SYSTEM AND METHOD FOR SUPPORTING AD HOC MULTILINGUAL NATURAL LANGUAGE DATABASE QUESTIONS WHICH ARE TRANSFORMED IN REALTIME TO DATABASE QUERIES USING A SUITE OF CASCADING, DEEP NEURAL NETWORKS WHICH ARE INITIALLY AND CONTINUOUSLY TRAINED THROUGH SCHEMA INFORMATION AND QUESTION-QUERY EXAMPLES

1 . A system to facilitate online ad hoc multilingual natural language database questions that are transformed in realtime to database queries using a suite of cascading Deep Neural Networks which are initially and continuously trained through schema information and question-query examples, the system comprising:

a processor; and

a memory device coupled to the processor and storing executable program instructions therein, which, when executed by the processor, cause the system to perform operations comprising:

receiving over a network from a client device a client communication including an ad hoc multilingual natural language question of a database;

cascading the natural language query through at most three successive attempts by three different instances of Deep Neural Networks components in the system, each of which is activated if the previous Deep Neural Network component is unable to predict an output database query.

2 . The system of claim 1 , wherein the submitted natural language question is cascaded to the first instance of a Deep Neural Network component in the system, whose corresponding model is initially and continuously trained in auto label generated natural language questions and their corresponding queries based on the database schema. This first instance of a Deep Neural Network component in the system, using fuzzy logic, attempts to match the submitted natural language question against its model, and if successful returns the matching SQL query for execution.

3 . The system of claim 1 , wherein if a match is not successful using the method in claim 2 , the system then cascades the submitted natural language question to the second instance of a Deep Neural Network component in the system, whose corresponding model is initially and continuously trained with previously submitted natural language questions and their corresponding queries, along with a set of inferred natural language questions based on previously submitted natural language questions. This second instance of a Deep Neural Network component in the system, attempts to resolve the linguistic ambiguity in the submitted natural language question using its model, and if successful returns the matching SQL query for execution.

4 . The system of claim 1 , wherein if a match is not successful using the method described in claim 3 , the system then cascades the submitted natural language question to the third instance of a Deep Neural Network component in the system, whose corresponding model is initially and continuously trained with a vast range of hypothetical natural language questions and their corresponding queries based on the database schema. This third instance of a Deep Neural Network component in the system, attempts to match the submitted natural language question against its model and if successful returns an SQL query for execution.