Research And Implementation Of Personalized Recommendation System Based On Logistics Data

With the rapid development and maturity of the Internet,online shopping has become one of the main shopping methods in the modern.As a result,the users and the goods are increasing sharply and resulting in a mass of logistics data.In the background of big data,the unequal reason of the logistics resources and user demand information leads to many issues.They include the higher logistics network transportation costs,unreasonable logistics resource scheduling and not timely business logistics decision-making and other issues.The traditional logistics data processing model can not accurately predict the user needs and the company will not be able to do a good logistics plan in advance.However,personalized recommendation can mine the user's preference information from logistics data by data mining technology,and find the user interested the goods according to the similarity calculation to achieve the user's accurate recommendation and provide effective data support for enterprise decision-making.At present,personalized recommendation system mainly consider the following questions: Firstly,it considers how to mine user information in the massive data to reflect the user's real preference information.Secondly,it considers how to use the preferred data set to train effective model.Thirdly,it considers choosing the appropriate recommendation algorithm.In order to study and realize the personalized recommendation system based on logistics data,this thesis focuses on the research and development of the recommended book system based on the book sales system of the online bookstore,and solves the issues of data sparsity,cold start and scalability in the traditional book recommendation system.The thesis proposes a domain feature-aware recommendation method based on collaborative filtering(DFAR).We use the feature values of the items that imply the user's preferences to extract the user's information indirectly.We use the existing tools to automatically extract the domain characteristics of the goods,and optimize user preference model through the multi-attribute decision method Analytic Hierarchy Process(AHP).Finally,the user preference model is combined with the collaborative filtering algorithm to generate recommendation results.Simulation results show that our method can effectively extract the domain features,relieve data sparsity and cold start problems,and improve the accuracy of the proposed method.Meanwhile,we combined with the Hadoop platform.we study personalized recommendation system based on the Hadoop platform.Facing the massive data,a distributed parallel web crawler is realized by using MapReduce parallel computing framework.At the same time,the time cost of traditional recommendation algorithm is too high in data processing and computation,so we implement a parallel DFAR recommendation method based on Hadoop,which greatly improves the efficiency of the algorithm to meet the needs of the users.Finally,we combine the actual application scenarios with analysis.We analyze the performance of the parallel DFAR recommendation method based on Hadoop,and design and implement an independent book recommendation system model by Java Web development technology.