| With the rapid development of Internet technology,E-commerce has come into a booming time.Because of the big potential business opportunities,all walks of life pay more attention on the e-commerce.However,due to the rapid growth of data volume in the e-commerce,the serious problems about data overload and data redundancy come out,which results that users cannot find the goods they need in the complicated data.This brings great challenges and big troubles to both merchants and users.To solve this problem,personalized recommendation technology came out and plays an active role in solving these problems.As a major component of e-commerce,there are also some problems about data overload and data redundancy in food industry.It is a big challenge for food industry to make appropriate recommendations for users in complex information.To solve this problem,this paper summarizes the characteristics of each algorithm by researching on popular recoumendation algorithms firstly.Secondly,this paper designs and implements a recommendation algorithm according to the data structure and the demands of food industry.At last,according to the business demands of the food industry,a personalized recommendation system was constructed.The main contents of this paper are as follows:1.To guarantee the performance of the recommendation system and prevent the sparse rating matrix,which results from the lack of the rating data,the paper use item-based collaborative filtering and adopt pre-score model to predict the food rating from the way user browsing and adding food to shopping cart.2.The recommended algorithm includes the improved Item-based CF and the implicit semantic model algorithm.The improved Item-based CF is the first and key step in the whole process,which fills the rating matrix and the similarity matrix.And this paper uses attenuation function of time and place to reflect user preferences.The implicit semantic model algorithm filters preliminary recommendation set by statistics. |
PDF Full Text Request