ABSTRACT

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).

This project utilizes information gleaned from social media about upcoming events to inform designated authorities in a timely manner so they can prepare mitigating action plans in case of emergency. Besides the extracted events themselves, harvested information may include (but is not limited to) images, posted messages, people?s sentiments and other surrounding context which will improve relevancy and trust of the information in understanding emergency situations. Extracted events become the source for investigating and analyzing spatial-temporal influences between events and cross-domain events to derive further insights. Potential applications are real-time tracking and monitoring of events for disaster relief, and forecasting of events for mitigation plans. Project outcomes will benefit researchers in information extraction and integration with interests in graph models and transfer learning; in addition to providing practical studying materials in areas such as deep learning, spatio-temporal data causality and analysis for students about disaster resilience and progressing towards community resilience in the long term. Moreover, the work will increase research capacity and collaborations to generate new research opportunities for students from underrepresented communities to pursue advanced degrees in computer science.

Social media data provides a means to identify happening events prior, during, and post disasters. It provides signals for designed authorities for reactions and mitigation planning. This research will use social media posts, machine learning, and transfer learning techniques in three thrusts: 1) Extract local and global events; 2) Embed surrounding context such as relevance and trust; 3) Analyze spatial-temporal relationship between events and cross-domain events for further insights. This project puts forth a novel approach to events analysis under the umbrella of graph neural network and transfer learning, leveraging recent advances and opportunities in deep learning. The resulting data-driven algorithms will be modelled emphasizing the socio-economic aspects of the consequences and cascading losses by allowing the system to adapt according to the community-based variables and the dynamics of the disasters. The findings will be disseminated via publications, source code, and data to reach diverse communities of researchers and students.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Please report errors in award information by writing to: awardsearch@nsf.gov.