Latent Ambiguity Handling in Natural Language Processing

A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to be specifically configured to implement a natural language processing system for detecting and resolving latent ambiguity in a natural language input at least by performing the following operations:

• processing, by a scoring component within an annotated semantic graph providing component, a semantic graph data structure to identify each node having a latently ambiguous semantic concept by applying a first statistical model that predicts a likelihood that the node is a latently ambiguous semantic concept and a second statistical model that predicts a likelihood that the node is an unambiguous node;
• marking, by the annotated semantic graph providing component, the semantic graph data structure to form an annotated semantic graph data structure based on a knowledge base in which nodes representing semantic concepts are linked by semantic relationships in the annotated semantic graph data structure, wherein one or more nodes are annotated to indicate a latent ambiguity of corresponding semantic concepts;
• processing, by a latent ambiguity processing component executing within the natural language processing system, the natural language input at least by performing activation of the annotated semantic graph data structure;
• applying, by the latent ambiguity processing component, in response to encountering an annotated node indicating a latent ambiguity of a corresponding semantic concept during processing of the natural language input, a pre-defined latent ambiguity process to the annotated node to resolve the latent ambiguity associated with the corresponding semantic concept present in the natural language input;
• and
• outputting, by an output component executing within the natural language processing system, a result of natural language processing of the natural language input based on the resolution of the latent ambiguity associated with the corresponding semantic concept present in the natural language input.

An apparatus comprising:

• at least one processor;
• and
• at least one memory coupled to the at least one processor, wherein the at least one memory comprises instructions which, when executed by the at least one processor, cause the at least one processor to be specifically configured to implement a natural language processing system for detecting and resolving latent ambiguity in a natural language input at least by performing the following operations:
• processing, by a scoring component within an annotated semantic graph providing component, a semantic graph data structure to identify each node having a latently ambiguous semantic concept by applying a first statistical model that predicts a likelihood that the node is a latently ambiguous semantic concept and a second statistical model that predicts a likelihood that the node is an unambiguous node;
• marking, by the annotated semantic graph providing component, the semantic graph data structure to form an annotated semantic graph data structure based on a knowledge base in which nodes representing semantic concepts are linked by semantic relationships in the annotated semantic graph data structure, wherein one or more nodes are annotated to indicate a latent ambiguity of corresponding semantic concepts;
• processing, by a latent ambiguity processing component executing within the natural language processing system, the natural language input at least by performing activation of the annotated semantic graph data structure;
• applying, by the latent ambiguity processing component, in response to encountering an annotated node indicating a latent ambiguity of a corresponding semantic concept during processing of the natural language input, a pre-defined latent ambiguity process to the annotated node to resolve the latent ambiguity associated with the corresponding semantic concept present in the natural language input;
• and
• outputting, by an output component executing within the natural language processing system, a result of natural language processing of the natural language input based on the resolution of the latent ambiguity associated with the corresponding semantic concept present in the natural language input.