Collaborative Filtering Algorithm Based On Symbolic Data Analysis And Its Application In Logistics Field

Along with the popularization and rapid development of the Internet technology, the number of users and the category of resources increased gradually, which caused the Internet information overload problems, recommend system can handle the problem effectively through fetching user behavior data. Through observing and recording user behavior or rate given, recommend system obtain the result by using the filter technology. Collaborative filtering recommend items to target users based on the ratings given by the nearest neighbors of the target users, Collaborative filtering is the most widely used and the most successful recommendation algorithm, Amazon, CDNow and MovieFinder all use the collaborative filter technology to improve service quality.But because of the users and resources'data quantity is too large, the sparsity of evaluation matrix is becoming more and more serious, which poses some key challenges for recommendation quality. Singular value decomposition (SVD) is a very important matrix decomposition technique, and a dimension reduction method in linear algebra. The symbolic data analysis (SDA) is a new data analysis method of processing mass data, with the thought of "data packing", it can reduce the amount of data greatly and at the same time, grasp the characteristics of sample on the whole. The paper put forward an improved collaborative filtering algorithm based on symbolic data analysis, which apply singular value decomposition and symbolic data analysis method to recommended system. The test done In EachMovie database showed us that the algorithm's recommendation quality is obviously superior to the traditional methods when the date is rather sparse.In the progressing course of reform and opening, our economy has made rapid development, together with a large amount of foreign capital enterprise pouring into our country to invest, manufacturing industry's transferring into, the strong support of the country and so on, our logistics industry has been developing rapidly in recent years. But we still have a long way to go to catch up with the developed countries. An important index measuring the development level of the logistics industry is the proportion of logistics cost in GDP. In 2010, the proportion of Chinese logistics total cost in GDP is about 18%, but to the developed countries it is only 10%. In the total cost of logistics, transportation cost accounts for more than half, so transportation cost reduction is the key to reduce logistics cost.The paper aims to study an improved collaborative filtering algorithm based on symbolic data analysis and then attempt to apply it to the logistics field. Try to build a logistics transportation information platform for public use which contains hundreds of thousands of logistics company. The system analyze the user's behavior, according to the requirements of users to recommend them the most likely suitable logistics company. At the same time, through this information platform more enterprise can realize coordination transportation which can reduce logistics cost greatly, so as to reduce the proportion of logistics cost in GDP,promote the fast and efficient development of the logistics industry,and then to stimulate the stable and healthy development of China's economy.