Text processing method and system based on discriminant generation model

The invention discloses a text processing method and system based on a discriminant generative model, and the method comprises the steps: obtaining a theme inputted by a user, and embedding the theme into a generative model through a writing material library knowledge embedding mode, thereby obtaining a preliminarily generated text copywriting; based on the preliminarily generated text copywriting, performing error check on the text copywriting through a reputation reconnaissance mechanism, judging and marking the error and omission of a model generation result, and obtaining error text information; according to error text information obtained by the reputation reconnaissance mechanism, the writing material library is combined to guide the generative model to correct original errors, and an error-corrected text copywriting is generated; performing knowledge extraction on the error-corrected text copywriting and the similar copywriting through an ETL-span algorithm, and performing credibility calculation on a