| With the rapid development of internet-based home decoration,more and more interior designers are uploading their design solutions to online trading platforms,providing users with more choices and convenience.However,as the number of design solutions increases,users find it difficult to quickly find their favorite designs among the massive amount of information.To enhance industry competitiveness,an increasing number of interior design solution trading platforms(hereinafter referred to as "platforms")have begun to provide personalized recommendation services to users through recommendation algorithms,which not only improve user experience but also increase user retention rates and transaction volumes of design solutions.Currently,content-based recommendation algorithms and collaborative filtering algorithms are common due to their simplicity and versatility,making them widely used on platforms.However,traditional content recommendation algorithms suffer from monotonous results,while collaborative filtering algorithms face challenges such as data sparsity,cold starts,and gray sheep.These issues require effective solutions.This thesis proposes a recommendation scheme for interior design solution platforms based on a combination of content and collaborative filtering algorithms.This algorithm not only incorporates user behavior and historical interest data but also considers the similarity between design solutions,enabling more accurate recommendations for users that meet their needs and interests.Furthermore,this thesis introduces a user interest model to capture changes in user interests and a project popularity penalty algorithm to reduce the impact of overly popular solutions,further optimizing the personalization and final recommendation results of the algorithm.Through experiments,it is concluded that the hybrid recommendation algorithm proposed in this thesis has better accuracy,coverage,and recall compared to traditional content or collaborative filtering recommendation algorithms.As a result,users are more satisfied with the recommendations and exhibit higher loyalty.This research provides an effective solution for the recommendation systems of interior design solution platforms and offers a reference for the application of hybrid recommendation algorithms in other fields. |
