Research Of Tea Consumers Network Community Detection Based On Improved GN Algorithm

In recently years, tea as a kind of characteristic agricultural product, the scale of the network trade is bigger and bigger. Personalized recommendation, as one of the hotspot research of the current e-commerce, will also be the inevitable trend of the future tea e-commerce.The existing personalized recommendation, the tea is regarded as ordinary commodities to recommend to consumers,overlooking the tea itself characteristics of agricultural products. Also the traditional personalized recommenda--tion mainly for a single user, there are few recommendations based on consumer network community.Community Detection is the precondition of the community recommendation.Firstly analyse the consumers contact between each other, then build consumers network,at last to differentiate different consumers to their communities. Thus to recommend according to community.This topic introduced a complex netw--ork of community detection technology. On the basis of constructing tea consumer network,mainly discussed the GN algorithm of community detection.According to the algorithm time complexity through higher problem, this paper proposes a improved GN algorithm implements accurate classification of tea consumers in the network community.The main content of the thesis includes the following aspects:(1) Method to construct the tea consumer network Subject to traditional collabo-rative filtering "user-project evaluation matrix" is mapped to the consumer in the network, the user as a node, the connection between the user as a boundary, the similarity between users as edge weights of network build tea consumers. In the paper combining with the characteristics of the tea itself, and puts forward a new method of user similarity calculation.(2) From the existing complex network of community found an improved designed the GN algorithm, on the basis of in-depth analysis and research, aiming at the shortcomings of the traditional designed and the GN algorithm complexity is too high. A new algorithm to find the center of the network node first, and then only consider the shortest path between center node and other nodes, so as to calculate each edge by the shortest path to get edge betweenness of the moment, at the same time use the end of the increment as designed.the GN algorithm module degree standards. Through the relevant experiments and simulation comparison of the traditional designed.the GN algorithm with improved designed.the GN algorithm on the accuracy and the running time difference, the experimental results show that the improved designed.the GN algorithm high classification accuracy and greatly reduces the running time.(3) Using the proposed method to construct the tea consumer network of part one.Built a tea consumer network on real date sets.Using the improved GN algorithm to divide the tea consumer network.Then Using software Pajek to visual the process.The result of the topic research provide reference for Subsequent personalized recommendation for the community.