针对在普适环境中系统实施自适应行为的问题,提出从实体粒度上去建模感知、推理以及作用能力的方法.该方法通过抽象封装物理和信息实体,构造行为实体,能够自主感知和融合行为上下文,建模多种用户行为(包括顺序、并发行为),同时建模行为与环境间的制约关系,通过运用Pe-tri网建模用户行为的变迁过程,计算当前行为实体的可中断度,来验证当前环境能否支持行为实体的发生,进而指导服务的自适应实施.将该方法应用于老年人用药辅助场景,设计及实现原型系统,并应用于真实智能家居环境下进行部署与测试.用户体验结果显示,系统能够提供自然的人机交互,保障用药安全,提高老年患者的合理用药水平.%For providing the adaptive assistances to determine when and how to make these assistances available according to the surroundings, the entities are built to model the diverse user behaviors (e. g. , sequential and concurrent behaviors) by integrating the capabilities on sensing, inferring and implementing. By computing the entities' interruption degree based on Petri Net, it is practicable to analyze whether an entity can act under the current environment or not, that means to decide whether an effective assistance can be triggered. We introduce the method in building a smart medication system which provides continuous medication monitoring and contextaware reminders. The experimental results show that our system could improve the medication adherence of elderly in a natural way.
展开▼