Recommendation And Analysis System Of Commodity Purchasability Based On Big Data

With the rapid development and wide application of Internet technology,people's ways of obtaining information and shopping have generally shifted from offline to online,bringing unlimited convenience,and also bringing massive and real-time updates to e-commerce platforms.For similar data,people cannot quickly and accurately find the target product from the overloaded data.At present,the product recommendation system generally recommends based on the existing information provided by the e-commerce platform,but cannot personally recommend from the perspective of consumers.In order to solve the above problems,this article proposes a commodity purchaseability based on big data Recommendation analysis system,which implements personalized recommendations from the consumer's perspective,and merchants can also adjust the sales plan according to the consumer's preferences in the system.This article uses natural language processing technology to analyze the review data of consumers on the e-commerce platform,complete the formulation of the candidate attribute extraction rule table,and extract the evaluation objects and triple relations according to the rule table.The research compares the three clustering algorithms of K-means,Mini Batch k-means,and Birch,uses contour coefficients to determine the clustering effect,proposes to cluster candidate attributes,and verifies that the clustering effect under this clustering condition is better.Use TF-IDF to assist the naming of clustering documents to realize the redefinition of commodity attributes The sentiment dictionary-based method is used to calculate the sentiment value of the product review data,and the overall review favorable rate is calculated based on the sentiment value to realize the recalculation of the product score.Use the corresponding recommendation algorithm to analyze the product attributes and scores to achieve personalized recommendation from the consumer's perspective.By researching the performance of different recommendation algorithms,analyzing the advantages and disadvantages of various recommendation algorithms and application scenarios,a personalized recommendation system including collaborative filtering recommendation,content-based recommendation,and implicit semantic model-based recommendation is designed and implemented.The performance of various recommendation algorithms is used to complement each other to solve the data sparsity and new user problems of collaborative filtering recommendation algorithms,and the poor scalability of model recommendation.Use big data technology for distributed storage and calculation of data,build a data storage platform Hadoop and data processing platform Spark on the server to achieve the calculation of the recommended algorithm.Finally,the Jeeplus development framework is used to design the front-end page display function of the commodity purchaseability recommendation system.