Abstract: The information interaction between huge heterogeneous entities existed in Internet of Things（IoT） leads to the internal contradiction for IoT. To solve this problem, a method of multi-agent-based inter-objects dynamic relation discovery is proposed. Firstly, abstract all entities in IoT to corresponding Agents. Secondly, employ ontology-based semantic matching and improved calculation method of inter-objects dynamic relation to calculate object dynamic relation for the individual. Finally, converse the original heterogeneous entity relation into the close-ranked dynamic network relation between Agents. The experimental verification reveals that the relation network constructed by the proposed method can obviously improve the interaction contradictions among heterogeneous entities. And compared with the original similar method, both accuracy and error rate have been improved, confirming the proposed method correct and feasible.

Key words: Internet of things, heterogeneous relation, entity, Agent, semantic matching, dynamic relation

摘要： 物联网中存在大量异构关系的实体，其信息间的交互造成了物联网的内在矛盾。针对这一问题，提出将物联网中实体抽象化为对应的Agent，并以个体人为中心，利用本体的语义匹配及改进的物间动态关系计算方法，计算出针对个体人的物间动态关系；将原本异构的实体关系转化为Agent间基于动态关系紧密度排序的网络关系。经实验验证，该方法构建的关系网络可明显改善异构实体间交互的矛盾，而且相比原有类似方法，该方法解决问题的准确率和误差率均有改善，说明了该方法的准确性和可行性。

关键词: 物联网, 异构关系, 实体, Agent, 语义匹配, 动态关系