Um método de reutilização do processo da descoberta de conhecimento em base de dados aplicado no setor agrícola

Abstract

The current dissertation was developed based on the methodology proposed by Fayyad, Piatetsky-saphiro, and Smyth, which consists of the extraction of implicit informa- tion, previously unknown and potentially useful, from stored data. From that assumption, we developed the current work, where we present proof of concept, with the primary ob- jective of speeding up the process of knowledge discovery using data from the agricultural sector from the reuse of task from the process. This process uses a set of data that most of the time requires adjustments or configurations during the execution of the knowledge discovery in databases (KDD) method, which creates a problem given the repetition of the settings in similar processes. This task turns the activity exhaustive for the domain special- ist, by having to repeat the entire configuration in a new method, spending the time that could be leveraged using a pre-execution learning knowledge — a guide to reuse of the knowledge discovery process in a database containing necessary settings for executing an initial pre-process, used on rice cultivation. In conclusion, in taking advantage of the tests performed in this culture have shown positive results in the bean crop, proving the premise of the process reuse knowledge discovery in other cultures. In the performed experiments, the method based on the knowledge of learning, had a relative gain of the duration of the process execution by 42.37 % more agile compared to the execution of the executed DCBD process. The tool WEKA was the one used for the execution of the process, using, as in the referred earlier to the methodology proposed by Fayyad, Piatetsky-Shapiro, and Smyth as an extremely fundamental theoretical reference.