面向空间信息智能分发动态化用户偏好模型研究_李新广.pdfVIP

40 5 Vol.40, No.5 2011 10 Acta Geodaetica et Cartographica Sinica Oct.,2011 :1001-1595(2011) 05-0646-09 李新广, 范明虎, 杜 武 , 430079 Research onDynamic ser Profile Model for Intelligent Distribution of Spatial Information LI Xinguang, FANMinghu, D Wu State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan niversity, Wuhan 430079, China Abstract : ser profile modeling method isthe key bottleneckthat restricts theory of intelligent distributionof spatial information to progress. Existing algorithms and systems of intelligent distribution of spatial information have drawbacks of inaccurate spatial location and biased utility, etc. and are mostly concerned on the contribution of the user. s retrieval behaviorto the profile model, but not considertimefactors at all, and pay little attentionto the role of userfeedback. Inviewof this, the theoriesand algorithms of the existing literatureare extended, by introdu- cing concepts andarithmetic of region number, interest degree, interest degree density, etc., and dynamic factors of weight attenuation function and user information feedback, etc. , to make a model able to adjust more accurately and in time with the change of user preference profile. The experimental results show that, compared withtraditional static models, the model can more effectively reflect the change of user preference profile. Key words : intelligent distribution of spatial information; dynamic user profile model; spatial information service; weight attenuation function; informationfeedback :用户偏好模型的构建方法是制约空间信息智能分发理 取得进展的关键瓶颈现有的空间信息智能分发算法 和系统存在空间范围定位不准确效用度计算存在偏差等缺陷,且大多关注用户的检索行为对偏好模型的贡献却均未考 虑时间因素的影响,也很少注意到用户反馈的作用鉴于此, 对现有文献的理 和算法进行扩展, 通过引入区域数兴趣 度兴趣度密度等概念和算法, 以及权值衰减函数和用户信息反馈等动态化因素, 使模型能够更为准确及时地随着用户 偏好特征的变化进行修正试验表明,相较于传统的静态模型而言,该模型能够更为有效地反映用户偏好特征的变