Design And Implementation Of Mixed Recommendation Method For Fruit Matching Based On User Behavior

Currently,It has been selling hot gradually, that fruits,vegetables and other agricultural and sideline products purchased on the internet. The subject takes fruits as an example to study,mainly around two issues: firstly, how to let users choose the favorite fruits by shopping online quickly; secondly, how to eat more healthy and reasonably. Therefore, to solve the above two issues,this topic launches the research mainly from two aspects: first of all,it is designed the recommended algorithm in order to help users to recommend their potential favorite fruits from the mass of goods; followed by, we develop the reasonable rules which can recommend the fruits collocating with the same effectiveness for users who might like. Finally, it is achieved a fruit matching and hybrid recommendation system, which based on the characteristics of user behavior , to provide users with purchase decisions.In view of the above problems, the main work of this subject includes:?The user behavior characteristics are analyzed and modeled throw the e-commerce platform.Then the weight of the users; interest is calculated by the combination weight method, that the user-commodity interest degree matrix can be generated; ? This paper analyzes the collaborative filtering recommendation algorithm based on user behavior(UBCFREC), improving the similarity calculation method and generating the user-commodity interest degree matrix to calculate the users similarity, that realizes the recommendation to the user's favorite product; ? This thesis puts forward the algorithm based on the rules of the nutritional combination recommendation(NCREC), that divided the recommendation results' sets of UBCFREC algorithm as input,matching the corresponding rule from the rule base and outputting the second new recommendation results;?From the perspective of increasing the sales of fruits and improving the dietary nutrition structure situation of consumers, the UBCFREC algorithm and the NCREC algorithm are combined to realize the hybrid recommendation algorithm. It is designed the APP recommendation algorithm. It is designed the APP of the recommended mall with fruits collocated and mixed, that simulates the realistic application scene of the algorithm.As to the advantages and disadvantages of the algorithm, the experimental environment was set up, and the algorithm testing and evaluation software was developed. The UBCFREC algorithm was designed by using the off-line experiment method. Above all, the experimental results are evaluated from the three aspects of algorithm precision, algorithm efficiency and recommendation accuracy.Eventually, the experimental results show that the proposed algorith-m can effectively improve the recommendation quality of the personalized recommendation system.