Abstract: According to the feature of Internet group events erupted and collective intelligence, a construction of Internet public opinion warning mechanism based on the bottom to top Internet metadata ontology is proposed. The information data in this warning and reliability ontology can be filtered and processed by using Higher Order Singular Value Decomposition（HOSVD） and this method can solve the problem of intelligence prefabricated with massive information in this database. It is observed that the defect of passive information retrieving in the original warning mechanism has been solved by the ontology which is built with users’ feedback in collective intelligence and efficiency of response about mechanism can be increased.

Key words: Internet group event, Internet public opinion, collective intelligence, ontology, Higher Order Singular Value Decomposition（HOSVD）

摘要： 针对网络群体事件的爆发与集体智慧工具的特点，提出利用自下而上的网络元数据建立本体信息库，构建网络群体事件的预警机制模型。采用高阶奇异值分解（HOSVD）实现对预警信息和可信度本体库中数据的过滤与处理，解决依据数据库中的海量信息进行情报预判的问题。以集体智慧中结合用户反馈信息的特点直接构成本体库，能够解决原有预警机制被动式信息检索的问题，有效提高机制的预警反应效率。

关键词: 网络群体事件, 网络舆情, 集体智慧, 本体库, 高阶奇异值分解（HOSVD）