基于变分自编码器的商品文本分类算法设计

Design of Commodity Text Classification Algorithm Based on Variational Auto-encoder

智能商品分类作为电商平台的重要组成部分越来越受到关注。以商品标题文本数据为研究对象,首先对现有模型中存在的问题进行了概述,其次对变分自编码器相关算法进行了介绍,最后为弥补商品标题文本数据中存在的不均衡问题,提出了一种基于变分自编码器的商品文本分类算法。在公开的商品标题分类数据集上进行的实验结果证明了提出算法的有效性。

As an important part of e-commerce platform, intelligent commodity classification has attracted more and more attention.Taking commodity title text data as the research object, this paper firstly summarizes the problems in the existing model, then introduces the relevant algorithms of variational auto-encoder, and finally proposes a commodity text classification algorithm based on variational auto-encoder to make up for the imbalance in commodity title text data. The experimental results on the open commodity title classification dataset show the effectiveness of the proposed algorithm.