Design And Implementation Of Women's Commodity Recommendation System Based On Hadoop

Due to the continuous maturity and rapid development of information and network technologies,the amount of information in the network era has been exploding in various fields(such as TV programs,movies,food,travel,music,books,news,Web pages,etc.).The same is true in the business sector,especially women's wear.As a consumer of information,we may always be troubled by the fact that we are faced with a wide variety of categories and embarrassing goods.Therefore,there is an urgent need for a recommendation system that can customize the mining of similar products for users to help users filter out useless products,improve user satisfaction with the platform,and bring greater profits to the enterprise.The Hadoop-based women's product recommendation system mainly includes timing tasks,offline calculations,data updates,and Web server modules.The scheduled task module is hosted by crontab.The main responsibility is to periodically upload user log information and commodity attribute related information files to HDFS,periodically call the offline computing module to perform calculations,and periodically call the data update module to synchronize data.The offline computing module is an offline computing task.The user log information and product attribute related information of the company Website are the main input data sources.HDFS is used as distributed storage and MapReduce is used as the distributed computing framework.The user customizes the list of recommended products and calculates a list of popular items on the Website,and saves the calculated results to the database.The main function of the data update module is to synchronize the latest data into the database,including data such as recommendation results,popular lists,and user logs.The Web server module is based on the python flask implementation,and mainly provides some external interface service implementation,including the user recommendation list result query interface,the Website popular commodity list query interface,the user log information query interface,the log add message request interface,and the commodity up and down shelf message request interface,etc.The test results show that the system can run stably for a long time,and the recommended results calculated by the user are basically in line with the user's interests and hobbies,and the expected goal can be achieved.It can be used as a system in the e-commerce field to recommend the preferred products for users.