Design And Implementation Of Personalized Gifts Recommendation System Based On Classification Domain

With the rapid development of internet, many people tend to choose gifts through gift web sites rather than go to gift stores. As there are many choices presented in gift web sites, users are facing an increasingly serious problem of information overload. Therefore, applying personalized recommendation service in gift web sites to help users make decisions quickly become an urgent problem.Personalized recommendation service applied in gift websites is a novel problem. Firstly, this paper introduces several common personalized recommendation algorithms, and then based on the characteristics of gift recommendations, designs and implements a personalized gifts recommendation system based on classification domain, puts forth two realization ways:recommendation based on user behavior and recommendation based on classes'relationship. This paper carries out the requirement analysis, system design, system implementation and system tests, analyzes the key issues in system and proposes practical solutions.This paper mainly achieves:(1) Using Java and regular expressions to capture gifts information from 20 classes of Taobao.com and saves the information in database; (2) Designs and implements personalized gifts recommendation based on user behavior. The implementation steps are:firstly records the users'behavior when they use the system and saves the behavior information in database, then extracts users'potential interest on various classes based on users'behavior, finally recommends gifts based on users' interest values. Specifically, a user's behavior means which classes the user clicks recently, the more often a class is clicked, the more user interested in it, thus the system should recommend more gifts in this class. (3) Designs and implements personalized gifts recommendation based on classes' relationship. Personalized gifts recommendation based on classes'relationship means provide personalized gifts recommendation to users according to the relationship between two classes, it based on the following thoughts:when user views a gift class, the system recommend gifts in classes which are more closely related to this class. The gifts classes'relationship information is saved in ClassRelate database. Information saved in the database is treated and is in more advanced form, conducive to generate accurate recommendation results.