DEEP LEARNING-BASED TARGET ADVERTORIAL GENERATING METHOD AND APPARATUS

摘要

A deep learning-based target advertorial generating method and apparatus. The method comprises: receiving related information of a target object, matching several adapted target headlines from a headline library according to the related information, the headlines in the headline library being expanded from acquired headlines by means of a third generation model (S1); inputting the target headline into the first generation model to generate at least one target introduction (S2); generating at least one piece of input information that conforms to a preset structure according to the related information and a preset rule, and inputting the input information into a second generation model to generate at least one target body (S3); and assembling the target headline, the target introduction, and the target body to obtain multiple target advertorials (S4). By using deep learning and natural language processing technology, the automatic, intelligent and diversified generation of marketing advertorials can be implemented, the investment of operators is reduced, the production efficiency of marketing advertorials is improved, the problem of low handwriting efficiency is effectively avoided, and the problem of dull template generation is also avoided.