电子商务用户反馈文本数据的可视化研究与实现

Research and Implementation of Visualization of Text Data Fed Back by E-commerce Users

中文关键词: 可视化设计  用户反馈数据  电商平台  语义分析

英文关键词: visual design  user feedback data  e-commerce platform  semantic analysis

中文摘要:

      目的 针对数以万计的电子商务交易量背后产生的海量评价数据，给用户带来的信息过载、检索低效等问题，探索合理的可视化方案，辅助用户进行购买决策，帮助商家提取有效反馈以帮助产品优化。方法 以京东平台商品的评论数据为例，提出一种电子商务用户反馈文本可视化方法，以口红为目标商品，采用网络爬虫技术从电商平台抓取用户评论文本数据并进行整理分析，针对文本数据采取自然语言处理技术进行语义分析并归类，得出可直接用于可视化的文本数据，进一步提取了口红色号、价格趋势、关注词频率三个评论数据种类，价格对比、口红色号可视化等用户指标，借助Tableau可视化工具对数据进行映射，建立了趋势图和分布图，得到直观的可视化图表，为用户提供了更美观且具有交互功能的信息浏览方式。结论 所提出的可视化方案采用视觉认知的方式提取关键信息，能有效降低用户浏览文本信息的时间成本，此方案的评估结果也获得了较高的评分，对推进电商平台数据可视化具有一定指导意义。

英文摘要:

      The work aims to explore reasonable visualization schemes to assist users in making purchasing decisions and help merchants extract effective feedback for product optimization, with respect to such problems as overloaded information and inefficient retrieval brought to users by huge amounts of evaluation data generated by tens of millions of e-commerce transactions. With the comment data of the products of Jingdong platform as an example, a method for visualizing the feedback text of e-commerce users was proposed. Taking the lipstick products as the target goods, the web crawler technology was used to capture the user’s comment text data from the e-commerce platform for their analysis. The text data were semantically analyzed and classified by natural language processing technology, and the text data directly used for visualization were obtained. Three kinds of comment data (lipstick color code, price trend and frequency of attention word), price comparison, lipstick color code visualization and other user indicator were extracted. Tableau visualization tools were used to map data, create trend and distribution maps, and get intuitive visual charts to provide users with more beautiful and interactive information browsing. The proposed visualization scheme can extract key information and effectively reduce the time cost for users to browse text information by means of visual cognition. The evaluation result of this scheme also obtains a high score. The proposed scheme has certain guiding significance for advancing the data visualization of e-commerce platforms.