Abstract: A machine reading comprehension system (MRCS) can analyze a larger-sized document that includes multiple pages to predict an answer to a query. For example, the document can have two, five, tens, or hundreds of pages. The MRCS divides the document into multiple sections with each section including a portion of the document. Each section is processed separately by one or more processing circuitries to determine a score for that section. The score indicates how related the section is to the query and/or a probability that the section provides a possible answer to the query. Once all of the sections have been analyzed, the sections are ranked by their scores and a subset of the ranked sections are processed again to determine a predicted answer to the query.

Abstract: A server, method, and non-transitory computer readable medium for ranking a plurality of data sources are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving an input query, identifying entities of the input query using conditional random fields, generating a normalized query and applying a support vector machine to the normalized query. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: A system, method and computer program is provided for generating customized text representations of audio commands. A first speech recognition module may be used for generating a first text representation of an audio command based on a general language grammar. A second speech recognition module may be used for generating a second text representation of the audio command, the second module including a custom language grammar that may include contacts for a particular user. Entity extraction is applied to the second text representation and the entities are checked against a file containing personal language. If the entities are found in the user-specific language, the two text representations may be fused into a combined text representation and named entity recognition may be performed again to extract further entities.

Abstract: A server, method, and non-transitory computer readable medium for classifying queries based on contextual information are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving a plurality of queries, analyzing the queries and determining a likelihood divergence and selecting a domain. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: Described herein are systems and methods that provide a natural language processing system (NLPS) that employs compressed word embeddings. An auto-encoder that includes encoder circuitry and decoder circuitry can be used to produce the compressed word embeddings. The decoder circuitry is trained to decompress the word embeddings with reduced or minimal differences between the original uncompressed word embeddings and the corresponding decompressed word embeddings. One or more parameters of the trained decoder circuitry are transferred to the NLPS, where the NLPS is then trained using the compressed word embeddings to improve the correctness of the responses or actions determined by the NLPS.

Abstract: A server, method, and non-transitory computer readable medium for ranking a plurality of data sources are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving an input query, identifying entities of the input query using conditional random fields, generating a normalized query and applying a support vector machine to the normalized query. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: A server, method, and non-transitory computer readable medium for ranking a plurality of data sources are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving an input query, identifying entities of the input query using conditional random fields, generating a normalized query and applying a support vector machine to the normalized query. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: A machine reading comprehension system (MRCS) can analyze a larger-sized document that includes multiple pages to predict an answer to a query. For example, the document can have two, five, tens, or hundreds of pages. The MRCS divides the document into multiple sections with each section including a portion of the document. Each section is processed separately by one or more processing circuitries to determine a score for that section. The score indicates how related the section is to the query and/or a probability that the section provides a possible answer to the query. Once all of the sections have been analyzed, the sections are ranked by their scores and a subset of the ranked sections are processed again to determine a predicted answer to the query.

Abstract: A machine reading comprehension system (MRCS) can analyze a larger-sized document that includes multiple pages to predict an answer to a query. For example, the document can have two, five, tens, or hundreds of pages. The MRCS divides the document into multiple sections with each section including a portion of the document. Each section is processed separately by one or more processing circuitries to determine a score for that section. The score indicates how related the section is to the query and/or a probability that the section provides a possible answer to the query. Once all of the sections have been analyzed, the sections are ranked by their scores and a subset of the ranked sections are processed again to determine a predicted answer to the query.

Abstract: A server, method, and non-transitory computer readable medium for ranking a plurality of data sources are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving an input query, identifying entities of the input query using conditional random fields, generating a normalized query and applying a support vector machine to the normalized query. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: A system, method and computer program is provided for generating customized text representations of audio commands. A first speech recognition module may be used for generating a first text representation of an audio command based on a general language grammar. A second speech recognition module may be used for generating a second text representation of the audio command, the second module including a custom language grammar that may include contacts for a particular user. Entity extraction is applied to the second text representation and the entities are checked against a file containing personal language. If the entities are found in the user-specific language, the two text representations may be fused into a combined text representation and named entity recognition may be performed again to extract further entities.

Abstract: A system, method and computer program is provided for generating customized text representations of audio commands. A first speech recognition module may be used for generating a first text representation of an audio command based on a general language grammar. A second speech recognition module may be used for generating a second text representation of the audio command, the second module including a custom language grammar that may include contacts for a particular user. Entity extraction is applied to the second text representation and the entities are checked against a file containing personal language. If the entities are found in the user-specific language, the two text representations may be fused into a combined text representation and named entity recognition may be performed again to extract further entities.

Abstract: A server, method, and non-transitory computer readable medium for ranking a plurality of data sources are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving an input query, identifying entities of the input query using conditional random fields, generating a normalized query and applying a support vector machine to the normalized query. The non-transitory computer readable medium is encoded with programming instructions to direct a processor to carry out the method.

Abstract: A method, system, and computer program product provide a conversation agent to process natural language queries expressed by a user and perform commands according to the derived intention of the user. A natural language processing (NLP) engine derives intent using conditional random fields to identify a domain and at least one task embodied in the query. The NLP may further identify one or more subdomains, and one or more entities related to the identified command. A template system creates a data structure for information relevant to the derived intent and passes a template to a services manager for interfacing with one or more services capable of accomplishing the task. A dialogue manager may elicit more entities from the user if required by the services manager and otherwise engage in conversation with the user. In one embodiment, the conversational agent allows a user to engage in multiple conversations simultaneously.

Abstract: A method, system, and computer program product provide a conversation agent to process natural language queries expressed by a user and perform commands according to the derived intention of the user. A natural language processing (NLP) engine derives intent using conditional random fields to identify a domain and at least one task embodied in the query. The NLP may further identify one or more subdomains, and one or more entities related to the identified command. A template system creates a data structure for information relevant to the derived intent and passes a template to a services manager for interfacing with one or more services capable of accomplishing the task. A dialogue manager may elicit more entities from the user if required by the services manager and otherwise engage in conversation with the user. In one embodiment, the conversational agent allows a user to engage in multiple conversations simultaneously.

Abstract: A server, method, and non-transitory computer readable medium for classifying a query into one of a plurality of classes are provided. The server includes a network interface, a memory storage unit and a processor. The method involves receiving a query applying a plurality of support vector machine models, calculating a probability, and determining a top class. The non-transitory computer readable medium is encoded with codes to direct a processor to carry out the method.

Abstract: A method, system, and computer program product provide a conversation agent to process natural language queries expressed by a user and perform commands according to the derived intention of the user. A natural language processing (NLP) engine derives intent using conditional random fields to identify a domain and at least one task embodied in the query. The NLP may further identify one or more subdomains, and one or more entities related to the identified command. A template system creates a data structure for information relevant to the derived intent and passes a template to a services manager for interfacing with one or more services capable of accomplishing the task. A dialog manager may elicit more entities from the user if required by the services manager and otherwise engage in conversation with the user. In one embodiment, the conversational agent allows a user to engage in multiple conversations simultaneously.

Abstract: Disclosed embodiments include systems and methods relevant to improvements to natural language processing used to determine an intent and one or more associated parameters from 5 a given input string. In an example, an input string is received and first and second different n-grams are applied to the input string. Recurrent neural network models are then used to generate output data based in part on the first and second different n-grams. In particular embodiments a recurrent neural network in both forward and backward directions specific to unigrams is applied. Intent detection and semantic labeling are applied to the output of the recurrent neural network models.

Abstract: A method and system are provided for processing natural language user queries for commanding a user interface to perform functions. Individual user queries are classified in accordance with the types of functions and a plurality of user queries may be related to define a particular command. To assist with classification, a query type for each user query is determined where the query type is one of a functional query requesting a particular new command to perform a particular type of function, an entity query relating to an entity associated with the particular new command having the particular type of function and a clarification query responding to a clarification question posed to clarify a prior user query having the particular type of function. Functional queries may be processed using a plurality of natural language processing techniques and scores from each technique combined to determine which type of function is commanded.

Abstract: Described herein are systems and methods for two-stage training of a spoken dialog system. The first stage trains a policy network using external data to produce a semi-trained policy network. The external data includes one or more known fixed dialogs. The second stage trains the semi-trained policy network through interaction to produce a trained policy network. The interaction may be interaction with a user simulator.