Abstract: 【Purpose/significance】The number of consumer reviews is huge and full of randomness,so it is necessary to analyze the re⁃view information.Information analysis can provide valuable reference for potential consumers when purchasing corresponding products,can provide consumers' feedback on products to businesses,and can also provide reference for the platform to improve services,so as to achieve a win-win situation among consumers, businesses and the platform.【Method/process】Taking some product reviews texts crawled from Jingdong Mall as the research object,the article extracts the commodity features in the reviews by using the method of word frequency and word cloud analysis,and analyzes the emotional tendency of five different types of commodity reviews through the calculation method of sentiment tendency【. Result/conclusion】The results show that for different types of goods,consumers have differ⁃

ent attribute preferences,and the number of product features that users pay attention to is also different.On this basis,targeted sugges⁃tions are put forward for businesses,e-commerce platforms and consumers

【. Innovation/limitation】With the method of comment min⁃ing,the paper studies from three perspectives of consumers,businesses and platforms,and puts forward the countermeasures.But the da⁃ta processing is still not accurate,and needs further research.