Expert conversation builder

• US 8,612,233 B2
• Filed: 01/05/2011
• Issued: 12/17/2013
• Est. Priority Date: 01/05/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Anand, Rangachari, Lee, Juhnyoung, Mueller, Erik T., Perez-Cedano, Melvin
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US12/985,050
• Publication Number:
  US 20120173243A1
• Time in Patent Office: 1,077 Days
• Field of Search: 704/275
• US Class Current: 704/275
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: H04M 2203/301 Management of recordings
• CPC Class Codes: H04M 2203/306 Prerecordings to be used during a voice call
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

An expert conversation builder contains a knowledge database that includes a plurality of dialogues having nodes and edges arranged as directed acyclic graphs. Users and authors of the system interface with the knowledge database through a graphical interface to author dialogues and to create expert conversations as threads traversing the node in the dialogues.

Claims

1. A method for conducting an expert conversation, the method comprising:

constructing a knowledge database comprising a dialogue repository comprising a plurality of dialogues, each dialogue representing a framework for creating at least one expert conversation and comprising a plurality of nodes and a plurality of edges extending between pairs of nodes; and

using the dialogue repository in combination with a runtime dialogue system executing on a computing system and in communication with the knowledge database to conduct the expert conversation between the runtime dialogue system and a user based on a given dialogue selected from the plurality of dialogues by presenting comments and questions derived from nodes in the given dialogue to the user, inputting data from the user in response to the presented comments and questions and identifying appropriate edges at each node based on the inputted data, wherein the expert conversation comprises a directed acyclic graph constructed from the nodes and edges of the selected dialogue.

2. The method of claim 1, wherein each node in each dialogue comprises questions to be presented to the user, a data entry point or combinations thereof.

3. The method of claim 1, wherein each node comprises an identifying text string comprising a subject for that node that identifies a region of the dialogue containing that node.

4. The method of claim 1, wherein each edge comprises an action undertaken in response to an answer provided by the user in response to a question contained at a node associated with that edge, to data submitted by users at a node associated with that edge or combinations thereof.

5. The method of claim 1, wherein at least one node in each dialogue comprises two or more associated edges extending from that node to two or more additional nodes.

6. The method of claim 1, wherein the nodes comprise multiple choice nodes, each multiple choice node comprising a single question with a plurality of answers, each answer associated with a given destination node linked to that multiple choice node by one of the plurality of edges.

7. The method of claim 1, wherein the nodes comprise text classifier nodes, each text classifier node comprising a text entry field and a trained text classifier to classify text entered by the user into one of a plurality of pre-defined classes where each class has an associated destination node linked to that text classifier node by one of the plurality of edges.

8. The method of claim 1, wherein the nodes comprise text search nodes, each teach search node comprising a single question and a plurality of text alternatives that are displayed in response to entered text, each text alternative matching at least a portion of the entered text.

9. The method of claim 1, wherein the nodes comprise decision nodes, each decision node comprising logic to determine a destination node linked to that decision node by one of the plurality of edges based on at least one of data and answers entered by the user at nodes other than that decision node.

10. The method of claim 1, wherein the step of constructing the knowledge database further comprises using a plurality of authors to collaboratively author the plurality of dialogues.

11. The method of claim 1, wherein the step of constructing the knowledge database further comprises:

restricting permission to modify specific portions of one or more dialogues to predetermined authors; and

tracking modifications to each dialogue with respect to authorship.

12. The method of claim 1, wherein the step of constructing the knowledge database further comprises using a web-based dialogue authoring tool running on a computing system to provide for author access to the dialogues in the knowledge database, the web-based dialogue authoring tool in communication with the knowledge database and at least one web browser.

13. The method of claim 1, wherein the step of using the dialogue repository and runtime dialogue system to conduct the expert conversation further comprises:

identifying one of the plurality of dialogues in the dialogue repository;

using the runtime dialogue system to present at least one of comments and questions from nodes within the identified dialogue to a user;

entering input from the user in response to the presented comments and questions; and

selecting edges to move between nodes based on the entered input.

14. The method of claim 1, further comprising using a graphical user interface to display to the user a current state of each expert conversation, a sequence of answers and entered data that constitutes the expert conversation, a list of instructions, a list of questions, a request for data entry box, a request for submission of comments or combinations thereof.

15. The method of claim 14, wherein the graphical user interface comprises a graphical display of the directed acyclic graph of each expert conversation.

16. The method of claim 1, further comprising:

changing previously inputted data from the user; and

modifying the directed acyclic graph of the expert conversation in response to the change in the previously inputted data.

17. The method of claim 1, further comprising logging all of the presented comments and questions from the nodes and the inputted data from the user for each expert conversation.

18. The method of claim 1, further comprising facilitating interactions between the user and additional users that participated in previous expert conversations based on the identified dialogue.

19. The method of claim 18, wherein the step of facilitating interactions comprises:

storing a set of conversation states for each previous expert conversation that was based on the identified dialogue;

associating at least one of the additional users with each stored set of conversation states;

assigning a success metric to each stored set of conversation states, each success metric measuring a degree of success accomplished by the previous expert conversation associated with the stored set of conversation states to which it has been assigned;

providing the user access to each set of stored conversation states; and

facilitating real-time conversations between the user and the additional users associated with the stored sets of conversation states.

20. The method of claim 1, wherein:

the expert conversation comprises at least one active thread, each active thread comprising a sequence of nodes traversed by the user in the expert conversation starting from one of a plurality of pre-defined start nodes within the identified dialogue; and

the step of using the dialogue repository and runtime dialogue system to conduct the expert conversation further comprises extending each active thread through the nodes and edges of the identified dialogue.

21. The method of claim 20, wherein the expert conversation comprises a plurality of simultaneously active threads and the method further comprises inactivating any active thread that traverses a node in the identified dialogue previously traversed by another active thread.

22. An expert conversation builder comprising:

a knowledge database comprising a dialogue repository comprising a plurality of dialogues, each dialogue representing a framework for creating at least one expert conversation and comprising a plurality of nodes and a plurality of edges extending between pairs of nodes;

a runtime dialogue system executing on a computing system and in communication with the knowledge database, the runtime dialogue system configured to conduct the expert conversation between the runtime dialogue system and a user based on a given dialogue selected from the plurality of dialogues by presenting comments and questions derived from nodes in the given dialogue to the user, inputting data from the user in response to the presented comments and questions and identifying appropriate edges at each node based on the inputted data; and

a web-based browser in communication with the runtime dialogue system and configured to display the comments and questions derived from nodes in the given dialogue to the user;

wherein the expert conversation comprises a directed acyclic graph constructed from the nodes and edges of the selected dialogue.

23. A method for conducting an expert conversation, the method comprising:

using a dialogue runtime system running on a computing system to select a sequence of nodes derived from a dialogue containing a plurality of nodes and edges between the nodes that is stored in a knowledge database in communication with the dialogue runtime system, wherein each node in the dialogue comprises one or more child nodes and at least one node in the dialogue comprises two or more parent nodes;

presenting a sequence of questions to a user in a web-based browser, each question derived from a node in the selected sequence of nodes;

selecting each subsequent node in the sequence based upon answers provided from the user in response to the present sequence of questions;

displaying a conversation thread to the user, the conversation thread comprising the selected sequence of nodes, and the expert conversation comprising the conversation thread;

wherein the conversation thread comprises a directed acyclic graph.

24. The method of claim 23, wherein at least one selected subsequent node in the selected sequence of nodes comprises a first previously traversed node that is a parent node of a second previously traversed node in the selected sequence, the first previously traversed node comprising an intermediate node in the dialogue between start nodes and end nodes in the dialogue.

25. A computer-readable storage medium containing a computer-readable code that when read by a computer causes the computer to perform a method for conducting an expert conversation, the method comprising:

constructing a knowledge database comprising a dialogue repository comprising a plurality of dialogues, each dialogue representing a framework for creating at least one expert conversation and comprising a plurality of nodes and a plurality of edges extending between pairs of nodes; and

using the dialogue repository in combination with a runtime dialogue system executing on a computing system and in communication with the knowledge database to conduct the expert conversation between the runtime dialogue system and a user based on a given dialogue selected from the plurality of dialogues by presenting comments and questions derived from nodes in the given dialogue to the user, inputting data from the user in response to the presented comments and questions and identifying appropriate edges at each node based on the inputted data, wherein the expert conversation comprises a directed acyclic graph constructed from the nodes and edges of the selected dialogue.