Label recommendation method and system based on stacked noise reduction automatic encoders

The invention discloses a label recommendation method and system based on a stacked noise reduction automatic encoder, and the method comprises the following steps: obtaining a historical behavior record of a user, and obtaining a user identifier, a product identifier and a label identifier according to the historical behavior record; performing noise destruction on the tag identifier, obtaining tag information, and obtaining user preferences and project features according to the user identifier and the project identifier; modeling an interaction relationship of the entities according to the user preferences, the project features and the label information, and performing dimension reduction processing on the interaction relationship of the entities; different coding layers are stacked according to a coding assembly of the automatic encoder to extract feature representations of different dimensions, and the feature representations of different dimensions are mapped to a deep fusion feature space; reconstructin