is valuable for emergency response efforts, analysis of the
39 state-of-the-art systems demonstrate a performance gap in identifying this data. We therefore find the current
state-of-the-art is insufficient for emergency responders? requirements, particularly for rare actionable information
for which there is little prior training data available.
|
|
|
Rob Grace, Shane Halse, Jess Kropczynski, Andrea Tapia, & Fred Fonseca. (2019). Integrating Social Media in Emergency Dispatch via Distributed Sensemaking. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Emergency dispatchers typically answer 911 calls and relay information to first responders; however, new workflows arise when social media analysts are included in emergency dispatch work. In this study we examine emergency dispatch workflows as distributed sensemaking processes performed among 911 call takers, dispatchers, and social media analysts during simulated emergency dispatch operations. In active shooter and water rescue scenarios, emergency dispatch teams including call takers, dispatchers, and social media analysts make sense of unfolding events by analyzing, aggregating, and synthesizing information provided by 911 callers and social media users during each scenario. Findings from the simulations inform design requirements for social media analysis tools that can help analysts detect, seek, and analyze information posted on social media during a crisis, and protocols for coordinating analysts? sensemaking activities with those of 911 call takers and dispatchers in reconfigured emergency dispatch workflows.
|
|
|
Sara Barozzi, Jose Luis Fernandez Marquez, Amudha Ravi Shankar, & Barbara Pernici. (2019). Filtering images extracted from social media in the response phase of emergency events. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: The use of social media to support emergency operators in the first hours of the response phases can improve the
quality of the information available and awareness on ongoing emergency events. Social media contain both textual
and visual information, in the form of pictures and videos. The problem related to the use of social media posts
as a source of information during emergencies lies in the difficulty of selecting the relevant information among
a very large amount of irrelevant information. In particular, we focus on the extraction of images relevant to an
event for rapid mapping purpose. In this paper, a set of possible filters is proposed and analyzed with the goal of
selecting useful images from posts and of evaluating how precision and recall are impacted. Filtering techniques,
which include both automated and crowdsourced steps, have the goal of providing better quality posts and easy
manageable data volumes both to emergency responders and rapid mapping operators. The impact of the filters on
precision and recall in extracting relevant images is discussed in the paper in two different case studies.
|
|
|
Shane Errol Halse, Rob Grace, Jess Kropczynski, & Andrea Tapia. (2019). Simulating real-time Twitter data from historical datasets. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: In this paper, we will discuss a system design for simulating social media data based on historical datasets. While many datasets containing data collected from social media during crisis have become publicly available, there is a lack of tools or systems can present this data on the same timeline as it was originally posted. Through the design and use of the tool discussed in this paper, we show how historical datasets can be used for algorithm testing, such as those used in machine learning, to improve the quality of the data. In addition, the use of simulated data also has its benefits in training scenarios, which would allow participants to see real, non-fabricated social media messages in the same temporal manner as found on a social media platform. Lastly, we will discuss the positive reception and future improvements suggested by 911 Public Service Answering Point (PSAP) professionals.
|
|
|
Sooji Han, & Fabio Ciravegna. (2019). Rumour Detection on Social Media for Crisis Management. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: We address the problem of making sense of rumour evolution during crises and emergencies. We study how
understanding and capturing emerging rumours can benefit decision makers during such event. To this end, we
propose a two-step framework for detecting rumours during crises. In the first step, we introduce an algorithm to
identify noteworthy sub-events in real time. In the second step, we introduce a graph-based text ranking method
for summarising newsworthy sub-events while events unfold. We use temporal and content-based features to
achieve the effective and real-time response and management of crises situations. These features can improve
efficiency in the detection of key rumours in the context of a real-world application. The effectiveness of our
method is evaluated over large-scale Twitter data related to real-world crises. The results show that our framework
can efficiently and effectively capture key rumours circulated during natural and human-made disasters.
|
|
|
Sophie Gerstmann, Hans Betke, & Stefan Sackmann. (2019). Towards Automated Individual Communication for Coordination of Spontaneous Volunteers. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: In recent years, spontaneous volunteers often turned out to be a critical factor to overcome disaster situations and
avoid further damages to life and assets. These Volunteers coordinate their activities using social media and
mobile devices but are not integrated in usual command and control structures of disaster responders. The lack of
professional disaster response knowledge leads to a waste of potential workforce or even dangerous situations for
the volunteers. In this paper, a novel approach for a centralized coordination of spontaneous volunteers through
disaster response professionals while using popular communication channels esp. messaging services (e.g.
Facebook Messenger, WhatsApp) is presented. The architecture of a volunteer coordination system focusing on
automated multi-channel communication is shown and the possibilities of a universal chatbot for individual
assignment and scheduling of volunteers are discussed. The paper also provides first insights in a demonstrator
system as a practical solution.
|
|
|
Starr Roxanne Hiltz, Amanda Hughes, Muhammad Imran, Linda Plotnick, Robert Power, & Murray Turoff. (2019). Requirements for Software to Support the use of Social Media in Emergency Management: A Delphi Study. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Social Media contain a wealth of information that could improve the situational awareness of Emergency Managers during a crisis, but many barriers stand in the way. These include information overload, making it impossible to deal with the flood of raw posts, and lack of trust in unverified crowdsourced data. The purpose of this project is to build a communications bridge between emergency responders and technologists who can provide the advances needed to realize social media?s full potential. We are employing a Delphi study survey design, which is a technique for exploring and developing consensus among a group of experts around a particular topic. Participants include emergency managers and technologists with experience in software to support the use of social media in crisis response, from many countries. The topics of the study are described and preliminary, partial results presented for Round 1 of the study, based on 33 responses.
|
|
|
Steve Peterson, Keri Stephens, Hemant Purohit, & Amanda Hughes. (2019). When Official Systems Overload: A Framework for Finding Social Media Calls for Help during Evacuations. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: During large-scale disasters it is not uncommon for Public Safety Answering Points (e.g., 9-1-1) to encounter
service disruptions or become overloaded due to call volume. As observed in the two past United States hurricane
seasons, citizens are increasingly turning to social media whether as a consequence of their inability to reach
9-1-1, or as a preferential means of communications. Relying on past research that has examined social media
use in disasters, combined with the practical knowledge of the first-hand disaster response experiences, this paper
develops a knowledge-driven framework containing parameters useful in identifying patterns of shared
information on social media when citizens need help. This effort explores the feasibility of determining
differences, similarities, common themes, and time-specific discoveries of social media calls for help associated
with hurricane evacuations. At a future date, validation of this framework will be demonstrated using datasets
from multiple disasters. The results will lead to recommendations on how the framework can be modified to make
it applicable as a generic disaster-type characterization tool.
|
|
|
Steven Sheetz, Andrea Kavanaugh, Edward Fox, Riham Hassan, Seungwon Yang, Mohamed Magdy, et al. (2019). Information Uses and Gratifications Related to Crisis: Student Perceptions since the Egyptian Uprising. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: People use diverse sources of information, e.g., newspapers, TV, Internet news, social media, and face-to-face
conversations, to make sense of crises. We apply uses and gratifications theory (UGT) and structural equation
modeling to illustrate how using internet-based information sources since the political uprisings in Egypt influence
perceptions of information satisfaction. Consistent with expectations we find that content and process gratifications
constructs combine to explain information satisfaction, while social gratifications do not significantly influence
satisfaction in the context of a crisis. This suggests that UGT is useful for evaluating the use of information
technology in a context where information is limited in quantity and reliability.
|
|
|
Thomas Spielhofer, Anna Sophie Hahne, Christian Reuter, Marc-André Kaufhold, & Stefka Schmid. (2019). Social Media Use in Emergencies of Citizens in the United Kingdom. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: People use social media in various ways including looking for or sharing information during crises or emergencies
(e.g. floods, storms, terrorist attacks). Few studies have focused on European citizens? perceptions, and just one
has deployed a representative sample to examine this. This article presents the results of one of the first
representative studies on this topic conducted in the United Kingdom. The study shows that around a third (34%)
have used social media during an emergency and that such use is more widespread among younger people. In
contrast, the main reasons for not using social media in an emergency include technological concerns and that the
trustworthiness of social media content is doubtful. However, there is a growing trend towards increased use. The
article deduces and explores implications of these findings, including problems potentially arising with more
citizens sharing information on social media during emergencies and expecting a response.
|
|
|
Valerio Lorini, Carlos Castillo, Francesco Dottori, Milan Kalas, Domenico Nappo, & Peter Salamon. (2019). Integrating Social Media into a Pan-European Flood Awareness System: A Multilingual Approach. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: This paper describes a prototype system that integrates social media analysis into the European Flood Awareness
System (EFAS). This integration allows the collection of social media data to be automatically triggered by flood
risk warnings determined by a hydro-meteorological model. Then, we adopt a multi-lingual approach to find
flood-related messages by employing two state-of-the-art methodologies: language-agnostic word embeddings
and language-aligned word embeddings. Both approaches can be used to bootstrap a classifier of social media
messages for a new language with little or no labeled data. Finally, we describe a method for selecting relevant and
representative messages and displaying them back in the interface of EFAS.
|
|
|
Xukun Li, Doina Caragea, Cornelia Caragea, Muhammad Imran, & Ferda Ofli. (2019). Identifying Disaster Damage Images Using a Domain Adaptation Approach. In Z. Franco, J. J. González, & J. H. Canós (Eds.), Proceedings of the 16th International Conference on Information Systems for Crisis Response And Management. Valencia, Spain: Iscram.
Abstract: Approaches for effectively filtering useful situational awareness information posted by eyewitnesses of disasters,
in real time, are greatly needed. While many studies have focused on filtering textual information, the research
on filtering disaster images is more limited. In particular, there are no studies on the applicability of domain
adaptation to filter images from an emergent target disaster, when no labeled data is available for the target disaster.
To fill in this gap, we propose to apply a domain adaptation approach, called domain adversarial neural networks
(DANN), to the task of identifying images that show damage. The DANN approach has VGG-19 as its backbone,
and uses the adversarial training to find a transformation that makes the source and target data indistinguishable.
Experimental results on several pairs of disasters suggest that the DANN model generally gives similar or better
results as compared to the VGG-19 model fine-tuned on the source labeled data.
|
|