Social recommendation system performance optimization method based on social importance and project category information

• Main Authors: LIU ZHIYUAN, MA YINGHONG, HU BIN
• Format: Patent
• Language: Chinese; English
• Published: 19.07.2024
• 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 relates to a social recommendation system performance optimization method based on social importance and item category information, and the method is realized based on a graph neural network. A similar user model, a social importance model and a project type information model of cold start users are established, user-project global information is fused through the models, after fusion, non-interactive projects are predicted and scored through a prediction model, and finally project recommendation is completed for the users. According to the method, preference embedding representation of a common user and a cold start user is remarkably enhanced, semantic representation of the user is enriched, and richer semantic connotation is provided for the user and a project. 本申请涉及一种基于社交重要性和项目种类信息的社交推荐系统性能优化方法，本方法基于图神经网络实现，在用户社交网络图SG、用户-项目图UIG和项目-种类图ICG的基础上，建立冷启动用户的相似用户模型、社交重要性模型、项目种类信息模型，通过上述模型对用户-项目全局信息进行融合，融合后通过预测模型对未交互项目进行预测评分，最终对用户完成项目推荐。本申请显著增强了普通用户和冷启动用户的偏好嵌入表示，丰富了用户的语义表征，为用户和项目提供了更加丰富的语义内涵。