Anti-fact sample generation method, model training method and device

• Main Authors: WANG WEI, LIU YI, WANG YI, ZHOU YU
• Format: Patent
• Language: Chinese; English
• Published: 27.05.2025
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR

The invention provides an anti-fact sample generation method and device and a model training method and device.The method is applied to the technical field of computers and comprises the steps that user feedback data generated by a user for a recommended subject matter is obtained, and a fact sample is generated based on the user feedback data; obtaining a first subject matter utility value corresponding to each subject matter to be recommended, and determining a second subject matter utility value corresponding to the recommended subject matter responded by the user and a third subject matter utility value corresponding to the recommended subject matter not responded by the user from the fact sample; comparing the utility value of the first subject matter with the utility value of the second subject matter and the utility value of the third subject matter, and generating a first anti-fact sample based on the to-be-recommended subject matter meeting a preset utility value condition; and inputting the first an