Schema Formalism for Semantic Summary Based on Labeled Graph from Heterogeneous Data

Graphs are used in various applications and to model real world objects. To understand the underlying characteristics of large graphs, graph summarization becomes a hot topic aiming to facilitate the identification of structure and meaning in data. The problem of graph summarization has been studied in the literature and many approaches for static contexts are proposed to summarize the graph in terms of its communities. These approaches typically produce groupings of nodes which satisfy or approximate some optimization function. Nevertheless, they fail to characterize the subgraphs and do not summarize both the structure and the content in the same approach. Existing approaches are only suitable for a static context, and do not offer direct dynamic counterparts. This means that there is no framework that provides summarization of mixed-source and information with the goal of creating a dynamic, syntactic, and semantic data summary. In this paper, the main contribution relies on summarizing data into a single graph model for heterogeneous sources. It's a schema-driven approach based on labeled graph. Our approach allows also to link the graph model to the relevant domain knowledge to find relevant concepts to provide meaningful and concise summary. After extracting relevant domain, we provide a personalized visualization model capable of summarize graphically both the structure and the content of the data from databases, devices, and sensors to reduce cognitive barriers related to the complexity of the information and its interpretation. We illustrate this approach through a case study on the use of E-health domain.