关键词: 跨境电商, 文本挖掘, 在线评论, 情感分析, 消费者需求

Abstract: In the history of bilateral trade in China and Russia, Russia has always been an important export market in China. Because of the geographical location of Russia, its people have high demand for winter clothing products. In this context, Chinese suppliers provide more targeted high quality winter clothing for the Russian clothing market, which will further promote the development of bilateral clothing trade. However, in the field of domestic consumer demand research, most of the research targets are domestic consumers, and foreign consumers account for only a few.
In order to promote the development of the foreign trade clothing industry in my country and Russia, and effectively guide domestic companies to enter the reverse custom production model, this article uses the Russian e-commerce platform Wildberries as the source of data to obtain online comments from down jacket clothing products. After preprocessing online comments, use the method of combining high frequency vocabulary extraction and cluster algorithm extracts consumer evaluation dimensions from it, obtains three first level evaluation dimensions, and ten second level evaluation dimensions; during the analysis of consumer demand, sentiment analysis is conducted on the manually annotated online comment dataset through three indicators: consumer evaluation dimension weight, emotional positivity rate, and progress to be improved, ultimately obtaining feedback information from clothing consumers. The results indicate that product quality has the greatest impact on the overall emotional orientation of consumers in the primary dimension, followed by product appearance and service quality. In terms of product quality, although warmth retention is the most important clothing characteristic of down jackets, it ranks only third in the overall emotional tendency of consumers; The top ranking is the fit, which is also the most in need of improvement. Among the other secondary dimensions, color and customer service have the greatest impact on the overall emotional orientation of consumers, both of which need further improvement. In the future, China's foreign trade clothing enterprises should not only increase corresponding sizes in the design and production process, but also consider consumer preferences for colors and warmth needs. Sales parties should pay attention to improving customer service quality in the service provision process to enhance consumer satisfaction, ultimately promoting domestic enterprises to enter the reverse customization production mode and promoting the development of bilateral trade.
In future research, in order to use data more effectively, deep learning technology can be used to automatically analyze online reviews, efficiently and accurately extract keywords and themes, and identify consumer preferences and feedback; Incorporate it into the training sample, you can better understand the all characteristics of the demand for down jacket consumers.

Key words: cross-border e-commerce, text mining, online review, emotion analysis, consumer demand