Matrix Decomposition Recommendation Algorithm Research Based On Forgetting Function And Multiattribute Feature Extraction

Complex network research has penetrated into different fields such as mathematics,life science,and engineering science,etc.It is one of the most extensive research topics in the study of complex networks,which is aimed at solving information overload and exploring people's hobbies and other aspects.In recent years,a variety of recommendation algorithms have emerged,and collaborative filtering algorithm is the most popular one.However,there are still some problems such as cold start,sparsity and timeliness,etc.On the basis of traditional collaborative filtering,the integration of matrix decomposition technology can greatly reduce the computational complexity and solve the problem of cold start-up and sparsity.In this paper,the matrix decomposition is taken as the main research content which can be combined with forgetting function and multi-attribute feature.Forgetting function is incorporated into the historical score,which makes the existing scoring information more efficient,and then the user feature attributes are applied to find similar users.Finally,an optimization model is established for matrix decomposition to give users better recommendations.(1)Matrix decomposition recommendation algorithm based on forgetting function and user attribute feature extraction.Because of the influence on the recommendation results,we take time factor into account.The new forgetting function is fitted according to the forgetting curve of people,and the original score is smoothed.Secondly,the reasonable application of the user's attributes can make recommendations more precisely.The user attributes will be used to list the relationships among users.It is shown that the relationship between user and item is represented by the score between them.Finally,time information and attribute information are integrated into the newly constructed optimization model of matrix decomposition.The convergence of the algorithm is proved in theory and verified to be of high recommendation accuracy comparing with the existing algorithms.(2)Research on matrix decomposition recommendation algorithm based on time attenuation weight.A continuous function is constructed according to the time-varyingcharacteristics of users' interests in the project,and the scores of different time are fitted.We also consider that some people interests in the project change slowly over time,so we keep the original rating matrix in the recommendation process.Finally,a new matrix decomposition optimization model is constructed by considering the score matrix combining time information and the original score matrix.Compared with the classical matrix decomposition algorithms,the effectiveness of the proposed algorithm is verified based on the recommended evaluation indexes.