Slandail: A Security System for Language and Image Analysis - Project No: 607691

Abstract

In major events, such as natural disasters, the amount of data shared through social media increases dramatically. It is difficult for individuals to monitor this large amount of information. Disaster response agencies, including fire services, police, ambulance services and disaster management agencies, use social networks – Facebook, Instagram – and social media systems – Twitter, WhatsApp – during the different phases of disaster management (preparedness, response, mitigation and recovery) both as inputs and outputs for relevant information (Lindsay 2016). SLANDAIL explored the impact of social media in emergencies and harnesses information generated during disaster events to enhance control room systems for emergency management. It synthesised expertise in emergency management, text and image analytics, communication, ethics and law to present a system for leveraging social media data for the security of citizens in vulnerable areas, supported by a customised License Agreement. Our research developed the necessary technical capabilities and addressed the social and ethical implications of an integrated control room system for emergency management. SLANDAIL enabled tracking of the informational evolution of disaster events using natural language processing and machine learning techniques, such as topic modelling and statistical machine learning, in English, German and Italian. The importance of language in disaster management resulted in the creation of guidelines for communications for disaster managers to enhance trust, authority, clarity and empathy. Non-verbal communication analysis, including facial expressions, vocal quality and gesture recognition in real-life announcements of impending disasters, were incorporated to improve emergency management communication techniques. SLANDAIL developed response tools to operate within an ethical framework. Our legal research culminated in the development of a model license agreement – the first of its kind – for a social media analysis system for disaster response, ensuring that the system’s operation complies with EU and national Data Protection legislation and with human rights initiatives related to privacy and security. SLANDAIL prototypes were evaluated positively by disaster management and disaster recovery experts, and partners have engaged technology transfer organisations for the subsequent commercialisation of the systems and technologies developed during the project with state organisations and venture funds. SLANDAIL produced a prototype that integrates social media input from text and image into disaster management control room software. The integrated system incorporates advances in text and image analytics and links software built using research from three universities and development by two SMEs. It was specified and tested by (i) police forces An Garda Siochana, Ireland, and Police Service of Northern Ireland, UK, (ii) army reserves Bundesministerium der Verteidigung, Saxony, Germany, (iii) civil protection authorities Protezione Civile de Veneto, Italy and (iv) media communications experts (Stillwater Communications, Ireland). It has been presented to large meetings of disaster managers in Ireland, Northern Ireland, Germany and Italy and evaluated by experts in emergency planning (Business Continuity Institute, Emergency Planning Society).