Dynamically trained models of named entity recognition over unstructured data

A system, comprising:

• a domain including a master data schema and domain knowledge ontology, the domain knowledge ontology comprising relationship definitions in the domain;
• and
• a computing server comprising a memory and at least one processor, the memory comprising instructions executable by the at least one processor, the instructions, when executed, cause the at least one processor to:
• identify targeted types of named entities of the domain from the master data schema according to at least the relationship definitions in the domain knowledge ontology by:
• converting entity types defined in the master data schema to a plurality of domain grammar definitions in a resource description framework format;
• deriving relationships among the entity types in the domain grammar definitions;
• augmenting the relationships among the entity types with the domain knowledge ontology;
• and
• identifying the targeted types of named entities from the plurality of domain grammar definitions;
• extract a plurality of named entities from unstructured data of the domain according to the targeted types of named entities identified;
• generate one or more sequences of named entities, at least one of the named entities in the sequences extracted from the unstructured data;
• assign entity labels to the named entities in the one or more sequences, the entity label for one of the named entities in a sequence assigned based at least on another entity label in the sequence;
• and
• generate, based on the entity labels, one or more natural language phrases describing a relationship of a set of named entities identified in the one or more sequences.

A non-transitory computer readable medium for storing computer code comprising instructions, when executed by at least one processor, cause the at least one processor to:

• receive a master data schema and domain knowledge ontology of a domain, the domain knowledge ontology comprising relationship definitions in the domain;
• identify targeted types of named entities of the domain from the master data schema according to at least the relationship definitions in the domain knowledge ontology by:
• converting entity types defined in the master data schema to a plurality of domain grammar definitions in a resource description framework format;
• deriving relationships among the entity types in the domain grammar definitions;
• augmenting the relationships among the entity types with the domain knowledge ontology;
• and
• identifying the targeted types of named entities from the plurality of domain grammar definitions;
• extract a plurality of named entities from unstructured data of the domain according to the targeted types of named entities identified;
• generate one or more sequences of named entities, at least one of the named entities in the sequences extracted from the unstructured data;
• assign entity labels to the named entities in the one or more sequences, the entity label for one of the named entities in a sequence assigned based at least on another entity label in the sequence;
• and
• generate, based on the entity labels, one or more natural language phrases describing a relationship of a set of named entities identified in the one or more sequences.