Systems and methods for document similarity matching

A system comprising:

• a non-transitory computer-readable medium configured to store machine-readable instructions;
• a processor configured to access the non-transitory computer-readable medium and execute the machine-readable instructions, the machine-readable instructions comprising:
• a document parser module programmed to retrieve and parse a given document to identify a plurality of sentences of the given document;
• a plurality of domain natural language classifiers (NLCs), each domain NLC being programmed to classify each identified sentence to determine a sentence confidence score for each identified sentence of the given document, wherein each domain NLC is trained based on domain training data associated with a respective domain of a plurality of domains;
• and
• a similarity determination module programmed to:
• determine a plurality of document confidence scores for the given document based on sentence confidence scores determined by the plurality of domain NLCs, each document confidence score characterizing a relevance of the given document to the respective domain of the plurality of domains;
• evaluate the plurality of document confidence scores for the given document to identify a greatest document confidence score among the plurality of document confidence scores for the given document;
• generate document similarity data identifying at least one document among documents associated with a corresponding domain of the plurality of domains based on an evaluation of the document confidence scores for the given document and document confidence scores for the documents associated with the corresponding domain;
• and
• compare differences in document confidence score between the greatest document confidence score for the given document and the document confidence scores for the documents associated with the corresponding domain to a difference threshold to identify the at least one document associated with the corresponding domain.

A method comprising:

• training a plurality of domain natural language classifiers (NCLs) based on domain training data associated with a respective domain of a plurality of different domains, wherein the domain training data for training a corresponding domain NLC of the plurality of NCLs characterizes a domain-specific taxonomy model comprising a plurality of nodes, each node of the plurality of nodes representing an intent of at least one sentence from a corresponding document associated with the respective domain of the plurality of domains;
• receiving and parsing a given document to identify each sentence of the given document;
• classifying each identified sentence of the given document via each domain NLC of the plurality of domain NLCs to determine a sentence confidence score for each identified sentence of the given document, each sentence confidence score being provided by each of the plurality of domain NLCs being associated with a respective node of the plurality nodes of the domain-specific taxonomy model;
• determining a plurality of document confidence scores for the document based on sentence confidence scores determined by the plurality of domain NLCs, each document confidence score characterizing a relevance of the given document to the respective domain of the plurality of domains;
• and
• generating document similarity data identifying at least one document among documents associated with a corresponding domain of the plurality of domains based on an evaluation of the document confidence scores for the given document and document confidence scores for the documents associated with the corresponding domain;
• and
• providing the document similarity data to an output device for displaying thereon document location information for the at least one document, wherein the document similarity result data includes the document location information.