An Empirical Comparison of Portuguese and Multilingual BERT Models for Auto-Classification of NCM Codes in International Trade

Abstract

The classification of goods involved in international trade in Brazil is based on the Mercosur Common Nomenclature (NCM). The classification of these goods represents a real challenge due to the complexity involved in assigning the correct category codes especially considering the legal and fiscal implications of misclassification. This work focuses on the training of a classifier based on Bidirectional En-coder Representations from Transformers (BERT) for the tax classification of goods with NCM codes. In particular, this article presents results from using a specific Portuguese Language tuned BERT model as well results from using a Multilingual BERT. Experimental results justify the use of these models in the classification process and also that the language specific model has a slightly better performance.

A Four-Step Cascade Methodology to Classify MCN Codes Using NLP Techniques

“…Recently, in [de Lima et al 2022] the authors made use of the BERT (Bidirectional Encoder Representations from Transformers) model to train only one classifier that aims to classify descriptions on its respective MCN chapter code. The authors divided their dataset into 96 chapters and focused the classification only inside a single chapter.…”

“…The works of [de Abreu Batista et al 2018] and [de Lima et al 2022] presented a classification of the MCN, considering only the two fist digits, specific chapter only. [Luppes et al 2019] shown a classification considering only the first 4 digits.…”