Communication networks of natural disasters: lexical-semantic indicators of social relevance in a corpus of newspaper reports

• Grant number: 11/16682-8
• Support Opportunities: Scholarships in Brazil - Master
• Start date: March 01, 2012
• End date: February 28, 2013
• Field of knowledge: Applied Social Sciences - Communications
• Principal Investigator: Margarethe Born Steinberger-Elias
• Grantee: Ariana Moura da Silva
• Host Institution: Centro de Engenharia, Modelagem e Ciências Sociais Aplicadas (CECS). Universidade Federal do ABC (UFABC). Ministério da Educação (Brasil). Santo André , SP, Brazil

Abstract

This research integrates an interdisciplinary project that put together the areas of Communication, Linguistic and Computing to model communication networks about disasters in Portuguese. Based on an assortment of journalistic reports (corpus) about natural disasters in Latin America, the research aims at the identification of linguistic indicators automatically capable of: a) recognizing vocabulary (lexicon) typical of natural disasters; b) classify and structure a hierarchy of news´ themes related to the corpus with the criteria of social relevance. The results from steps a) and b) will be validated in the step c) set against a corpus of Web pages with contents on natural disasters. The main drive behind the research is to develop solutions for an efficient and rapid monitoring of information with social relevancy in urgent situations. In this context, communication networks might be of substantial aid, performing the evaluation of contents of messages according to their user-relevance, eliminating superfluous information, attaching users with similar search terms and even generating help warns, if needed. Capturing information on networks must be based on devices capable of recognizing and interpreting human language reports. The project attempts to create a device module able to recognize the typical lexicon of disasters in Portuguese as well as defining and structuring a hierarchy of contents by social relevance in this semantic domain. The methodology will be based on statistical and semantic grouping techniques applied to a sample of journalistic texts and validated against a sample of Web pages. (AU)