Design And Implementation Of An E-commerce Personalized Recommendation System Based On Collaborative Filtering

With the continuous improvement of mobile Internet technology and mobile phone application software development technology,most smartphone users have higher requirements for mobile business functions,especially the function of using mobile phones for online shopping.With its fast dissemination of commodity information,convenient shopping function,by the majority of users like and use;Secondly,with the rapid development of Internet information technology and e-commerce,human beings have gradually stepped into the era of information overload,"huge and bloated" e-commerce data makes it difficult for users to obtain effective data information in time.Therefore,this paper combines personalized recommendation technology with e-commerce technology to achieve more efficient electricity.The main research work of E-commerce personalized recommendation system is as follows:Firstly,aiming at the problem of low accuracy of feature extraction from personalized recommendation data,this paper proposes a Text Rank algorithm based on comprehensive weight,which uses TF-IDF value,keyword part of speech,keyword position,keyword length and other index data,and so on.The rule of G1 weight method is used to synthesize the weight of keywords,and then the Text Rank algorithm is used to iterate the key word weights to get the final comprehensive weights,which can effectively represent the importance of keywords,and pass the accuracy rate.Recall rate and F1_Score mean to measure Text Rank algorithm based on synthetic weight,traditional Compared with TF-IDF algorithm and Text Rank algorithm,Text Rank based on synthetic weight has better keyword extraction effect.Secondly,in this paper,based on clustering optimization,this paper proposes a co-filtering algorithm based on cluster optimization to optimize the co-filtering recommendation algorithm,and completes the system commodity recommendation function,and effectively solves the problem of online purchase users to find high-quality commodity resources from many commodity information.Based on clustering optimization,the co-filtration recommendation algorithm mainly collects user interest,commodity content,user score and purchase behavior,and uses clustering algorithm to cluster the collected data sample space.The evaluation absolute error is used to compare the user collaborative filtering recommendation algorithm,the content-based collaborative filtering recommendation algorithm and the cluster optimization-based collaborative filtering recommendation algorithm.It is concluded that the cooperative filtering recommendation algorithm based on clustering optimization has better recommendation effect.Finally,the paper uses the object-oriented design method to analyze and design the requirements of the system,using Android Studio development tools,JAVA technology and SQLite database to develop a well-functioning e-commerce system on mobile terminals.The system can realize many common functions of online merchandise trading,such as: commodity information viewing,commodity quick search,popular commodity activities,characteristic commodity market,commodity classification,shopping cart,user registration and login,and so on.The stability and reliability of the system data is good,which provides a more efficient e-commerce system for the vast number of online shopping users,and improves the e-commerce system of the online shopping users.Shopping efficiency.