Identification of health concepts in social networks

• Autor(a) principal: Matos, André Pereira de
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/14716
• topic: Engenharia de computadores; Redes sociais - Análise de dados; Recuperação da informação; Medicamentos - Consumo

Social Networks are, without doubt, one of the fastest growing trends on the Internet nowadays and the amount of information generated is huge. On the other hand, search engines such as Google, Yahoo and Bing became global interfaces for information search. From an enterprise point of view, this phenomenon has lead to changes in strategies to deal with information representation, indexing and searching. Engines like Lucene, Nutch, Solr, Sphinx, ElasticSearch, among others, were born and became important platforms of software engineering. The amount of information available on the Internet on a daily basis, can be utterly valuable for multiple goals - social, scientific, politic, economic. For instance, in the case of health related subjects, institutions keep information on patients such as pathologies, conditions, treatments, exam results, pathways, etc. Also, governmental institutions divulge more and more information on health subjects in the most varied aspects. On the other hand, health research is one of the most active areas, resulting on network publishing of new results. Adding to the aforementioned, the way people share information on their health status can be valuable to detect pandemics and also to identify clinical conditions and their causes. In this project we proposed to develop a system capable of processing information retrieved from social networks, identifying symptoms, disorders and drugs. From an engineering point of view, the process consisted in accessing varied data sources to obtain information, create dictionaries and models of natural language processing to identify text patterns and associations between concepts. In the future, these associations can be valuable, for instance, to detect adverse reaction to drugs.