Training of chatbots from corpus of human-to-human chats

1. A computer-based method to prepare initial training patterns for an Artificial Intelligence (Al) based chatbot comprising:

accessing, by a computer processor, a corpus in computer-readable memory having a plurality of digital weighted conversation models of text-based interlocutory conversations, wherein each digital weighted conversation model contains a plurality of annotations and a plurality of paths, and wherein each of the plurality of the paths in each of the plurality of the digital weighted conversation models is associated with a weight;

selecting, by a computer processor, a plurality of the text-based interlocutory conversations which meet at least one annotation value and in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models;

converting, by a computer processor, each weight associated with each of the plurality of the selected conversations into initial training pattern values according to at least one Artificial Intelligence (Al) based automated conversational agent system; and

exporting, by a computer processor, the training pattern values to at least one Artificial Intelligence (Al) based automated conversational agent system platform.

2. The computer-based method as set forth in claim 1 further comprising, subsequent to the exporting, receiving, by a computer processor, and subsequent to one or more digital weighted conversation models being added to the corpus, repeating the steps of accessing, selecting, converting and exporting to update the training pattern values.

3. The computer-based method as set forth in claim 1 further comprising, subsequent to the exporting, receiving, by a computer processor, and subsequent to one or more digital weighted conversation models being removed from the corpus, repeating the steps of accessing, selecting, converting and exporting to update the training pattern values.

4. The computer-based method as set forth in claim 1 wherein the accessing further comprises retrieving the plurality of digital weighted conversation models from a corpus stored at least partially in a database.

5. The computer-based method as set forth in claim 1 wherein the accessing further comprises receiving the plurality of digital weighted conversation models from a corpus at least partially via a digital data network.

6. The computer-based method as set forth in claim 1 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to an intent annotation value.

7. The computer-based method as set forth in claim 1 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to a topic annotation value.

8. The computer-based method as set forth in claim 1 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a conversation according to an outcome annotation value.

9. The computer-based method as set forth in claim 1 wherein the selecting further comprises providing a prompt to a user on a user interface device of a computer to select one or more illustrated paths.

10. The computer-based method as set forth in claim 9 wherein the selecting of a plurality of the text-based interlocutory conversations in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models further comprises displaying a representation the paths and weights to enable the user to select paths having a minimum weight.

11. The computer-based method as set forth in claim 9 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a displayed conversation according to a displayed intent annotation value.

12. The computer-based method as set forth in claim 9 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a displayed conversation according to a displayed topic annotation value.

13. The computer-based method as set forth in claim 9 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a displayed conversation according to a displayed outcome annotation value.

14. The computer-based method as set forth in claim 1 wherein the Al-based automated conversational agent system further comprises one or more systems selected from the group consisting of a chatbot, an interactive voice response (IVR) systems, and a voicebot, and wherein the training pattern comprises one or more outputs selected from the group consisting of a sample prompt, an entity, a flow, an intent, an utterance, an outcome, a speech act, a turn grouping, a topic, a phase, a sentiment, a clarifying question, a clarifying statement, a conversation summary, a promise, a next-best turn, a next-best action, an agent activity, a business process, and an event.

15. A computer program product to prepare initial training patterns for an Artificial Intelligence (Al) based chatbot comprising:

a computer-readable memory which is not a transitory propagating signal per se; and

one or more program instructions embodied by the computer-readable memory configured to, when executed by a computer processor, cause the processor to:

access a corpus in computer-readable memory having a plurality of digital weighted conversation models of text-based interlocutory conversations, wherein each digital weighted conversation model contains a plurality of annotations and a plurality of paths, and wherein each of the plurality of the paths in each of the plurality of the digital weighted conversation models is associated with a weight;

select a plurality of the text-based interlocutory conversations which meet at least one annotation value and in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models;

convert each weight associated with each of the plurality of the selected conversations into initial training pattern values according to at least one Artificial Intelligence (Al) based automated conversational agent system; and

export the training pattern values to at least one Artificial Intelligence (Al) based automated conversational agent system platform.

16. The computer program product as set forth in claim 15 wherein the program instructions further comprise program instructions to, subsequent to the exporting, receiving, and subsequent to one or more digital weighted conversation models being added to the corpus, repeat the instructions of accessing, selecting, converting and exporting to update the training pattern values.

17. The computer program product as set forth in claim 15 wherein the program instructions further comprise program instructions to, subsequent to the exporting, receiving, and subsequent to one or more digital weighted conversation models being removed from the corpus, repeat the instructions of accessing, selecting, converting and exporting to update the training pattern values.

18. The computer program product as set forth in claim 15 wherein the accessing further comprises retrieving the plurality of digital weighted conversation models from a corpus stored at least partially in a database.

19. The computer program product as set forth in claim 15 wherein the accessing further comprises receiving the plurality of digital weighted conversation models from a corpus at least partially via a digital data network.

20. The computer program product as set forth in claim 15 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to an intent annotation value.

21. The computer program product as set forth in claim 15 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to a topic annotation value.

22. The computer program product as set forth in claim 15 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a conversation according to an outcome annotation value.

23. The computer program product as set forth in claim 15 wherein the selecting further comprises providing a prompt to a user on a user interface device of a computer to select one or more illustrated paths.

24. The computer program product as set forth in claim 23 wherein the selecting of a plurality of the text-based interlocutory conversations in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models further comprises displaying a representation the paths and weights to enable the user to select paths having a minimum weight.

25. The computer program product as set forth in claim 23 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed intent annotation value.

26. The computer program product as set forth in claim 23 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed topic annotation value.

27. The computer program product as set forth in claim 23 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed outcome annotation value.

28. The computer program product as set forth in claim 15 wherein the Al-based automated conversational agent system comprises one or more systems selected from the group consisting of a chatbot, an interactive voice response (IVR) systems, and a voicebot, and wherein the training pattern comprises one or more outputs selected from the group consisting of a sample prompt, an entity, a flow, an intent, an utterance, an outcome, a speech act, a turn grouping, a topic, a phase, a sentiment, a clarifying question, a clarifying statement, a conversation summary, a promise, a next-best turn, a next-best action, an agent activity, a business process, and an event.

29. A system for preparing initial training patterns for an Artificial Intelligence (Al) based chatbot comprising:

at least one computer processor configured to execute program instructions;

a computer-readable memory which is not a transitory propagating signal per se; and

one or more program instructions embodied by the computer-readable memory configured to, when executed by a computer processor, cause the processor to:

access a corpus in computer-readable memory having a plurality of digital weighted conversation models of text-based interlocutory conversations, wherein each digital weighted conversation model contains a plurality of annotations and a plurality of paths, and wherein each of the plurality of the paths in each of the plurality of the digital weighted conversation models is associated with a weight;

select a plurality of the text-based interlocutory conversations which meet at least one annotation value and in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models;

convert each weight associated with each of the plurality of the selected conversations into training pattern values according to at least one Artificial Intelligence (Al) based automated conversational agent system; and

export the training pattern values to at least one Artificial Intelligence (Al) based automated conversational agent system platform.

30. The system as set forth in claim 29 wherein the program instructions further comprise program instructions to, subsequent to the exporting, receiving, and subsequent to one or more digital weighted conversation models being added to the corpus, repeat the instructions of accessing, selecting, converting and exporting to update the training pattern values.

31. The system as set forth in claim 29 wherein the program instructions further comprise program instructions to, subsequent to the exporting, receiving, and subsequent to one or more digital weighted conversation models being removed from the corpus, repeat the instructions of accessing, selecting, converting and exporting to update the training pattern values.

32. The system as set forth in claim 29 wherein the accessing further comprises retrieving the plurality of digital weighted conversation models from a corpus stored at least partially in a database.

33. The system as set forth in claim 29 wherein the accessing further comprises receiving the plurality of digital weighted conversation models from a corpus at least partially via a digital data network.

34. The system as set forth in claim 29 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to an intent annotation value.

35. The system as set forth in claim 29 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least annotation value further comprises selecting a conversation according to a topic annotation value.

36. The system as set forth in claim 29 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value further comprises selecting a conversation according to an outcome annotation value.

37. The system as set forth in claim 29 wherein the selecting further comprises providing a prompt to a user on a user interface device of a computer to select one or more illustrated paths.

38. The system as set forth in claim 37 wherein the selecting of a plurality of the text-based interlocutory conversations in which at least one path meets at least one weight threshold according to the plurality of digital weighted conversation models further comprises displaying a representation the paths and weights to enable the user to select paths having a minimum weight.

39. The system as set forth in claim 37 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed intent annotation value.

40. The system as set forth in claim 37 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed topic annotation value.

41. The system as set forth in claim 37 wherein the selecting of a plurality of the text-based interlocutory conversations which meet at least one annotation value comprises selecting a displayed conversation according to a displayed outcome annotation value.

42. The system as set forth in claim 29 wherein the Al-based automated conversational agent system comprises one or more systems selected from the group consisting of a chatbot, an interactive voice response (IVR) systems, and a voicebot, and wherein the training pattern comprises one or more outputs selected from the group consisting of a sample prompt, an entity, a flow, an intent, an utterance, an outcome, a speech act, a turn grouping, a topic, a phase, a sentiment, a clarifying question, a clarifying statement, a conversation summary, a promise, a next-best turn, a next-best action, an agent activity, a business process, and an event.

43. An improved data storage and retrieval system for a computer memory, comprising:

means for configuring said computer memory according to a weighted conversation model data structure, said weighted conversation model data structure comprising:

at least one conversation topic;

a plurality of recorded conversation paths associated with each conversation topic; and

a plurality of recorded interlocutor turns associated with each conversational path within each topic;

at least one weight value associated with each of the plurality of recorded interlocutor turns, each of the recorded paths, or both recorded turns and recorded paths; and

means for transforming the conversation model data structure into initial training pattern values according to at least one Artificial Intelligence (Al) based automated conversational agent system.

44. The improved data storage and retrieval system as set forth in claim 43 wherein a conversational path is designated in the weighted conversation model data structure as a dominant path according to at least one weight value in the weighted conversation model data structure.

45. The improved data storage and retrieval system as set forth in claim 43 wherein the weighted conversation model data structure further comprises one or more sub-paths within the plurality of recorded conversational paths.

46. The improved data storage and retrieval system as set forth in claim 43 wherein the each weight value associated with each of the plurality of recorded interlocutor turns is determined according to analysis of a corpus of a text-recorded interlocutor conversations.

47. The improved data storage and retrieval system as set forth in claim 43 wherein the Al-based automated conversational agent system comprises one or more systems selected from the group consisting of a chatbot, an interactive voice response (IVR) systems, and a voicebot, and wherein the training pattern comprises one or more outputs selected from the group consisting of a sample prompt, an entity, a flow, an intent, an utterance, an outcome, a speech act, a turn grouping, a topic, a phase, a sentiment, a clarifying question, a clarifying statement, a conversation summary, a promise, a next-best turn, a next-best action, an agent activity, a business process, and an event.