中文摘要：

基于iLBS系统中SP服务器的发布／订阅（pub／sub）中间件技术，提出借助于贝叶斯网络来预测用户行为的一种新颖的pub／sub模型（UBPM）．新模式不仅以用户目前所在位置作为通告的唯一标准，而且考虑了用户环境信息中前后台信息的同步，并使用贝叶斯网络对移动用户的行为做出预测．因此，有效地解决了传统pub／sub系统中病态和冗余消息通告的问题，并提高了消息的精确性．实验结果表明相对于现有预测模型，UBPM预测模型更加有效．

英文摘要：

Based on the pub/sub middle ware of existing SP server in iLBS system, a novel pub/ sub model is presented to predict user behavior using the Bayesian network. The new pub/sub model not only takes users＇ current location as unique criterion of the notification, but also considers users＇ environment information of fore/background synthetically, and uses Bayesian network to predict the future behavior of mobile users, thereby solves abnormity and redundancy notification in traditional pub/sub system effectively and enhances the notifications＇ precision. The result of experimental testing shows that the prediction model of UBPM is more efficient than existing prediction models.