机构地区: 北京邮电大学
出 处: 《无线电工程》 2006年第3期16-19,共4页
摘 要: 将用户的需求抽象为可表示、可量化、可感知的特征是未来移动业务的重要特点,用户偏好提取算法是探索这一问题的关键。分析了用户偏好提取算法的数学结构、技术特点、算法类型及研究面临的挑战。针对异构网络环境下移动用户的业务需求特点,提出将传统用户偏好提取技术与马尔可夫决策过程建模方法相结合,创建用户偏好评估模型。解决动态判决环境下基于不完整信息的智能判决问题。对研究用户体验的评价问题和业务与业务环境的适配问题提供了新的思路。 The user preference extraction is a key of the development of expression, quantzation aria awareness of me user requirements in the future mobile service. The paper analyses the mathematical structure, technical feature, algorithm type and challenges of user preference extraction algorithm. For the service requirement features of mobile user under the heterotypie network environment, we propose a new model for evaluating user preference by combining traditional preference methods with MDP (Markov Decision Process) to realize the decision scenario with dynamic and incomplete information. The model proposed is a new attempt, it provides the thinking for the research of user experience evaluation and adaptability of between service and environment.
关 键 词: 偏好提取 马尔可夫决策过程 效用理论 智能学习
领 域: [自动化与计算机技术]