基于反馈感知注意力机制的互联网电视智能推荐平台设计与实现  

Design and Implementation of an Internet TV Intelligent Recommendation Platform Based on Feedback-aware Attention Mechanism

作　　者：李鸣 肖云 曾泽基 高明青 郁延书 王学明 Li Ming;Xiao Yun;Zeng Zeji;Gao Mingqing;Yu Yanshu;Wang Xueming(Future TV Co.,Ltd.,Tianjin 300308,China)

机构地区：[1]未来电视有限公司,天津300308

出　　处：《广播与电视技术》2025年第4期21-24,共4页Radio & TV Broadcast Engineering

摘　　要：本文探讨了在智能电视广泛普及的背景下,如何有效应对网络视听内容井喷式增长带来的用户选择困境。针对现有媒体智能推荐服务在内容安全、舆论引导及传播能力上的不足,论文提出运用深度神经网络与注意力机制等先进算法,深入挖掘家庭用户的观影行为与特征,以实现对用户画像的精准描绘,为用户打造更加安全、精准和个性化的新视听体验,更准确地把握用户需求,进而优化推荐算法,提升推荐系统的效率和精准度。This paper explores how to effectively address the dilemma of user choice brought about by the exponential growth of online audio-visual content against the backdrop of the widespread popularity of smart TVs.Addressing the deficiencies of existing media intelligent recommendation services in content security,public opinion guidance,and communication capabilities,this paper proposes the application of advanced algorithms such as deep neural networks and attention mechanisms to deeply mine the viewing behaviors and characteristics of household users.This aims to achieve a precise depiction of user personas,thereby creating a safer,more precise,and personalized new audio-visual experience for users.By more accurately grasping user needs,the recommendation algorithm can be further optimized,enhancing the efficiency and accuracy of the recommendation system.

关 键 词：反馈感知 注意力机制 深度神经网络 互联网电视 智能推荐 

分 类 号：G220.7[文化科学]