Research And Application Of The Big Data Analysis In Logistics Information

With the development of economy and the popularization of Internet, largenumber of industries such as logistics industry have faced both opportunities andchallenges which are caused by the explosive growth of data. In order to improvelogistics efficiency and reduce logistics cost, It’s a significant work to extractpotential and meaningful information from the numerous and complex logisticsinformation in the field of logistics informatization.In this paper, we propose a novel method to deal with the problem of logisticsinformation. We extract the key attributes from the short text data and receive thestructured data by analyzing and processing the unstructured short text messages inthe logistics information system. Then we design two optimal path algorithms basedon BFS(breadth first search) and dynamic programming to achieve the informationpush service in logistics industry. In order to improve the efficiency of the system, westudy the parallel processing technology on GPU and achieve the information pushservice on CUDA. The research we have done not only make contributions toaccelerate the algorithm and reduce the runtime of the algorithm but also help toimprove the efficiency of the algorithm and enhance the availability of the system.The main work is as follows:(1) In this paper, we study the unstructured short text messages in the logisticsinformation system by using the heuristics algorithm which is based on thecharacteristic of the messages and the idea of divide and conquer. After analyzing thedata in the logistics information system, we conclude the characteristic of the dataaccording to the context and create grammar to describe the logistics messages as aformal language. Then we segment the logistics messages and extract the keyattributes and attributes values from the short text data based on the grammar. Themethods we propose transform the unstructured text data to structured data and makedata storage standardized in logistics industry.(2)In this paper, we discuss two algorithm of message matching based on optimalmatching line, namely the optimal matching line algorithm based on BFS (breadthfirst search) and the optimal matching line algorithm based on dynamic programming. The main idea of the first algorithm is to establish the directed graph model accordingto the structured data we get from the previous work. Then we compute the optimalline on the basis of BFS and Pruning strategy. To solve the complex of computingproblem coming with the first algorithm when the data increase, we propose thesecond method: the dynamic programming. We establish the undirected graph modelaccording to the highway information, and then we select the path node which makethe cost lowest based on the idea of dynamic programming. By using two algorithmof optimal matching line, we can compute the service area or the service line of thecustomers which are given the attributes. Then we select the suitable messages whichmatch with the customers feature and achieve the information push service inlogistics industry.(3) In this paper, we study the parallel processing technology on GPU andpropose parallel algorithms to compute the optimal line on CUDA. The algorithmsbased on CUDA can accelerate the speed of the algorithms and then improve theinteractivity of the information push service system in logistics.