Slándáil partners organised the Emotions, Metaphors, Ontology and Terminology (EMOT) Workshop as part of Language Resources and Evaluation Conference (LREC) on May 23rd, 2016. The workshop presented advances in analytical methods of verbal and non-verbal communication during impending disasters. The audience consisted of academics and experts in information extraction and text analytics, disaster management and communication professionals.
The workshop consisted of presentations that focused on the terminology, type of content and emotional involvement of messages sent and received during disasters, particularly in social media. Statistical and semantic processing of these messages facilitates the creation of useful databases and ontologies, which can be used by emergency managers to situate and confront a disaster more efficiently and effectively. If the emergency handlers understand what kind of information is disseminated in different types of social media, they can direct their efforts with a greater confidence. The topics covered included the emotive and terminological language used in disaster related data, automatic extraction and evaluation of terminology from messages, the social media communication by civilians and volunteers, disaster monitoring and mitigation and the creation of a disaster ontology.
The authors who contributed to this workshop have been working on how to specify, design and trial disaster management systems that use social media as one of their inputs. Their focus is on extracting information from continuous data streams including text, speech and images. Khurshid Ahmad, the project coordinator, opened the session introducing Slándáil to the audience. University of Padua partners presented two works, one mainly focused on emotions and non-verbal communication during disasters and the other focusing on terminology and software training.
Busà and Cravotta presented work carried out on gestures and speech of Italian reporters working in areas threatened by natural disasters like floods. They suggest that hand gestures and the modulation of voice may influence the perception of the speakers’ emotional involvement and professionalism. Speakers’ emotional involvement can correlate with the severity of an impending disaster. Musacchio, Panizzon, Zhang and Zorzi presented their work on the use of terminology and ontology for detecting impending and current emergencies in social media streams. Their work was aimed at identifying communicative strategies, linguistic characteristics and sentiment of text messages that can be used to filter disaster-related communications on social media.
Maria Spyropoulou from Trinity College Dublin gave a presentation on the content found in textual and audial disaster-related messages. It showed that speech recordings from the preparedness or recovery phase of a disaster are more suitable for sentiment analysis. A further paper was presented as a feasibility study aimed at building terminology and ontology in the domain of disaster management. A corpus-based text analytic system was evaluated, CiCui, which combines frequency, collocation and linguistic analyses to extract candidates terminologies from corpora comprised of domain texts from diverse sources. CiCui was assessed against four terminology extraction systems and the initial results show that it has an above average precision in extracting disaster-related terms.
INFAI presented work on topic modelling and information extraction from Facebook and Twitter. The paper studies the results of a German case study on social media use during the flood 2013 in Central Europe. In their investigation, the researchers applied state-of-the-art Natural Language Processing technology, mainly topic modeling, to test and demonstrate its usefulness for computer-based media analysis in modern disaster management.
The audience actively responded to presentations with questions and comments and the workshop became a welcoming forum for people sharing ideas and advances in the field. Project Slándáil received very positive feedback from attendants.
The proceedings of the workshop are available at LREC online.