Optimising Crisis Management Systems Through Data-Driven Risk Communication Models: An ISM Approach to Enhancing Organisational Resilience

Abstract

Crisis Management Systems (CMSs) serve as fundamental elements in emergency response, enabling control over natural disasters, pandemics, and cybersecurity incidents. CMSs often perform inadequately when handling unstructured emergency communication and delayed messages. Frequent emergencies, along with their complex nature, necessitate response approaches that extend beyond conventional methods. The failure of traditional CMS systems to detect and organise messages promptly leads to time-consuming response delays, which in turn result in coordination breakdowns and the wider spread of false information. This study addresses these problems through a data-driven risk communication system that integrates NLP with ISM to achieve its objectives. The research utilises Kaggle’s Multilingual Disaster Response data repository, which comprises actual crisis messages exceeding 30,000 entries. The text cleaning process and tokenisation, along with sentiment analysis, were performed using NLP methods before the SVM-based classification of urgent messages. The Implementation of ISM began after message classification, where urgent messages were allocated to specific hierarchical positions, with sentiment serving as a determining factor in the detection process. Evaluation standards confirmed that the model functioned effectively. Testing revealed the exceptional performance of the SVM, as it identified messages with 99% accuracy in both urgent and non-urgent categories. Only a minimal number of urgent communications needed reclassification according to the confusion matrix results. High-priority urgent messages with negative sentiment were properly identified, as most messages had a low-priority status according to the ISM-based priority distribution. The research results demonstrate that an ISM-NLP hybrid model effectively improves communication efficiency in CMSs. The crisis communication system based on this approach provides an intelligent and structured method for managing emergency responses, enabling fast decision-making and swift emergency responses.