Systems methods and computer-readable storage media for real-time automated conversational agent

Abstract

Systems, methods and computer-readable storage media for natural language understanding in combination with real-time automated responses verification in conversation agent messaging are described.

1 . A system for operating a real-time automated conversational agent, characterized by human crowd verified responses, said system comprises:
 (a) at least one user computing device comprising:
 (I) a micro-processor programmable to operate a terminal interface of a conversational agent; 
 (II) at least one input/output (I/O) hardware component configured to:
 (i) provide an incoming message collected at said terminal interface; 
 (ii) communicate a verified response to said user via at said terminal interface; 
 
 (III) a communication port configured to:
 (i) transmit said incoming message collected at said terminal interface; 
 (ii) receive said verified response communicable to said user via at said terminal interface; 
 
   (b) at least one natural language programming (NLP) server, operatively connected to said at least one user computing device, said at least one NLP server comprising:
 (I) a computer-readable storage medium configured to store:
 (i) a plurality of predefined verified responses; 
 (ii) at least one set of rules for determining that there is a verified response matching said incoming message with a matching score exceeding a predefined matching score threshold, among said plurality of predefined verified responses; 
 (iii) at least one set of rules for censoring and/or filtering said incoming message; 
 
 (I) a micro-processor programmable to:
 (i) determine that there is said verified response matching said incoming message with said matching score exceeding said predefined matching score threshold among said plurality of verified responses, by executing said at least one set of rules for determining; 
 (ii) censor and/or filter said incoming message, according to said at least one set of rules for censoring and/or filtering, by executing said at least one set of rules for censoring and/or filtering; 
 
 (II) a communication port configured to:
 (i) receive said incoming message, collected at said terminal interface of said at least one user computing device; 
 (ii) transmit said verified response to said at least one user computing device; 
 (iii) transmit a censored/filtered incoming message from said at least one NLP server; 
 (iv) receive an indication of said verified response, among said plurality of verified responses; 
 
   (c) at least one challenge server, operatively connected to said at least one NLP server, said at least one challenge server comprising:
 (I) a computer-readable storage medium configured to store:
 (i) metrics of registered participants; 
 (ii) at least one set of rules for constructing a challenge; 
 
 (II) a micro-processor programmable to construct a challenge from said censored/filtered incoming message, by executing said at least one set of rules for constructing said challenge; 
 (III) a communication port configured to:
 (i) receive said censored/filtered incoming message from said at least one NLP server; 
 (ii) transmit said challenge from said at least one challenge server; 
 (iii) receive responses to said challenge; 
 (iv) transmit to said at least one NLP server said indication of said verified response, among said plurality of verified responses; 
 
   (d) at least one participant computing device, operatively connected to said at least one challenge server, comprising:
 (I) at least one input/output (I/O) hardware component configured to:
 (i) communicate said challenge to a participant; 
 (ii) collect a response of said participant to said challenge; 
 
 (II) a communication port configured to:
 (i) receive said challenge from said at least one challenge server; 
 (ii) transmit said response to said challenge, collected by said least one input/output (I/O) hardware component, to said at least one challenge server. 
 
   
     
     
         2 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said at least one input/output (I/O) hardware component is selected from the group consisting of: a graphical user interface (GUI), a voice user interfaces (VUI), a display, a screen touch, a computer keyboard, a virtual keyboard, a drawing pad, a speech-to-text conversation engine, a microphone, a computer pointing device. 
     
     
         3 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said computer-readable storage medium of said challenge server is further configured to store at least one member selected from the group consisting of:
 (a) a list of currently available participants;   (b) a plurality of optional/non-verified responses a set of rules   (c) at least one set of rules for constructing a statistically reliable sample from said list of currently available participants;   (d) at least one set of rules for constructing a challenge from said censored/filtered incoming message and said plurality of optional/non-verified responses;   (e) a set of rules for determining that said responses collected by said challenge server attain credibility above a predefined threshold level.   
     
     
         4 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said computer-readable storage medium of said challenge server is further configured to store a set of rules for conducting an offline non-real-time challenge, to a group of participants larger than said statistically reliable sample, in order to re-verify said credibility of said verified response among said larger group of participants. 
     
     
         5 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said computer-readable storage medium of said challenge server is further configured to store a set of rules for recalculating said metrics of said registered participants. 
     
     
         6 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said NLP server is further configured to perform machine learning of said verified responses. 
     
     
         7 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said NLP server is further configured for updating said computer-readable storage medium of said NLP server with said verified responses. 
     
     
         8 . The system for operating a real-time automated conversational agent, as in  claim 1 , wherein said set of rules for constructing said challenge on said computer-readable storage medium of said challenge server further comprises a set of rules for construction of multiple challenges for ensuring adequate challenge construction by said challenge server, wherein a portion of interest in said incoming message is identified and a plurality of challenges is constructed by said challenge server. 
     
     
         9 . A computer network implemented method, of operating a real-time automated conversational agent, said method comprises:
 (a) collecting an incoming message, inputted by a user, by a terminal interface of a conversational agent;   (b) transmitting said incoming message, collected at said terminal interface, from said conversational agent to a natural language programming (NLP) server;   (c) determining by said NLP server that there is no a verified response, among a plurality of predefined verified responses in a database of said NLP server, matching said incoming message with a matching score exceeding a predefined matching score threshold;   (d) generating a filtered incoming message by said NLP server by performing at least type of editing selected from the group consisting of: censoring said incoming message and filtering said incoming message, so as to remove and/or obstruct any personal sensitive information of said user;   (e) transmitting said filtered incoming message from said NLP server to a challenge server;   (f) constructing by said challenge server a challenge from said filtered incoming message, received from said NLP server;   (g) transmitting said challenge from said challenge server to at least one participant computing device;   (h) communicating said challenge by an output hardware component of said at least one participant computing device to at least one participant;   (i) collecting by an input hardware component of said at least one participant computing device at least one response of said at least one participant to said challenge;   (j) transmitting said at least one response of said at least one participant to said challenge, from said at least one participant computing device to said challenge server;   (k) determining that said response received by said challenge server has a credibility score exceeding a predefined threshold level, by at least one server selected from the group consisting of: said NLP server and said challenge server;   (l) transmitting a verified response with said credibility score exceeding said predefined threshold level, from said NLP server to said terminal interface, of said conversational agent.   
     
     
         10 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises at least one step selected from the group consisting of:
 (a) establishing an up-to-date list of currently available participants;   (b) defining a statistically reliable sample from said list of currently available participants;   (c) transmitting said filtered incoming message and a plurality of verified responses from said database of said NLP server to said challenge server.   
     
     
         11 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises conducting an offline non-real-time challenge to a group of participants larger than said statistically reliable sample, in order to re-verify said credibility of said verified response among said larger group of participants. 
     
     
         12 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises method of operating a real-time automated conversational agent, as in  claim 9 , further comprises recalculating said metrics of said registered participants. 
     
     
         13 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises performing NLP machine learning of said verified responses. 
     
     
         14 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises updating a computer-readable storage medium of said NLP server with said verified responses. 
     
     
         15 . The computer network implemented method of operating a real-time automated conversational agent, as in  claim 9 , further comprises construction of multiple challenges for ensuring adequate challenge construction by said challenge server, wherein a portion of interest in said incoming message is identified and a plurality of challenges is constructed by said challenge server. 
     
     
         16 . A computer-readable storage medium of a challenge server in a system operating a real-time automated conversational agent, having computer-executable instructions stored thereon, which when executed by a computer micro-processor of said challenge server, causing said challenge server to:
 (a) receive a filtered incoming message, which has been subjected by an NLP server to at least one type of editing selected from the group consisting of: censoring an incoming message and filtering said incoming message;   (b) construct a challenge from said incoming message, by incorporating at least a portion of said filtered incoming message;   (c) transmit said challenge to at least one participant computing device, to be further communicated via an output hardware component of said at least one participant computing device to a participant;   (d) receive a response to said challenge collected via an output hardware component of said at least one participant computing device from said participant;   (e) determining that a response has a credibility score exceeding a predefined threshold level;   (f) transmitting a verified response that has said credibility exceeding said predefined threshold level to said NLP server.   
     
     
         17 . The computer-readable storage medium of a challenge server, as in  claim 16 , further comprises at least one set of computer-executable instructions stored thereon, which when executed by said computer micro-processor of said challenge server, causing said challenge server to perform at least one operation selected from the group consisting of:
 (a) establishing an up-to-date list of currently available participants;   (b) defining a minimally reliable statistical sample from said list of currently available participants;   (c) selecting a statistically reliable sample of participants from said list of currently available participants;   (d) receiving said filtered incoming message and a plurality of verified responses from said database of said NLP server;   (e) conducting an offline non-real-time challenge to a group of participants larger than said statistically reliable sample, in order to re-verify said credibility of said verified response among said larger group of participants;   (f) recalculating said metrics of said registered participants.   
     
     
         18 . The computer-readable storage medium of a challenge server, as in  claim 16 , further comprises at least one set of computer-executable instructions stored thereon, which when executed by said computer micro-processor of said challenge server, causing said challenge server to establish an up-to-date list of said currently available participants by performing at least one operation selected from the group consisting of:
 (a) actively inquiring computing devices associated with registered participants and collecting the notifications from said computing devices;   (b) retrieving from a database of said challenge server said up-to-date list of currently available participants, wherein said list is compiled independently by said challenge server, by collecting notifications periodically sent by active computing devices of said currently available participants.   
     
     
         19 . The computer-readable storage medium of a challenge server, as in  claim 16 , further comprises computer-executable instructions stored thereon for construction of multiple challenges for ensuring adequate challenge construction by said challenge server, wherein a portion of interest in said incoming message is identified and a plurality of challenges is constructed by said challenge server.