Research On Recommendation Method Based On User Profile

Social media sites,e-commerce sites,and other data have shown explosive growth,and product recommendation methods have emerged.However,in practical application scenarios,how to recommend the right products to the right consumers at the right time is an issue that ecommerce companies have always been concerned about,and user profile can solve this problem well.The current product recommendation methods based on user profile have the following problems: 1.Social network information of users is rarely considered.2.Mostly based on a single tag for profile,often ignore the tag collection based on user interest profile.3.The recommended results of existing methods are often poorly interpretable.Therefore,in response to these problems,this article is to study a recommendation method based on user profile.The specific work mainly includes the following aspects.First,define related concepts,and give the definition and formal description of product recommendation problems based on user profile.Secondly,a user interest profile model based on personal social network is proposed.On this basis,a product recommendation method based on user profile is proposed.The main idea of this method is to calculate the similarity between user profile and products based on the constructed user interest profile,and recommend products with high similarity to users.The method also considers the popularity of the product,obtains the highly popular product by using the product label weight algorithm,and then calculates the similarity between the user interest profile and the highly popular product to complete the product recommendation.The user interest profile model description and construction method UIPSN based on personal social network is given;the idea and implementation steps of the product recommendation method based on user profile are described,and an improvement considering the diversity and novelty of the product is given algorithm.Finally,design and implement the experiment.The experimental results on the Douban dataset and the Netflix dataset prove the feasibility and correctness of the product recommendation method based on user profile.