An example of taxonomy 

Context in source publication

Context 1

... the end of object property extraction process, the generator stores all information which is used later during the instance generation phase. The instances generator has been implemented in two successive incremental versions. The first version generates instances with data properties only. The subsequent version produces instances with both data and object properties. In this section, we provide the implementation detail of both versions. The first version of the generator can be used in a special case. To be specific, it extracts instances from the ontologies which represent taxonomies only. A taxonomy is a specific type of ontology, where relations between concepts and their child nodes are defined hierarchically without defining any complex property such as symmetric property of concepts. A taxonomy is an ontology without properties. Figure 3 provides an example of a taxonomy. During instance generation from such ontologies, the data properties of classes must be propagated to all of its child nodes (sub-classes) in the hierarchy. The data properties can be any of the following types: Boolean , Double , Float , Integer , and String . Instances are generated in two steps. First, the generator copies the given ontology file (loaded onto the main memory) and writes instances at the end of that file. Note that, in some cases, the generator creates a new file and adds instances there instead of adding them to the ontology file. Then, in the second step, the generator writes the instances to the file stored in secondary storage. This approach frees main memory significantly and thus, prevents memory overflow and optimizes the performance. In the extraction phase, the generator defines an association between concepts and the domains of data properties inherited from the super-classes. This information are stored in TClass . This enables the generator to be more efficient. more efficient because, for any concept, the generator creates its instances without the need to search its super-attributes. Furthermore, our generator uses a buffer instead of accessing secondary storage devices (hard disks), which enhances its performance. Once the buffer exceeds a predetermined storage capacity, its contents are moved to a file using the WRITE() function. Then the buffer is automatically reset to empty state by the generator. The pseudocode in Algorithm 4 demonstrates the implementation of the initial version and how it works. It is worth noting that this algorithm is integrated into the final version. Below, we provide the list of steps which show how instances are generated using the initial ...