Open AccessClassifying Descriptions of Goods with Artificial Neural Networks
Author(s) -
Vinícius Di Oliveira,
Marcelo Ladeira
Publication year - 2019
Publication title -
revista singular
Language(s) - English
Resource type - Journals
ISSN - 2596-1322
DOI - 10.33911/singular-etg.v1i1.19
Subject(s) - artificial neural network , computer science , artificial intelligence , sample (material) , machine learning , chromatography , chemistry
The present study aims to evaluate the performance of an artificial neural network in the classification of merchandise descriptions indicated in electronic bills, legal document used to record all commercial transactions in Brazil. For this, a significant sample of the actual descriptions will be used as well as a overlook about the performance of the neural network with a KNN and a GBM algorithms forecasting the category of the merchandise each description refers. This paper brings a method for classifying descriptions of goods with Artificial Neural Networks. The descriptions are small non structured texts, maximum of 120 characters, relating to goods traded in commercial transactions.