Aplicação do text mining para incorporação de informações sócio-econômicas em sistemas objetivos de previsão de safra.

Abstract

Several academic studies and governmental efforts have been undertaken to predict, with trust, the planted area and the productivity, in the intention of esteem the Brazilian agricultural harvests officially. The official estimate is based on systematic data, for municipal district, with information picked through interviews by rural establishments. However factor as public politics, climatic phenomena, crop management, diseases, pests, and others have to be considered in the global calculation of the harvests. Under this optics, the present study presents the technique of text mining for incorporation economic factors in the process of harvests forecast. These factors were analyzed in the context of journalistic news through the software Eurekha. This tool formed groupings with index of acceptable similarities