基于时间片划分和多元数据融合的异质媒体网络社会事件发现

为了能在大规模、多异质的网络环境下进行网络社会事件的有效检测,提出了一种基于时间片划分和多元数据融合的异质媒体网络社会事件发现方法.该方法首先采用时间片划分的方法,结合用户信息和时间信息来建立用户-时间(user-time,UT)数据模型以减小数据规模;然后通过多元数据线性叠加来整合不同元数据间的相似度并用基于密度的算法以完成社会事件的发现.在最新的SED 2014数据集上进行对比,实验结果表明:该方法与现有方法相比,具有数据处理速度快、事件发现准确率高的优点.

Time-slice and Multi Metadata Fusion for Multimedia Social Event Detection

To detect network event efficientlyin the large scaled and heterogeneous network environment, this research puts forward a new approach based on time-slice and multi metadata fusion for Multimedia Social Event Detection. Firstly, a user-time model by time-slicing with user information was constructed to reduce the scale of data. Secondly, the multi metadata fusion method and density-based clustering ( DBSCAN) algorithm were applied to detect social events. The comparison experiments of the latest dataset-SED2014 indicate that the new approach is faster and more accurate to detect the network social event compared with the existing methods.