Research And Application Of Neural Collaborative Filtering Recommendation Algorithm Based On Context Information Fusion

In the era of big data,the problem of "information overload" has seriously hindered Internet users from quickly obtaining the information they need.Recommendation systems are currently one of the effective methods to solve this problem.In recent years,traditional recommendation algorithms have been unable to meet market demand.In practical recommendation systems,traditional recommendation algorithms are often used in combination with new technologies,such as deep learning techniques widely used in the field of image recognition and context aware computing to mine the user’s context.This paper proposes a neural collaborative filtering recommendation model(TCNeu CF)based on context information fusion,and applies it to the development of personalized movie recommendation systems for movie scenes to help users quickly discover favorite movies and promote the development of recommendation algorithms.The research work of this article is as follows:(1)Select the basic recommendation model.This article selects user history ratings as an implicit feedback behavior and obtains user preferences based on this.In addition,experiments were conducted on the recommendation model based on generalized matrix decomposition,the recommendation model based on multi-layer perceptron,and the hybrid recommendation model combining generalized matrix decomposition and multi-layer perceptron to select the optimal basic recommendation model.The experiments found that the hybrid recommendation model had the best effect under the same conditions.(2)The hybrid recommendation model introduces a time context.Considering that the user’s context can help improve recommendation accuracy,this paper introduces a temporal context based on the hybrid recommendation model.Experiments have found that the improved model significantly outperforms the hybrid recommendation model in terms of recommendation effectiveness.(3)Adjust parameters to improve the recommendation effect.In order to improve the accuracy of recommendation,this paper conducts parameter adjustment experiments on three parameters,namely,learning rate,recommendation length,and iteration number,to determine the optimal parameters.This article uses the Movie Lens dataset to evaluate the effectiveness of the recommended algorithm using two evaluation indicators: Normalized Discounted Cumulative Gain(NDCG)and Hit Rate(HR).Experiments have proven the effectiveness of the proposed neural collaborative filtering recommendation algorithm that fuses context information.In addition,the movie recommendation system developed in this paper has enabled the algorithm to be applied.