Commodity Text Classification Based E-commerce Category and Attribute Mining

This paper proposes a commodity text classification-based e-commerce category and attribute mining method. The fastText is utilized to classify the commodity text data and establish the first-level and leaf category mapping between two different e-commerce platforms. Based on the category mapping, the system can mine the attributes and attributes' values that the target leaf category does not have, while its corresponding leaf category from another platform does. Additionally, the longest common subsequence (LCS) algorithm is utilized to calculate the attribute similarity. The experiments on two different e-commerce platforms show that the proposed method not only can help the e-commerce platform establish and improve the category architecture but also enhances user experience.