Holistic Statistical Open Data integration based on integer linear programming

Integrating several Statistical Open Data (SOD) tables is a very promising issue. Various analysis scenarios are hidden behind these statistical data, which makes it important to have a holistic view of them. However, as these data are scattered in several tables, it is a slow and costly process to use existing pairwise schema matching approaches to integrate several schemas of the tables. Hence, we need automatic tools that rapidly converge to a holistic integrated view of data and give a good matching quality. In order to accomplish this objective, we propose a new 0-1 linear program, which automatically resolves the problem of holistic OD integration. It performs global optimal solutions maximizing the profit of similarities between OD graphs. The program encompasses different constraints related to graph structures and matching setup, in particular 1:1 matching. It is solved using a standard solver (CPLEX) and experiments show that it can handle several input graphs and good matching quality compared to existing tools.