Symbolic data not only serve to summarize large data sets, but also lead to more complex data tables, thus enabling the manipulation of huge data sets (Billard and Diday, 2003). In order to handle the underlying concepts, a statistical metadata model is required. In this section we develop step by step a statistical metadata model designed especially for the symbolic data environment in order to capture the metainformation needed for the creation of symbolic objects (SOs) and symbolic data tables (SDTs) and the successful implementation of symbolic data analysis methods. The metadata model under consideration should hold metainformation for the classical (original) data (survey variables, statistical units, frame population, etc.) and the symbolic data. More specifically, it should store metainformation both for the main stages of the processes of the classical data analysis, and for the symbolic data analysis procedures. It is also desirable for the model to store the processing history, from the creation of the original variables to the creation of the SDT and the application of certain statistical methods and/or the visualization of the final results.

Finally, the applicability of the model is verified by providing an example based on a data table recorded in Bock and Diday (2000, p. 58), illustrating the capability of the model in the simultaneous manipulation of both data and metadata, as well as for the improvement of the visualization of SOs.