Research On Context-aware Recommendation Method Based On Gradient Boosting Decision Tree

With the maturity of mobile Internet technology,user preferences show more context sensitivity.How to dig deeply into the user's preferences in different contexts and integrate context factors into the information recommendation process to achieve context-aware recommendation has become a hot and difficult issue in the field of information recommendation.Based on this,this paper constructs a method for calculating the context attribute weights based on the gradient boosting decision tree,and on this basis,this paper constructs a context-aware recommendation method based on the gradient boosting decision tree,in order to provide new ideas for the subsequent research on context-aware recommendation.At first,this paper sorts out the research on context-aware recommendation at home and abroad,and points out the problems in the existing research,so as to lay the foundation for the follow-up work of this paper.Then this paper deeply explores the user's contextualized preferences from the perspective of context attribute weight.In view of the existing context attribute weight calculation research,the context attribute weight calculation is regarded as a static process,and the relationship between context attributes is ignored.In this paper,the Boosting algorithm,which is widely used in the field of feature selection,is introduced into the context attribute weight calculation,and this paper builds a method for calculating the context attribute weights based on the gradient boosting decision tree.This method regards context attribute weights as a dynamic feature selection process,and obtains the influence degree of context attributes on user selection based on considering the relationship between context attributes.It provides a reference for the key context attribute recognition and multi-scenario attribute relationship processing problems in the context perception recommendation research.On this basis,this paper through contextual modeling to deeply integrates context information,and proposes a context-aware recommendation method based on gradient boosting decision tree.This method combines the similarity of user ratings and the similarity of the user's information resource category preference in the current context instance to realize the nearest neighbor search in a single context dimension,which al eviates the impact of data sparsity on the calculation of user similarity.Then based on the weight of the nearest neighbor search results and context attribute weights to obtain the recommendation results of fusion context information,to achieve context-aware recommendatio n.Finally,this paper uses the classic Movie-lens dataset to verify,and compares the proposed method with the traditional collaborative filtering recommendation method.The Experimental results prove the effectiveness of the method for calculating context attribute weights and the context-aware recommendation method proposed in this paper.