Deep recommendation model fusing comment timeliness and credible neighbor score influence

The invention discloses a deep recommendation model fusing comment timeliness and credible neighbor score influence. A deep feature extraction module fusing comment timeliness is used for extracting user preference deep features and article deep features from a comment data set; decomposing a scoring matrix for the scoring data set through a shallow feature extraction module fused with credible neighbor scoring influence to obtain a user implicit vector and an article implicit vector, and calculating user scoring information to obtain the credible neighbor scoring influence; fusing the deep feature of the article and the implicit vector of the article to obtain an article feature; fusing the user preference deep feature and the user implicit vector to obtain a user intermediate vector, and fusing the user intermediate vector and the credible neighbor score influence to obtain the user preference; and performing full connection layer processing on the user preferences and the article features to obtain a predi