System for extracting semantic triples for building a knowledge base

A computer-implemented method of filtering semantic triples generated for building a knowledge base including data patterns defining concepts associated with semantic triples derived from unstructured text, the method comprising:

• receiving a semantic triple including a subject, a relation, and an object;
• comparing the relation to a pre-defined set of relevant relations;
• accepting the semantic triple in an event that the relation is present in the pre-defined set of relevant relations, and rejecting the semantic triple in an event that the relation is not present in the pre-defined set of relevant relations;
• and
• if the semantic triple is accepted, transmitting the accepted semantic triple for inclusion as a data pattern in the knowledge base, and if the semantic triple is rejected, discarding the semantic triple.

A non-transitory computer readable medium including instructions which, when executed by a processor, cause the processor to perform a method of generating a semantic triple for building a knowledge base to include data patterns associated with semantic triples derived from unstructured text, the method comprising:

• providing a sentence associated with unstructured text including a main verb and a taxonomic verb, the taxonomic verb describing a taxonomic relationship;
• generating a first frame including a first subject, a first relation, and a first object, the first relation including the main verb, wherein the first subject and the first object each include a phrase including one or more sub-strings;
• generating a second frame including a second subject, a second relation, and a second object, the second relation including the taxonomic verb, wherein the second subject and the second object each include a phrase including one or more sub-strings;
• identifying a common sub-string in the first subject and either the second subject or the second object, or a common sub-string in the first object and either the second subject or the second object;
• generating a semantic triple using the first frame and replacing the first subject or the first object having the common sub-string with the second subject or the second object having the common sub-string;
• and
• transmitting the semantic triple for inclusion as a data pattern in a knowledge base.