Artificial intelligence-based methods for harmonized system code translation: A review

Harmonized System (HS) is a list of classifications of goods that are made systematically with the aim of facilitating discretion, trade transactions, transportation and statistics. The HS code is determined by certain exporters / importers based on the description of the goods. Although it speeds up the customs clearance process by standardizing traded goods, it remains difficult to effectively have proper HS codes for each traded item. It is an extremely difficult task to provide accurate customs identification. This identification system remains complicated for non-experts. Importers / exporters and customs can determine different HS Codes on the same goods description. This happens because of differences in interpretation of goods descriptions. This research conducted a survey of several studies that have been conducted to translate HS codes using intelligent systems. Several studies have developed applications for determining the HS Code by requiring human intervention. Other studies have successfully developed a model of classification of goods based on HS codes. The input used varies between text, image and a combination of both. Based on the evaluation of the obtained model performance, the combination of text and image provides the best performance, with the highest f-score.