Research On Customer 's Common Tendency Acquisition Algorithm In Online Shopping

The rapid development of electronic commerce, making online shopping become a trend, but in the era of big data, in the face of vast amounts of commodity information, it’s easy to less the customers’ interest in shopping. The emergence of the recommendation system to a certain extent, solved the problem, and the recommendation algorithm plays an important role in the recommendation system.In this paper, we use the collection of similarity theory, the relevant theories of complex network, and knowledge of link prediction from the perspective of customers. The historical purchase records is used to analyze customer relationship, in order to eliminate the influence of the abnormal customers on the calculation of the similarity, we add the weighted RA.Based on the link prediction methods building the obtaining algorithm of customers’ common tendency and then based on the related customers’ shopping common tendency to gain goods which the target customers are most likely to buy. By comparing and analyzing the results, which are the model and experimental verification, we conclused the feasibility of the algorithm.For the construction of a model mainly adopts the following method:(1) Set operations. By making college student as a customer to buy goods, to get a customer-goods collection, set operation to get customers-commodity relationship matrix.(2) Similarity algorithm. In view of the two cases on whether the customer is the isolated points, the integrated use of cosine similarity coefficient and relative euclidean distance coefficient to solve similar to the degree of customers from both considering the difference of the sample data change rule in also considering the difference of the numerical value of the sample data, which is more accurate to solve the customer’s similarity. In order to eliminate the influence of mainstream customer to computing similarity, on the basis of general similarity degree of value as the calculation of weight weighted RA, using pajek to build similar relationship network of customer.(3) The common tendency of acquisition algorithm. For the different numbers of similar customers, we use different methods to recommend separately, based on the related customers’ shopping common tendency to gain goods which the target customers are most likely to buy. The model results Experimental verification, the feasibility of the algorithm.