Data mining e processamento digital de sinais para a análise de estruturas poéticas

Abstract

This work aims to validate the application of digital signal processing in literary structures. In this way, the analysis is validated based on the principles of data mining for the mathematical understanding of poems in the Portuguese language. Pattern recognition is a rapidly developing field that supports studies in areas such as the search for metrics from the rhyme obtained in poetic texts. The validation of such characteristic is accomplished with the application of autocorrelation as a mathematical tool. From this analysis, it is assumed that the higher values of autocorrelation represent the greater similarity of the text. This tool enables the interpretation of signals as we search for patterns of repetition in poems. Thus, the best result of the analysis considers only the end of each verse, where it is assumed that there is rhyme. In these cases, the highest values of autocorrelation are obtained, although these values do not present significant relevance for the standardization of the text.