SDM-RDFizer. The SDM-RDFizer is an RML engine capable of creating a knowledge graph from multiple data sources formats like CSV, TSV, JSON, and XML and relational databases like MySQL and Postgres following the rules established in an RML mapping. The SDM-RDFizer implements the data structures Predicate Tuple Table (PTT) and Predicate Join Tuple Table (PJTT) to execute duplicated removal and joins more efficiently. The PTT stores for each predicate P the triples that have been generated so far. The key encodes the subject and object of the triple. The PJTT stores the subjects of the triples generated by a join. PJTT is an index hash table where the key encodes each value of the attributes in the join condition. The value is a set of subject values in the second source associated with the values of the attributes in the hash key. Additionally, the physical operators Simple Object Map (SOM), Object Reference Map (ORM), and Object Join Map (OJM) have been implemented in the SDM-RDFizer. SOM generates an RDF triple from the execution of a simple Predicate Object Map. Each generated triple is checked against the associated PTT. If the triple was already generated before, it is discarded. If not, it will be added to the knowledge graph and the PTT will be updated accordingly. ORM extends SOM by using the subject of one Triples Map as the object of another Triples Map. The condition for this operator to work is that both Triples 2 xxxxx://xxxxx.xxxx.xxx/ 3 xxxxx://xx.xxxxx-xxxxxxxxx.xx/seas/index.html 4 xxxxx://xxxxxxxx.xxx.xx/cerise/cim-profile/ 5 xxxxx://xxx.x0.xxx/TR/vocab-ssn/ 6 xxxxx://xxxxxxx.xxxxxxxxx.xxx/ontologies/oema/index-en.html 7 xxxxx://x0x-xxx-xx.xxxxxx.xx/bot/ 8 xxxx://xxx.xxxxxxx-xxxxx.eu/urban-enery-ontology Maps must use the same data source. Afterward, the same process as with SOM is followed, i.e., the triples are checked against to the corresponding PTT to avoid duplicate triples. The OJM is an extension of ORM with the difference where the Triples Maps can be defined over different data sources and there exists a join condition between them. The corresponding PJTT is used in an index join where the outer table corresponds to the child map and the inner table to the PJTT. If an entry e exists with the same hash key, all the subjects in e are used to generate the resulting RDF triples. Finally, a similar procedure as before is followed to avoid the generation of duplicate triples. Xxxxxxxx et al. provide a more detailed description of the operators imple...
SDM-RDFizer. The SDM-RDFizer implements efficient data structures, data caching techniques, and query optimization strategies to scale up to a large number of mapping rules and data sets. As a result, semantic enrichment can be efficiently computed. The outcome of evaluations of the implemented techniques can be found in the reports [7], [8], and [9]. The SDM-RDFizer implements the data structures Predicate Tuple Table (PTT) and Predicate Join Tuple Table (PJTT) to execute duplicated removal and joins more efficiently. The PTT stores for each predicate P the triples that have been generated so far. The key encodes the subject and object of the triple. The PJTT stores the subjects of the triples generated by a join. PJTT is an index hash table where the key encodes each value of the attributes in the join condition. The value is a set of subject values in the second source associated with the values of the attributes in the hash key. Additionally, the physical operators Simple Object Map (SOM), Object Reference Map (ORM), and Object Join Map (OJM) have been implemented in the SDM-RDFizer. SOM generates an RDF triple from the execution of a simple Predicate Object Map. Each generated triple is checked against the associated PTT. If the triple was already generated before, it is discarded. If not, it will be added to the knowledge graph and the PTT will be updated accordingly. ORM extends SOM by using the subject of one Triples Map as the object of another Triples Map. The condition for this operator to work is that both Triples Maps must use the same data source. Afterward, the same process as with SOM is followed, i.e., checking against the PTT to avoid duplicated triples. The OJM is an extension of ORM with the difference that the Triples Maps can be defined over different data sources and there exists a join condition between them. The corresponding PJTT is used in an index join where the outer table corresponds to the child map and the inner table to the PJTT. If an entry e exists with the same hash key, all the subjects in e are used to generate the resulting RDF triples. Finally, a similar procedure as before is followed to avoid the generation of duplicate triples. Xxxxxxxx et al. [8] [9] provide a more detailed description of the operators implemented by SDM-RDFizer.
