Aprendizado de máquina e processamento de linguagem natural aplicado na predição de marcas a partir de dados de marketplaces

Abstract

Natural Language Processing, as well as Machine Learning, has been widely used to process and classify data in the search for information. This work aims to propose a machine learning model capable of classifying brands from product descriptions. For this, data collected by Crawlers from the main e-commerce systems in Latin America was used. First, a treatment was performed on the data, which includes removing, filtering and transforming them so that they are ready to be used. NLP was used to vectorize the product description and then three machine learning algorithms were used. This computational prototype was built using Python and significant were results obtained, demonstrating that the algorithm is very promising.