Semantic networks can now be created in a regular user’s schema. Such networks are called schema-private semantic networks because the associated database objects are created in the network owner’s schema, and the network owner has exclusive privileges to those objects. (DBA users also have such privileges, and the network owner or a DBA can grant and revoke the privileges for other users.)

In previous releases, the only scenario was a single semantic network owned by the MDSYS user and asvailable to the entire database. That scenario is still supported, but you are encouraged to use schema-private semantic networks instead.

For more information, see Semantic Networks.

The feature name previously called RDF Semantic Graph is now called RDF Knowledge Graph.

The terms semantic network and semantic data are still widely used throughout this guide.

RDF data can easily be loaded into memory to take advantage of the Oracle Database In-Memory feature. A semantic network can now be loaded into memory with the SEM_APIS.ENABLE_INMEMORY procedure. In addition, in-memory virtual columns can be used at the virtual model level to add lexical values for RDF terms to the in-memory representation of the MDSYS.RDF_LINK$ table, thus reducing the number of joins required to evaluate SPARQL queries.

For more information, see RDF Support for Oracle Database In-Memory.

Semantic networks can now be created with list-hash composite partitioning. With this scheme, a semantic network is initially list-partitioned by model id, and then each partition is subpartitioned using a hash of the RDF predicate ID. Composite partitioning improves SPARQL query performance through increased parallelization and better query optimizer statistics.

For more information, see Semantic Networks.

Staging tables that contain RDF quads with long literals (RDF object values greater than 4000 bytes in size) can now be efficiently loaded with SEM_APIS.BULK_LOAD_FROM_STAGING_TABLE.

New options have also been added to SEM_APIS.LOAD_INTO_STAGING_TABLE for better handling of long literals when loading from an external table. The VC_ONLY option loads only RDF quads with object values not larger than 4000 bytes into a staging table, and the CLOB_ONLY option loads only RDF quads with object values larger than 4000 bytes. These options allow a very efficient two-phase bulk load where VARCHAR-only data is loaded in one bulk load operation and CLOB-only data is loaded in a second bulk load operation.

Turtle and Trig RDF formats can now be directly loaded into Oracle Database without the need for third-party tools. SPARQL LOAD operations executed through SEM_APIS.UPDATE_MODEL can now parse and insert RDF data serialized in Turtle and Trig formats in addition to the N-Triple and N-Quad formats that were previously supported.

RDF Views can now be created directly from R2RML mappings specified in Turtle or N-Triple format. New R2RML_STRING and R2RML_STRING_FMT arguments have been added to SEM_APIS.CREATE_RDFVIEW_MODEL so that an R2RML mapping string can be used to create an RDF View model.

You can use Oracle SQL Developer to create RDF-related objects and use RDF and OWL features.

For more information, see RDF Support in SQL Developer.