Minera??o de opini?es aplicada a m?dias sociais

• Main Author: Souza, Marlo Vieira dos Santos e
• Other Authors: Vieira, Renata
• Format: Others
• Language: Portuguese
• Published: Pontif?cia Universidade Cat?lica do Rio Grande do Sul 2015
• Subjects: INFORM?TICA; PROCESSAMENTO DA LINGUAGEM NATURAL; RECUPERA??O DA INFORMA??O; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Online Access: http://tede2.pucrs.br/tede2/handle/tede/5219

Made available in DSpace on 2015-04-14T14:50:01Z (GMT). No. of bitstreams: 1 448645.pdf: 990116 bytes, checksum: 3ad832cb853385fcbb55975315370396 (MD5) Previous issue date: 2012-03-19 === The competitive environment has become more dynamic in the last few decades due to the great development of information and comunication technologies and to the globalization process.A company manager must, thus, always be well informed about the competitive landscape before making strategic decisions. In this sense, the Competitive Intelligence (CI) emerges as a discipline that aims to systematize the collection and analysis of information in the competitive environment willing to assist decision making. There is, however, an increasing amount of information being produced and released in Internet and traditional media, which become unwieldy. Associated with this, managers still suffer with time constraints to respond to the market stimuli and remain competitive. Thus, it is necessary to maintain a constant staff monitoring the competitive environment to be able to handle the amount of information from this various sources. We believe that the application Text Analysis techniques can help in various stages of such process. This work presents a proposal to use such techniques to aid the process of Competitive Intelligence. We discuss the use of Sentiment Analysis techniques coupled with Named Entity Recognition in texts from social media - especially Twitter - which helps in the analysis of the attitudes of the consumer market towards a brand. We also present a system implementing the proposed techniques, the evaluations made with it and present our conclusions. === O ambiente competitivo se tornou, nas ultimas d?cadas, mais din?mico gra?as ?s tecnologias de informa??o e comunica??o e ? globaliza??o. O gestor, assim, precisa estar sempre bem informado sobre o panorama competitivo antes de tomar decis?es estrat?gicas. Nessa dire??o, a Intelig?ncia Competitiva (IC) surge como uma disciplina que pretende sistematizar a obten??o e an?lise de informa??es do ambiente competitivo com fun??o de auxiliar a tomada de decis?o. H? entretanto uma quantidade crescente de informa??o sendo produzida e disponibilizada em meios como a Internet e m?dias tradicionais, as quais se tornam de dif?cil manejo. Associado a isso, os gestores sofrem ainda com restri??es temporais para responder ao est?mulo do mercado e manteremse competitivos. Dessa forma, ? necess?rio manter uma equipe de monitoramento constante do ambiente competitivo para que se possa lidar com a quantidade de informa??o proveniente de diversas fontes. Acreditamos que a aplica??o de t?cnicas de An?lise de Texto podem auxiliar nas diversas fases do processo de IC. O presente trabalho apresenta uma proposta de utiliza??o de tais t?cnicas para auxiliar o processo de Intelig?ncia Competitiva. Discutimos aqui a utiliza??o de um m?todo de An?lise de Sentimentos aliado ao Reconhecimento de Entidades Nomeadas em textos provenientes de m?dias sociais - particularmente o Twitter - que permitam analisar as atitudes do mercado consumidor quanto a uma determinada marca. S?o apresentados ainda o sistema desenvolvido, as avalia??es realizadas e as conclus?es que tiramos.