Abstract: Systems, methods, and other embodiments associated with equivalence reasoning are described. One example method includes iteratively inputting batches of unprocessed equivalence pairs from a semantic model to an operating memory. In the operating memory, one or more cliques for the input batches are built until no further batches remain. A clique designates a canonical representative resource for a group of equivalent resources as determined from the equivalence pairs. The one or more cliques are built for the input batches to a clique map in a remote access memory. The clique map is returned for use by the semantic model.

Abstract: Systems, methods, and other embodiments associated with data sources adapted for parallel inference on triples associated with a semantic model are described. One example method includes creating a source table that is partitioned on triple predicate and stores triples for entailment. The source table may store compact triple identifiers that have been mapped to triple identifiers from the semantic model.

Abstract: A method is provided for allowing programmers to specify program execution control semantics using standard programming language syntax even when the standard language does not provide a language construct for specifying execution control. In a similar manner, the approach provides programmers the ability to extend the expressiveness of a language by introducing statements expressed in the syntax of a target programming language. A program written in a first programming language may be translated into statements of a second programming language, where the target programming language is more expressive than the first. This language-based approach preserves the standard syntax of the first programming language, allowing a program written with semantic extensions to be compiled and run according to the standard on any standards-compliant system.

Abstract: Disclose is an improved approach for caching lexical values for RDF data. A compact, efficient resource ID to value mapping is stored in memory, where the cache is stored in memory such that the overhead of accessing and updating it is minimized. The advantage of this approach is improved query performance, since the database is performing less work with the in-memory RDF resource caching. In addition, throughput and performance of the database are also improved.

Abstract: Systems, methods, and other embodiments associated with equivalence reasoning are described. One example method includes iteratively inputting batches of unprocessed equivalence pairs from a semantic model to an operating memory. In the operating memory, one or more cliques for the input batches are built until no further batches remain. A clique designates a canonical representative resource for a group of equivalent resources as determined from the equivalence pairs. The one or more cliques are built for the input batches to a clique map in a remote access memory. The clique map is returned for use by the semantic model.

Abstract: Systems, methods, and other embodiments associated with equivalence reasoning are described. One example method includes constructing a clique map that maps groups of equivalent resources to a representative canonical resource. The clique map is constructed by iteratively inputting batches of equivalence pairs from a semantic model to operating memory and building, in operating memory, one or more cliques for the input batch. Cliques for various input batches of equivalence pairs are merged in memory to produce the clique map for use by the semantic model. Triples associated with the semantic model may be consolidated by replacing resources that are in a clique with their corresponding representative canonical resource.

Abstract: Systems, methods, and other embodiments associated with data sources adapted for parallel inference on triples associated with a semantic model are described. One example method includes creating a source table that stores triples for entailment in a manner that is adapted for parallel inference. The source table may be partitioned by triple predicate or may store compact triple identifiers that have been mapped to triple identifiers from the semantic model.

Abstract: An un-indexed, partitioned temporary table and an exchange table are used in the inferencing of semantic data in a relational database system. The exchange table has the same structure as a semantic data table storing the semantic data. In the inferencing process, a new partition is created in the semantic data table. Inference rules are executed on the semantic data table, and any newly inferred semantic data generated is added to the temporary table. Once no new data is generated, the inferred semantic data is copied from the temporary table into the exchange table. Indexes that are the same as indexes for the semantic data table are built for the exchange table. The indexed data in the exchange table is then exchanged into the new partition in the semantic data table. By use of the un-indexed, partitioned temporary table, incremental maintenance of indexes is avoided, thus allowing for greater efficiency.

Abstract: Disclose is an improved approach for caching lexical values for RDF data. A compact, efficient resource ID to value mapping is stored in memory, where the cache is stored in memory such that the overhead of accessing and updating it is minimized. The advantage of this approach is improved query performance, since the database is performing less work with the in-memory RDF resource caching. In addition, throughput and performance of the database are also improved.

Abstract: Systems, methods, and other embodiments associated with incremental inference are described. One example method includes updating existing or old triples in a semantic model with triples resulting from the addition of new triples. The updating is performed by separating inference rules into joining steps that are performed on first and second predicates for the inference rule. A first joining step joins results of execution of the first predicate on the new triples with the results of execution the second predicate on the union of the old and new triples to produce newly inferred triples. A second joining step joins results of execution of the first predicate on the union of the old and new triples with the results of execution the second predicate on the new triples to produce newly inferred triples.

Abstract: A method is provided for allowing programmers to specify program execution control semantics using standard programming language syntax even when the standard language does not provide a language construct for specifying execution control. In a similar manner, the approach provides programmers the ability to extend the expressiveness of a language by introducing statements expressed in the syntax of a target programming language. A program written in a first programming language may be translated into statements of a second programming language, where the target programming language is more expressive than the first. This language-based approach preserves the standard syntax of the first programming language, allowing a program written with semantic extensions to be compiled and run according to the standard on any standards-compliant system.

Abstract: Systems, methods, and other embodiments associated with data sources adapted for parallel inference on triples associated with a semantic model are described. One example method includes creating a source table that stores triples for entailment in a manner that is adapted for parallel inference. The source table may be partitioned by triple predicate or may store compact triple identifiers that have been mapped to triple identifiers from the semantic model.

Abstract: Systems, methods, and other embodiments associated with incremental inference are described. One example method includes updating existing or old triples in a semantic model with triples resulting from the addition of new triples. The updating is performed by separating inference rules into joining steps that are performed on first and second predicates for the inference rule. A first joining step joins results of execution of the first predicate on the new triples with the results of execution the second predicate on the union of the old and new triples to produce newly inferred triples. A second joining step joins results of execution of the first predicate on the union of the old and new triples with the results of execution the second predicate on the new triples to produce newly inferred triples.

Abstract: Systems, methods, and other embodiments associated with equivalence reasoning are described. One example method includes constructing a clique map that maps groups of equivalent resources to a representative canonical resource. The clique map is constructed by iteratively inputting batches of equivalence pairs from a semantic model to operating memory and building, in operating memory, one or more cliques for the input batch. Cliques for various input batches of equivalence pairs are merged in memory to produce the clique map for use by the semantic model. Triples associated with the semantic model may be consolidated by replacing resources that are in a clique with their corresponding representative canonical resource.

Abstract: An un-indexed, partitioned temporary table and an exchange table are used in the inferencing of semantic data in a relational database system. The exchange table has the same structure as a semantic data table storing the semantic data. In the inferencing process, a new partition is created in the semantic data table. Inference rules are executed on the semantic data table, and any newly inferred semantic data generated is added to the temporary table. Once no new data is generated, the inferred semantic data is copied from the temporary table into the exchange table. Indexes that are the same as indexes for the semantic data table are built for the exchange table. The indexed data in the exchange table is then exchanged into the new partition in the semantic data table. By use of the un-indexed, partitioned temporary table, incremental maintenance of indexes is avoided, thus allowing for greater efficiency.