Research on Multidimensional Algorithm Optimization of Accurate Path of Marxist Education Communication in New Media Environment

With the explosive growth of information resources, recommender systems play a pivotal role in alleviating information overload and have been widely adopted in many services.This paper combines Bayesian personalized ranking method and graph convolutional neural network to construct an accurate recommendation model based on the dissemination of Marxist education information, and realize the multi-dimensional optimization of the recommendation path.Compared with several traditional collaborative filtering recommendation models, this paper's model achieves better results in terms of RMSE, Precision, and coverage indicators, which verifies the effectiveness of this paper's model in information dissemination recommendation.In addition, compared with other six recommendation models such as SVD, Social_MF and CUNE, this paper's model is only slightly worse than the CUNE model in RMSE indexes, and is smaller than other models in RMSE and MAE indexes, with smaller prediction error and higher recommendation accuracy.It shows that the model in this paper can effectively mine the user's interest preferences and the personalized characteristics of the items, and realize the multi-dimensional optimization of the precise path of Marxist education dissemination.