Christian Reuter, Gerhard Backfried, Marc-André Kaufhold, & Fabian Spahr. (2018). ISCRAM turns 15: A Trend Analysis of all ISCRAM-Papers 2004-2017. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 445–458). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: In 2004, Information Systems for Crisis Response and Management (ISCRAM) was a new area of research. Pioneering researchers from different continents and disciplines found fellowship at the first ISCRAM work-shop. Around the same time, the use of social media in crises was first recognized in academia. In 2018, the 15th ISCRAM conference will take place, which gives us the possibility to look back on what has already been achieved with regard to IT support in crises using social media. With this article, we examine trends and devel-opments with a specific focus on social media. We analyzed all papers published at previous ISCRAMs (n=1339). Our analysis shows that various platforms, the use of language and coverage of different types of disasters follow certain trends – most noticeably a dominance of Twitter, English and crises with large impacts such as hurricanes or earthquakes can be seen.
|
Diana Fischer, Carsten Schwemmer, & Kai Fischbach. (2018). Terror Management and Twitter: The Case of the 2016 Berlin Terrorist Attack. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 459–468). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: There is evidence that people increasingly use social networking sites like Twitter in the aftermath of terrorist attacks to make sense of the events at the collective level. This work-in-progress paper focuses on the content of Twitter messages related to the 2016 terrorist attack on the Berlin Christmas market. We chose topic modeling to investigate the Twitter data and the terror management theory perspective to understand why people used Twitter in the aftermath of the attack. In particular, by connecting people and providing a real-time communication channel, Twitter helps its users collectively negotiate their worldviews and re-establish self-esteem. We provide first results and discuss next steps.
|
Yajie Li, Amanda Lee Hughes, & Peter D. Howe. (2018). Communicating Crisis with Persuasion: Examining Official Twitter Messages on Heat Hazards. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 469–479). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Official crisis messages need to be persuasive to promote appropriate public responses. However, little research has examined the content of crisis messages from a persuasion perspective, especially for natural hazards. This study deductively identifies five persuasive message factors (PMFs) applicable to natural hazards, including two under-examined health-related PMFs: health risk susceptibility and health impact. Using 2016 heat hazards as a case study, this paper content-analyzes heat-related Twitter messages (N=904) posted by eighteen U.S. National Weather Service Weather Forecast Offices according to the five PMFs. We find that the use of descriptions of hazard intensity is disproportionately high, with a lack of use of other PMFs. We also describe different types of statements used to signal the two health-related PMFs. We conclude with implications and recommendations relevant to practitioners and researchers in social media crisis communication.
|
Mahshid Marbouti, Craig Anslow, & Frank Maurer. (2018). Evaluation results for a Social Media Analyst Responding Tool. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 480–492). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: We take a human-centered design approach to develop a fully functional prototype, SMART (“Social Media Analyst Responding Tool”), informed by emergency practitioners. The prototype incorporates machine learning techniques to identify relevant information during emergencies. In this paper, we report the result of a user study to gather qualitative feedback on SMART. The evaluation results offer recommendations into the design of Social Media analysis tools for emergencies. The evaluation findings show the interest of emergency practitioners into designing such solutions; it reflects their need to not only identify relevant information but also to further perceive the outcome of their actions in social media. We found out there is a notable emphasis on the sentiment from these practitioners and social media analysis tools need to do a better job of handling negative sentiment within the emergency concept.
|
Jan Wendland, Christian Ehnis, Rodney J. Clarke, & Deborah Bunker. (2018). Sydney Siege, December 2014: A Visualisation of a Semantic Social Media Sentiment Analysis. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 493–506). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Sentiment Analyses are widely used approaches to understand and identify emotions, feelings, and opinion on social media platforms. Most sentiment analysis systems measure the presumed emotional polarity of texts. While this is sufficient for some applications, these approaches are very limiting when it comes to understanding how social media users actually use language resources to make sense of extreme events. In this paper, a Sentiment Analysis based on the Appraisal System from the theory of communication called Systemic Functional Linguistics is applied to understand the sentiment of event-driven social media communication. A prototype was developed to analyze Twitter data using the Appraisal System. This prototype was applied to tweets collected during and after the Sydney Siege 2014, a hostage situation in a busy café in Sydney. Because the Appraisal System is a theorised functional communication method, the results of this analysis are more nuanced than is possible with traditional polarity based sentiment analysis.
|
Shane Errol Halse, Aurélie Montarnal, Andrea Tapia, & Frederick Benaben. (2018). Bad Weather Coming: Linking social media and weather sensor data. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 507–515). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: In this paper we leverage the power of citizen supplied data. We examined how both physical weather sensor data (obtained from the weather underground API) and social media data (obtained from Twitter) can serve to improve local community awareness during a severe weather event. A local tornado warning was selected due to its small scale and isolated geographic area, and only Twitter data found from within this geo-locational area was used. Our results indicate that during a severe weather event, an increase in weather activity obtained from the local weather sensors does correlate with an increase in local social media usage. The data found on social media also contains additional information from, and about the community of interest during the event. While this study focuses on a small scale event, it provides the groundwork for use during a much larger weather event.
|
Shane Halse, Jomara Binda, & Samantha Weirman. (2018). It's what's outside that counts: Finding credibility metrics through non-message related Twitter features. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 516–528). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Social media data, such as Twitter, enables crisis response personnel and civilians to share information during a crisis situation. However, a lack of information gatekeeping processes also translates into concerns about both content and source credibility. This research aims to identify Twitter metrics which could assist with the latter. A 2 (average number of hashtags used) x 2 (ratio of tweets/retweets posted) x 2 (ratio of follower/followee) between-subjects experiment was conducted to evaluate the level of influence of Twitter broker metrics on behavioral intention and the perception of source credibility. The findings indicate that follower/followee ratio in conjunction with hashtag usage approached a significant effect on perceived source credibility. In addition, both Twitter awareness metrics and dispositional trust played an important role in determining behavioral intentions and perceived source credibility. Implications and limitations are also discussed.
|
Hussein Mouzannar, Yara Rizk, & Mariette Awad. (2018). Damage Identification in Social Media Posts using Multimodal Deep Learning. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 529–543). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Social media has recently become a digital lifeline used to relay information and locate survivors in disaster situations. Currently, officials and volunteers scour social media for any valuable information; however, this approach is implausible as millions of posts are shared by the minute. Our goal is to automate actionable information extraction from social media posts to efficiently direct relief resources. Identifying damage and human casualties allows first responders to efficiently allocate resources and save as many lives as possible. Since social media posts contain text, images and videos, we propose a multimodal deep learning framework to identify damage related information. This framework combines multiple pretrained unimodal convolutional neural networks that extract features from raw text and images independently, before a final classifier labels the posts based on both modalities. Experiments on a home-grown database of labeled social media posts showed promising results and validated the merits of the proposed approach.
|
Aibek Musaev, Kimberly Stowers, & Jonghun Kam. (2018). Harnessing Data to Create an Effective Drought Management System. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 544–552). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Drought is a complex climate phenomenon with slow emergence and quick vanish, which makes it hard for stakeholders to respond to drought timely. To reduce the vulnerability of our society to future drought, a better understanding of how society responds to drought is critical. Here, we propose a pilot study about social response to a recent California drought through social media. In this study, we identify the most important users using an extension of PageRank algorithm. We investigate the key drivers of the public activity in February, 2014 during the California drought. We also create a word cloud visualization from the most retweeted tweets. Lastly, we specify the information sources from those tweets. The findings of this study inform us that big data can help us to improve the current drought response plans through fundamental understanding of social response to drought, which is applicable to other natural hazards.
|
Firoj Alam, Ferda Ofli, Muhammad Imran, & Michael Aupetit. (2018). A Twitter Tale of Three Hurricanes: Harvey, Irma, and Maria. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 553–572). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: People increasingly use microblogging platforms such as Twitter during natural disasters and emergencies. Research studies have revealed the usefulness of the data available on Twitter for several disaster response tasks. However, making sense of social media data is a challenging task due to several reasons such as limitations of available tools to analyze high-volume and high-velocity data streams. This work presents an extensive multidimensional analysis of textual and multimedia content from millions of tweets shared on Twitter during the three disaster events. Specifically, we employ various Artificial Intelligence techniques from Natural Language Processing and Computer Vision fields, which exploit different machine learning algorithms to process the data generated during the disaster events. Our study reveals the distributions of various types of useful information that can inform crisis managers and responders as well as facilitate the development of future automated systems for disaster management.
|
Apoorva Chauhan, & Amanda Lee Hughes. (2018). Social Media Resources Named after a Crisis Event. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 573–583). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Crisis Named Resources (CNRs) are the social media accounts and pages named after a crisis event. CNRs typically appear spontaneously after an event as places for information exchange. They are easy to find when searching for information about the event. Yet in most cases, it is unclear who manages these resources. Thus, it is important to understand what kinds of information they provide and what role they play in a response. This paper describes a study of Facebook and Twitter CNRs around the 2016 Fort McMurray wildfire. We report on CNR lifecycles, and their relevance to the event. Based on the information provided by these resources, we categorize them into 8 categories: donations, fundraisers, prayers, reactions, reports, needs and offers, stories, and unrelated. We also report on the most popular CNR on both Facebook and Twitter. We conclude by discussing the role of CNRs and the need for future investigation.
|
Takuya Oki. (2018). Possibility of Using Tweets to Detect Crowd Congestion: A Case Study Using Tweets just before/after the Great East Japan Earthquake. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 584–596). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: During large earthquakes, it is critical to safely guide evacuation efforts and to prevent accidents caused by congestion. In this paper, we focus on detecting the degree of crowd congestion following an earthquake based on information posted to Social Networking Services (SNSs). This research uses text data posted to Twitter just before/after the occurrence of the Great East Japan Earthquake (11 March 2011 at 02:46 PM JST). First, we extract co-occurring place names, proper nouns, and time-series information from tweets about congestion in the Tokyo metropolitan area (TMA). Next, using these extracted data, we analyze the frequency and spatiotemporal characteristics of these tweets. Finally, we identify expressions that describe the degree of crowd congestion and discuss methods to quantify these expressions based on a questionnaire survey and tweets that contain a photograph.
|
Grégoire Burel, & Harith Alani. (2018). Crisis Event Extraction Service (CREES) – Automatic Detection and Classification of Crisis-related Content on Social Media. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 597–608). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Social media posts tend to provide valuable reports during crises. However, this information can be hidden in large amounts of unrelated documents. Providing tools that automatically identify relevant posts, event types (e.g., hurricane, floods, etc.) and information categories (e.g., reports on affected individuals, donations and volunteering, etc.) in social media posts is vital for their efficient handling and consumption. We introduce the Crisis Event Extraction Service (CREES), an open-source web API that automatically classifies posts during crisis situations. The API provides annotations for crisis-related documents, event types and information categories through an easily deployable and accessible web API that can be integrated into multiple platform and tools. The annotation service is backed by Convolutional Neural Networks (CNNs) and validated against traditional machine learning models. Results show that the CNN-based API results can be relied upon when dealing with specific crises with the benefits associated with the usage word embeddings.
|
Rob Grace, Jess Kropczynski, & Andrea Tapia. (2018). Community Coordination: Cooperative Uses of Social Media in Community Emergency Management. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 609–620). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Emergency managers continue to struggle with a lack of staff, information processing tools, and sufficient trust in citizen-reported information to coordinate the use of social media in their communities. To understand possibilities for overcoming these barriers, we conduct interviews with emergency managers using scenarios describing the projective activities of community volunteers disseminating and monitoring social media. We find that coordinating social media use in communities requires alignments with local incident management systems and, in particular, existing sociotechnical infrastructure for managing citizen-reported information. These alignments open limited roles for community volunteers, notably coordinating the redistribution of official information; and stand to reshape the workflows and infrastructures of incident management systems by pushing emergency dispatchers to proactively process indirect reports of incidents obtained on social media, and integrate tools that can access hyperlocal data, curate incident and situational reports, and facilitate sensemaking among officials confronted with multiple information sources.
|
Sofia Eleni Spatharioti, Sara Wylie, & Seth Cooper. (2018). Does Flight Path Context Matter? Impact on Worker Performance in Crowdsourced Aerial Imagery Analysis. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 621–628). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Natural disasters result in billions of dollars in damages annually and communities left struggling with the difficult task of response and recovery. To this end, small private aircraft and drones have been deployed to gather images along flight paths over the affected areas, for analyzing aerial photography through crowdsourcing. However, due to the volume of raw data, the context and order of these images is often lost when reaching workers. In this work, we explored the effect of contextualizing a labeling task on Amazon Mechanical Turk, by serving workers images in the order they were collected on the flight and showing them the location of the current image on a map. We did not find a negative impact from the loss of contextual information, and found map context had a negative impact on worker performance. This may indicate that ordering images based on other criteria may be more effective.
|
Laura Petersen, Laure Fallou, Grigore Havarneanu, Paul Reilly, Elisa Serafinelli, & Rémy Bossu. (2018). November 2015 Paris Terrorist Attacks and Social Media Use: Preliminary Findings from Authorities, Critical Infrastructure Operators and Journalists. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 629–638). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Crisis communication is a key component of an effective emergency response. Social media has evolved as a prominent crisis communication tool. This paper reports how social media was used by authorities, critical infrastructure operators and journalists during the terrorist attacks that hit Paris on 13th November 2015. A qualitative study was conducted between January and February 2017 employing semi-structured interviews with seven relevant stakeholders involved in this communication process. The preliminary critical thematic analysis revealed four main themes which are reported in the results section: (1) social media is used in crisis times; (2) authorities gained situational awareness via social media; (3) citizens used social media to help one another; and (4) communication procedures changed after these critical events. In conclusion, authorities, citizens and journalists all turned to social media during the attack, both for crisis communication and for increasing situational awareness.
|
William R. Smith, Keri K. Stephens, Brett Robertson, Jing Li, & Dhiraj Murthy. (2018). Social Media in Citizen-Led Disaster Response: Rescuer Roles, Coordination Challenges, and Untapped Potential. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 639–648). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Widespread disasters can overload official agencies' capacity to provide assistance, and often citizen-led groups emerge to assist with disaster response. As social media platforms have expanded, emergent rescue groups have many ways to harness network and mobile tools to coordinate actions and help fellow citizens. This study used semi-structured interviews and photo elicitation techniques to better understand how wide-scale rescues occurred during the 2017 Hurricane Harvey flooding in the Greater Houston, Texas USA area. We found that citizens used diverse apps and social media-related platforms during these rescues and that they played one of three roles: rescuer, dispatcher, or information compiler. The key social media coordination challenges these rescuers faced were incomplete feedback loops, unclear prioritization, and communication overload. This work-in-progress paper contributes to the field of crisis and disaster response research by sharing the nuances in how citizens use social media to respond to calls for help from flooding victims.
|
Leon Derczynski, Kenny Meesters, Kalina Bontcheva, & Diana Maynard. (2018). Helping Crisis Responders Find the Informative Needle in the Tweet Haystack. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 649–662). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Crisis responders are increasingly using social media, data and other digital sources of information to build a situational understanding of a crisis situation in order to design an effective response. However with the increased availability of such data, the challenge of identifying relevant information from it also increases. This paper presents a successful automatic approach to handling this problem. Messages are filtered for informativeness based on a definition of the concept drawn from prior research and crisis response experts. Informative messages are tagged for actionable data – for example, people in need, threats to rescue efforts, changes in environment, and so on. In all, eight categories of actionability are identified. The two components – informativeness and actionability classification – are packaged together as an openly-available tool called Emina (Emergent Informativeness and Actionability).
|
Samuel Lee Toepke. (2018). Leveraging Elasticsearch and Botometer to Explore Volunteered Geographic Information. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 663–676). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: In the past year, numerous weather-related disasters have continued to display the critical importance of crisis management and response. Volunteered geographic information (VGI) has been previously shown to provide illumination during all parts of the disaster timeline. Alas, for a geospatial area, the amount of data provided can cause information overload, and be difficult to process/visualize. This work presents a set of open-source tools that can be easily configured, deployed and maintained, to leverage data from Twitter's streaming service. The user interface presents data in near real-time, and allows for dynamic queries, visualizations, maps and dashboards. Another VGI challenge is quantifying trustworthiness of the data. The presented work shows integration of a Twitter-bot assessment service, which uses several heuristics to determine the bot-ness of a Twitter account. Architecture is described, Twitter data from a major metropolitan area is explored using the tools, and conclusions/follow-on work are discussed.
|
Venkata Kishore Neppalli, Cornelia Caragea, & Doina Caragea. (2018). Deep Neural Networks versus Naive Bayes Classifiers for Identifying Informative Tweets during Disasters. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 677–686). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: In this paper, we focus on understanding the effectiveness of deep neural networks by comparison with the effectiveness of standard classifiers that use carefully engineered features. Specifically, we design various feature sets (based on tweet content, user details and polarity clues) and use these feature sets individually or in various combinations, with Naïve Bayes classifiers. Furthermore, we develop neural models based on Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) with handcrafted architectures. We compare the two types of approaches in the context of identifying informative tweets posted during disasters, and show that the deep neural networks, in particular the CNN networks, are more effective for the task considered.
|
Kiran Zahra, Muhammad Imran, & Frank O Ostermann. (2018). Understanding eyewitness reports on Twitter during disasters. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 687–695). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Social media platforms such as Twitter provide convenient ways to share and consume important information during disasters and emergencies. Information from bystanders and eyewitnesses can be useful for law enforcement agencies and humanitarian organizations to get firsthand and credible information about an ongoing situation to gain situational awareness among other uses. However, identification of eyewitness reports on Twitter is challenging for many reasons. This work investigates the sources of tweets and classifies them into three types (i) direct eyewitnesses, (ii) indirect eyewitness, and (iii) vulnerable accounts. Moreover, we investigate various characteristics associated with each kind of eyewitness account. We observe that words related to perceptual senses (feeling, seeing, hearing) tend to be present in direct eyewitness messages, whereas emotions, thoughts, and prayers are more common in indirect witnesses. We believe these characteristics can help make more efficient computational methods and systems in the future for automatic identification of eyewitness accounts.
|
Rachel Samuels, John Eric Taylor, & Neda Mohammadi. (2018). The Sound of Silence: Exploring How Decreases in Tweets Contribute to Local Crisis Identification. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 696–704). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Recent research has identified a correlation between increasing Twitter activity and incurred damage in disasters. This research, however, fails to account for localized emergencies occurring in areas in which people have lost power, otherwise lack internet connectivity, or are uncompelled to Tweet during a disaster. In this paper, we analyze the correlation between daily Tweet counts and FEMA Building Level Damage Assessments during Hurricane Harvey. We find that the absolute deviation of Tweet counts from steady state is a potentially useful tool for the evolving information needs of emergency responders. Our results show this to be a more consistent and persistent metric for flood damage across the full temporal extent of the disaster. This shows that, when considering the varied information needs of emergency responders, social media tools that seek to identify emergencies need to consider both where Tweet counts are increasing and where they are dropping off.
|
Alan Aipe, Asif Ekbal, Mukuntha NS, & Sadao Kurohashi. (2018). Linguistic Feature Assisted Deep Learning Approach towards Multi-label Classification of Crisis Related Tweets. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 705–717). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Micro-blogging site like Twitter, over the last decade, has evolved into a proactive communication channel during mass convergence and emergency events, especially in crisis stricken scenarios. Extracting multiple levels of information associated with the overwhelming amount of social media data generated during such situations remains a great challenge to disaster-affected communities and professional emergency responders. These valuable data, segregated into different informative categories, can be leveraged by the government agencies, humanitarian communities as well as citizens to bring about faster response in areas of necessity. In this paper, we address the above scenario by developing a deep Convolutional Neural Network (CNN) for multi-label classification of crisis related tweets.We augment deep CNN by several linguistic features extracted from Tweet, and investigate their usage in classification. Evaluation on a benchmark dataset show that our proposed approach attains the state-of-the-art performance.
|
Songhui Yue, Jyothsna Kondari, Aibek Musaev, Songqing Yue, & Randy Smith. (2018). Using Twitter Data to Determine Hurricane Category: An Experiment. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 718–726). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Social media posts contain an abundant amount of information about public opinion on major events, especially natural disasters such as hurricanes. Posts related to an event, are usually published by the users who live near the place of the event at the time of the event. Special correlation between the social media data and the events can be obtained using data mining approaches. This paper presents research work to find the mappings between social media data and the severity level of a disaster. Specifically, we have investigated the Twitter data posted during hurricanes Harvey and Irma, and attempted to find the correlation between the Twitter data of a specific area and the hurricane level in that area. Our experimental results indicate a positive correlation between them. We also present a method to predict the hurricane category for a specific area using relevant Twitter data.
|
Reza Mazloom, HongMin Li, Doina Caragea, Muhammad Imran, & Cornelia Caragea. (2018). Classification of Twitter Disaster Data Using a Hybrid Feature-Instance Adaptation Approach. In Kees Boersma, & Brian Tomaszeski (Eds.), ISCRAM 2018 Conference Proceedings – 15th International Conference on Information Systems for Crisis Response and Management (pp. 727–735). Rochester, NY (USA): Rochester Institute of Technology.
Abstract: Huge amounts of data that are generated on social media during emergency situations are regarded as troves of critical information. The use of supervised machine learning techniques in the early stages of a disaster is challenged by the lack of labeled data for that particular disaster. Furthermore, supervised models trained on labeled data from a prior disaster may not produce accurate results, given the inherent variation between the current and the prior disasters. To address the challenges posed by the lack of labeled data for a target disaster, we propose to use a hybrid feature-instance adaptation approach based on matrix factorization and the k nearest neighbors algorithm, respectively. The proposed hybrid adaptation approach is used to select a subset of the source disaster data that is representative for the target disaster. The selected subset is subsequently used to learn accurate Naive Bayes classifiers for the target disaster.
|
