基于人工智能算法的商品归类研究与应用

Research and Application of Commodity Classification Based on Artificial Intelligence Algorithm

准确高效的商品归类有助于进出口企业提升通关速度、降低通关成本。为帮助进出口企业传统商品归类在效率和准确率等方面实现进一步提升,本文利用企业申报数据,构建了基于双向转换编码器表征(Bidirectional Encoder Representations from Transformers,BERT)与文本卷积神网络(Text Convolutional Neural Network,Text CNN)联合模型的商品归类算法,并完成归类系统开发及验证。以企业实际商品申报数据进行测试,归类准确率达95%以上,取得了较好的应用效果。

Accurate and efficient commodity classification helps import and export enterprises to improve the speed of customs clearance and reduce the cost of customs clearance.To improve efficiency and accuracy in classifying traditional goods for import and export enterprises,this paper uses enterprise declaration data to construct a commodity classification algorithm based on a combined model of Bidirectional Encoder Representations from Transformers(BERT)and Text Convolutional Neural Network(TextCNN),and completes the development and verification of the classification system.According to the test of the actual commodity declaration data of those enterprises,the single accuracy rate is over 95%,achieving good application results.