中文摘要：

提出了实体状态预测方法，基于传感器采集的实体状态原始数据，预测实体在用户搜索时刻的状态，设计了实体匹配估计方法，依据实体的预测状态对实体的匹配状态进行分类，并估计其与搜索需求的匹配概率，返回匹配概率较高的实体作为搜索结果．结果表明，所提机制在物联网中基于内容的实体搜索的查全率与查准率方面均有较大的性能增益．

英文摘要：

An entity state prediction method based on the raw entity state data observed by the attached sensor was proposed to predict the state of entities at the query time. Moreover, an entity matching esti- mation approach was designed to classify these entities and estimate their matching probabilities with the search needs based on the state of entities predicted. The entities that own higher matching probabilities will be returned as the search results. Numerical results show that the search mechanism proposed can achieve a better recall ratio and precision performance in terms of content-based entity search in the Internet of Things.