Inteligência competitiva na cafeicultura: mineração textual em notícias publicadas na web
Data
2016-07-26
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Viçosa
Resumo
A cafeicultura tem um papel significativo para o agronegócio no Brasil, mas é uma atividade de risco elevado pela variação de preço do café que causa impactos em diferentes setores de sua cadeia produtiva. Isto exige dos agentes Inteligência Competitiva para monitorar o ambiente competitivo por meio de um processo contínuo e sistemático de coleta e análise de informações para tomada de decisões em gestão de risco. Notícias com informações que influenciam o mercado de café e afetam a dinâmica da sua cadeia produtiva são publicadas na web diariamente. Entretanto, lidar com o volume e velocidade dessas informações não é uma tarefa trivial, consome recursos humanos, tempo e restringe a análise à capacidade de busca e leitura dos especialistas. E a automatização do processo, apesar do avanço da tecnologia, esbarra em obstáculos no campo sintático como ruídos nos dados e semântico como ambiguidade da linguagem e ausência de contexto. Neste cenário, por meio do processo iterativo do método Design Science Research, foi possível, juntamente com especialistas, adquirir conhecimento sobre requisitos de inteligência para cafeicultura e construir artefatos para coletar e classificar automaticamente, pela perspectiva de IC, notícias da web sobre eventos que impactam o mercado de café. Uma avaliação estatística mostrou correlação entre a ocorrência cronológica destes eventos e a série de preço e volatilidade do café, enquanto uma avaliação qualitativa por especialista apontou a relevância das notícias para análise de requisitos de inteligência na cafeicultura. Estes resultados apontam viabilidade de um indicador de evidências qualitativas vindas da web que, a saber, sua influência e erro, e confrontado com uma análise qualitativa, permita perceber aumento de volatilidade e viés para delinear um cenário de tomada de decisões para gestão de risco. Desta forma, a pesquisa corrobora a possiblidade de promover Inteligência Competitiva para apoiar decisões sobre gerenciamento de risco e competitividade na cafeicultura por meio de Mineração Textual em notícias publicadas na web.
Coffee production plays a significant role in Brazilian agribusiness. However, it is a high-risk activity given the impacts in different sectors of the production chain caused by coffee price variation. This demands Competitive Intelligence for the agents to monitor the competitive environment by means of a continuous and systematic process of information gathering and analysis for the decision- making in risk management. News with information that influence the coffee market and that affect the dynamics of its production chain are daily published online. However, dealing with the volume and speed of this information is not an easy task. It consumes human resources, time and restricts the capacity analysis of the specialists seeking and reading. The automation of the process, despite the advance in technology, meets obstacles in the syntactic field, such as residue on the data, and semantics, such as language ambiguity and absence of context. In this scenery, it was possible to acquire knowledge with the specialists regarding the intelligence for coffee production, by means of the iterative process of the Design Science Research method, and construct artifacts to automatically collect and classify, in the Competitive Intelligence perspective, internet news on events that influence the coffee market. A statistical evaluation showed correlation between the chronologic occurrence of these events and the price series and coffee volatility, while a qualitative evaluation made by specialist pointed the relevance of the news for analyzing the intelligence requisites in coffee production. These results point to the viability of an indicator for qualitative evidence derived from the internet, and its influence and error, while confronting the qualitative analysis, allowing us to perceive an increase in volatility and bias to design a decision-making scenery for risk management. Thus, this research corroborates the possibility of promoting Competitive Intelligence to support decisions regarding risk management and competitiveness in coffee production by means of Text Mining in news published in the internet.
Coffee production plays a significant role in Brazilian agribusiness. However, it is a high-risk activity given the impacts in different sectors of the production chain caused by coffee price variation. This demands Competitive Intelligence for the agents to monitor the competitive environment by means of a continuous and systematic process of information gathering and analysis for the decision- making in risk management. News with information that influence the coffee market and that affect the dynamics of its production chain are daily published online. However, dealing with the volume and speed of this information is not an easy task. It consumes human resources, time and restricts the capacity analysis of the specialists seeking and reading. The automation of the process, despite the advance in technology, meets obstacles in the syntactic field, such as residue on the data, and semantics, such as language ambiguity and absence of context. In this scenery, it was possible to acquire knowledge with the specialists regarding the intelligence for coffee production, by means of the iterative process of the Design Science Research method, and construct artifacts to automatically collect and classify, in the Competitive Intelligence perspective, internet news on events that influence the coffee market. A statistical evaluation showed correlation between the chronologic occurrence of these events and the price series and coffee volatility, while a qualitative evaluation made by specialist pointed the relevance of the news for analyzing the intelligence requisites in coffee production. These results point to the viability of an indicator for qualitative evidence derived from the internet, and its influence and error, while confronting the qualitative analysis, allowing us to perceive an increase in volatility and bias to design a decision-making scenery for risk management. Thus, this research corroborates the possibility of promoting Competitive Intelligence to support decisions regarding risk management and competitiveness in coffee production by means of Text Mining in news published in the internet.
Descrição
Tese de Doutorado defendida na Universidade Federal de Viçosa.
Palavras-chave
Inteligência competitiva, Mineração textual, Mercado de café
Citação
LIMA JÚNIOR, P. O. Inteligência competitiva na cafeicultura: mineração textual em notícias publicadas na web. 2016. 221 f. Tese (Doutorado em Administração) - Universidade Federal de Viçosa, Viçosa-MG. 2016.