Avaliação do desempenho do algoritmo ML-kNN em classificação de textos livres de atividades econômicas

Automatic text classification is still a challenging in the literature, specially for multi-label classification. In this work we evaluate the performance of the Multi-Label k-Nearest Neighbor algorithm for a multi-labeled dataset with more than 1,000 possible labels to be assigned to each one of the documents in the dataset. The results are promising.