Descoberta de conhecimento com uso de técnicas de mineração de textos aplicadas em documentos textuais da investigação policial brasileira

Abstract

The aim of this study is to analyze how text mining techniques applied to textual documents of Brazilian police investigation can promote knowledge discovery. The research collected documents from the police investigation and submitted them to the text mining process. The study used the techniques of case folding, tokenization, custom stopwords, bag of words and TF-IDF in order to extract results in ngrams. The results were presented with word clouds. In the research, k-means were used to cluster the sets of trigrams, identifying in each clusters the most representative terms of the clusters. The use of text mining techniques on these documents was intended to extract non-trivial knowledge. The techniques of text mining, or discovery of knowledge in a textual database, have the purpose of discovering unobservable patterns when analyzed by human manipulation of large volumes of documents. The results found favored the discovery of knowledge in the identification of entities and connections, as well as thematic categories of the investigation