A Shopping Guide Text Generation System Based on Deep Neural Network

Abstract

Many people shop on Taobao, Jingdong and other online platforms in China. More and more products are advertised through self-media. Shopping guide text is an effective means to improve the effectiveness of advertising. However, self-media companies need to hire a lot of professional writer to write shopping guide text, which leads to high labor cost. In this paper, we proposed a shopping guide text generator, which can automatically generate shopping guide text given an image of the product. In this paper, we focus on the shopping guide text generation of clothes. The proposed text generator consists of a convolutional neural network, a recurrent neural network with long-short-term-memory (LSTM) over shopping guide text, and a structured module which evaluates the related degree between the image and shopping guide text. The experimental results show that the proposed shopping guide text generation system can generate attractive text to advertise the given clothes.