Example 1: Quality Assessment of RDF Data Cubes
Prior to publishing the resulting RDF data on an existing data portal and thus enabling other users to download and exploit the data for various purposes, every dataset should be validated to ensure it conforms to the RDF Data Cube model.
Table 3. Example values for a Data Cube structure representing the Serbian economic statistics
|
Component property |
Concept description |
Identifier |
Code list |
|
Dimension |
Geographical region |
rs:geo |
cl:geo |
|
Dimension |
Time |
rs:time |
cl:time |
|
Dimension |
Economic activity |
rs:activityNACEr2 |
cl:nace rev2 |
|
Attribute |
Unit of measurement |
sdmx- attribute:unitMeasure |
cl:esa95-unit |
|
Measure |
Observed value |
sdmx-measure:obsValue |
Fig. 4. RDF Data Cube quality assessment
The data validation step is covered by the LOD2 stack, i.e. through the following software tools:
• The RDF Data Cube Validation Tool[1];
• CubeViz, a tool for visualization of RDF Data Cubes[2].
The RDF Data Cube Validation Tool aims at speeding-up the processing and publishing of Linked Data in RDF Data Cube format. Its main use is validating the integrity constraints defined in the RDF Data Cube specification. It works with the Virtuoso Universal Server as a backend and can be run from the LOD2 Statistical Workbench environment.
The main benefits of using this component are improved understanding of the RDF Data Cube vocabulary and automatic repair of identified errors. Figure 4 shows the component in action: the integrity constraints and their status are shown on the left side, while the results of analysis are shown on the right. A list of resources that violate the constraint, an explanation about the problem, and if possible, a quick solution to the problem is offered to the user. Once an RDF
Data Cube satisfies the Data Cube integrity constraints, it can be visualized with CubeViz. More details can be found in the LOD2 Stack Documentation[3].
Example 2: Filtering, Visualization and Export of RDF Data Cubes
The facetted browser and visualization tool CubeViz can be used to filter observations to be visualized in charts interactively. Figure 5 shows an exploration session that comprises of the following steps:
Fig. 5. RDF Data Cube exploration and analysis
1. Select one out of the available datasets in the RDF graph;
2. Choose observations of interest by using a subset of the available dimensions;
3. Visualize the statistics by using slices, or
4. Visualize the statistics in two different measure values (millions of national currency and percentages).
Example 3: Merging RDF Data Cubes
Merging[4] is an operation of creating a new dataset (RDF Data Cube) that compiles observations from the original datasets (two or more), and additional resources (e.g. data structure definition, component specifications). In order to obtain meaningful charts the observed phenomena (i.e. serial data) have to be described on the same granularity level (e.g. year, country) and expressed in same units of measurement (e.g. euro, %). Therefore alignment of the code lists used in the input data is necessary before the merging operation is performed.
