A Unified Semantic Model for Cross-Media Events Analysis in Online Social Networks

The gap between the raw data from various data sources and the diverse intelligent applications has been an obstacle in the field of events analysis in online social networks. Most existing analysis systems focus on data from a certain single online social network platform and a limited range of analysis applications. To comprehensively understand events, the sources of data usually include multiple online social network platforms and different existing corpora. Thus, it is necessary to build a bridge to handle the online social data from different sources and support various analysis application requirements. In this paper, a unified semantic model for events analysis is proposed. The model contains well-designed classes and properties to tackle the lack of unified representation, and provenance information is also taken into consideration. The reasoning is supported to check the consistency of the data and to discover hidden knowledge such as tacit classification and implicit relationships. The schema mapping and data transformation methods are provided to handle the heterogeneous data from various online social network platforms and datasets. The design of the cross-media event analysis system is also presented. The comparison shows the advantages of this paper. The case study shows the applicability and effectiveness of our model and system.