Convolutional state modeling for planning natural language conversations

Abstract

In one aspect, a computerized method useful for, with an ensemble of Natural Language Understanding and Processing methods converting a set of user actions into machine queries, includes the step of providing a knowledge model. The method includes the step of receiving a natural language user query; preprocesses the natural language user query for further processing as a preprocessed user query. The preprocessing includes the step of chunking a set of sentences of the natural language query into a set of smaller sentences and retaining the references between chunks of the set of sentences. The method includes the step of, with the preprocessed user query. For each chunk of the chunked preprocessed user query the method implements the following steps.

What is claimed is: 
     
         1 . A computerized method useful for, with an ensemble of Natural Language Understanding and Processing methods, converting a set of user actions into one or more machine queries over a convoluted data model, comprising:
 providing a convoluted knowledge model;   receiving a natural language user query;   preprocessing the natural language user query for further processing as a preprocessed user query, wherein the preprocessing comprises;
 chunking a set of sentences of the natural language user query into a set of smaller sentences, and 
 retaining the references between chunks of the set of sentences; 
   with the preprocessed user query, for each chunk of the chunked preprocessed user query:
 using a Name Entity Recognition (NER) ensemble to extract a domain specific name entity from the chunked preprocessed user query, 
 using a sentiment analysis technique to determine a sentiment of each chunk of the chunked preprocessed user query, 
 using a classification technique to produce one or more classes for each chunk of the chunked preprocessed user query, 
   for each chunk:
 translating each chunk to a standalone system query and a contextual system query, 
 querying the knowledge model using the standalone system query and the contextual system query of each chunk to determine a closest state in the knowledge model, 
 returning a set qualified decisions from the knowledge model that match the standalone system query and the contextual system query of each chunk, wherein each qualified decision is scored, 
 ranking the set of qualified decisions based on the score of each qualified decision, and 
 detecting a winner state in the knowledge model as a highest ranked member of the set of qualified decisions. 
   
     
     
         2 . The computerized method of  claim 1 , wherein the contextual system query comprises a set of variables used to query the knowledge model in a vector space model. 
     
     
         3 . The computerized method of  claim 2 , wherein the standalone system query comprises a translated query using data provided by the user without using the contextual state data to produce a qualified state in the knowledge model. 
     
     
         4 . The computerized method of  claim 3 , wherein the NER ensemble is used to locate and classify a set of named entities into pre-defined categories. 
     
     
         5 . The computerized method of  claim 4 , wherein the knowledge model comprises a model that represents knowledge in form of states in a Vector Space Model represented by tensors or set of vectors or variables. 
     
     
         6 . The computerized method of  claim 5 , wherein the winner state is derived using a scoring model based on a cosine similarity distance and a soft cosine similarity distance between the contextual system query, the context-free system query and the knowledge model state. 
     
     
         7 . The computerized method of  claim 5 , wherein the system states are atomic states or hybrid states. 
     
     
         8 . The computerized method of  claim 6 , wherein the system states are represented by tensors or set of variables in a vector space model. 
     
     
         9 . The computerized method of  claim 6 , wherein the winner state updates a context object, and wherein the context object is used for a future contextual query or a successive chunk. 
     
     
         10 . A computerized system useful for, with an ensemble of Natural Language Understanding and Processing methods converting a set of user actions into machine queries, comprising:
 at least one processor configured to execute instructions;   a memory containing instructions when executed on the processor, causes the at least one processor to perform operations that:
 provide a knowledge model; 
 receive a natural language user query; 
 preprocess the natural language user query for further processing as a preprocessed user query, wherein the preprocessing comprises; 
 chunk a set of sentences of the natural language user query into a set of smaller sentences, and 
 retain the references between chunks of the set of sentences. 
   
     
     
         11 . The computerized system of  claim 7 , wherein the memory causes the at least one processor to perform operations that:
 with the preprocessed user query, for each chunk of the chunked preprocessed user query:
 use a Name Entity Recognition (NER) ensemble to extract a domain specific name entity from the chunked preprocessed user query, 
 use a sentiment analysis technique to determine a sentiment of each chunk of the chunked preprocessed user query, 
 use a classification technique for topic modeling each chunk of the chunked preprocessed user query, 
 append each sequential chunk to a previously analyzed chunk, 
 translate each chunk to a standalone system query and a contextual system query, 
 query the knowledge model using the standalone system query and the contextual system query of each chunk to determine a closest state in the knowledge model, 
 return a set qualified decisions from the knowledge model that match the standalone system query and the contextual system query, wherein each qualified decision is scored, 
 rank the set of qualified decisions based on the score of each qualified decisions, and 
 detect a winner state in the knowledge model as a highest ranked member of the set of qualified decisions. 
   
     
     
         12 . The computerized system of  claim 8 , wherein the contextual system query comprises a set of variables used to query the knowledge model in a vector space model. 
     
     
         13 . The computerized system of  claim 9 , wherein the standalone system query comprises a translated query using data provided by the user without using the contextual state data to produce a qualified state in the knowledge model. 
     
     
         14 . The computerized system of  claim 10 , wherein the NER ensemble is used to locate and classify a set of named entities into pre-defined categories. 
     
     
         15 . The computerized system of  claim 11 , wherein the knowledge model comprises a model that represents knowledge in form of states in a Vector Space Model represented by tensors or set of vectors or variables. 
     
     
         16 . The computerized system of  claim 12 , wherein the winner state is derived using scoring model based on a cosine similarity and a soft cosine similarity distance between the contextual system query and the knowledge model state.