SYSTEM AND METHOD FOR AUTOMATIC CALL FLOW DETECTION

• US 20080304650A1
• Filed: 06/11/2007
• Published: 12/11/2008
• Est. Priority Date: 06/11/2007
• Status: Active Grant

• Current Assignee: Enghouse Interactive Incorporated (Enghouse Systems Limited)
• Original Assignee: Syntellect Incorporated (Enghouse Systems Limited)
• Inventors: Rush, James Bryan, Catlin, Jon
• Granted Patent: US 8,423,635 B2
• Time in Patent Office: Days
• US Class Current: 379/298
• CPC Class Codes: H04M 3/24 with provision for checking the normal operation
• CPC Class Codes: H04M 3/493 Interactive information services, e.g. directory enquiries ; Arrangements therefor, e.g. interactive voice response [IVR] systems or voice portals
• CPC Class Codes: H04M 3/4938 comprising a voice browser which renders and interprets, e.g. VoiceXML

Abstract

A method and exploration engine for automatically detecting a call flow path of a voice application having a plurality of dialog states, the method comprising the steps of: identifying, in an automated manner, the plurality of dialog states of the voice application through execution of the voice application; and identifying at least one relationship between the plurality of dialog states. The method may further comprise some of the following steps: creating a unique identifier (ID) for one of the plurality of dialog states, wherein the unique ID is based on information received when executing the voice application, wherein the information comprises attributes associated with one of the plurality of dialog states; storing the unique ID and other information associated with the dialog state; and graphically illustrating a call flow path of the voice application based on the unique ID and other information stored in the database.

Claims

1. A method for automatically detecting a call flow path of a voice application having a plurality of dialog states, said method comprising the steps of:

identifying, in an automated manner, said plurality of dialog states of said voice application through execution of said voice application; and

identifying at least one relationship between said plurality of dialog states.

2. The method of claim 1, wherein identifying said plurality of dialog states further comprises the steps of:

providing a signal to a web server running said voice application, wherein said signal initiates a call with said web server;

receiving a first set of information from said web server;

executing instructions in said first set of information and receiving a second set of information;

creating a unique identifier (ID) for one of said plurality of dialog states, wherein said unique ID is based on said first and second sets of information;

providing an input to said web server to cause said voice application to move to another dialog state; and

storing said unique ID and other information associated with said dialog state.

3. The method of claim 2, wherein said first and second sets of information comprises attributes associated with one of said plurality of dialog states;

and further comprising the step of graphically illustrating a call flow path of said voice application based on the unique ID and other information stored in said database.

4. The method of claim 3, wherein said attributes comprise at least one of grammars, audio files, text, tags, and commands;

and wherein said first set of information is in one of a VXML document, a CCXML document, and a SALT document.

5. The method of claim 2, further comprising the steps of:

discovering possible inputs associated with one of said plurality of dialog states, wherein said possible inputs further include invalid inputs;

storing said possible inputs in a work queue;

sending a signal to said web server that mimics a response that said voice application expects for a first one of said possible inputs; and

noting in a work queue that said first one of the possible inputs has been explored.

6. The method of claim 2, wherein said unique ID is a hash based on attributes associated with said first set of information.

7. The method of claim 1, further comprising the step of automatically identifying at least one of the following:

states where there is no way to get operator assistance;

menus that have more than a preset number of choices;

tasks that allow more than a preset number of retries;

tasks that involve too many inputs to complete;

hang-ups;

transfers;

infinite call loops, application behavior change, and prompts that are longer than a preset length of time.

11. The method of claim 1, wherein said unique ID is a hash based on attributes associated with said first set of information.

25. The method of claim 1, wherein identifying said plurality of dialog states further comprises the steps of:

initiating a simulated phone call with said voice application by reading a designated static page;

receiving a first set of information from said designated static page;

executing instructions in said first set of information and receiving a second set of information;

creating a unique identifier (ID) for one of said plurality of dialog states, wherein said unique ID is based on said first and second sets of information;

providing an input to said voice application to cause said voice application to move to another dialog state; and

storing said unique ID and other information associated with said dialog state.

26. The method of claim 1, further comprising the step of automatically identifying states that may have faults.

27. The method of claim 26, wherein said faults comprise at least one of the following:

invalid or missing application documents, invalid or missing grammar documents, invalid or missing audio documents, user defined error conditions, and situations where the voice application fails to respond within a specified time.

8. A method for detecting, in an automated manner, a call flow path of a voice application having a plurality of dialog states, the method comprising the steps of:

sending a signal to a web server running said voice application, wherein said voice application is one of;

a VXML application, a CCXML application, and a SALT application;

receiving, for each of said plurality of dialog states in said voice application, a document from said web server, said document comprising attributes, and said attributes comprising at least one of;

grammars, sound clips, text, and tags;

creating a unique identifier (ID) for each of said plurality of dialog states based on said attributes;

identifying for a particular dialog state of said plurality of dialog states, and based on at least one grammar in said document, a set of possible inputs expected by said particular dialog state;

sending one of said set of possible inputs to said web server and queuing remaining said set of possible inputs for later exploration; and

storing information related to each encountered dialog state of said plurality of dialog states, wherein said information comprises said unique ID and transition information indicating, for each individual dialog state, how said each individual dialog state is connected to other dialog states.

9. The method of claim 8, wherein transition information comprises one of said set of possible inputs and an error condition, and wherein said error condition comprises one of a non-response and an invalid response.

10. The method of claim 8, further comprising the steps of:

identifying for each dialog state, those dialog states that input to that dialog state and those dialog states that are output from that dialog state; and

creating an output representative of the call flow of said voice application.

12. The method of claim 8, further comprising the steps of:

graphically illustrating the call flow path of the voice application based on said unique ID and other information stored in said database.

13. The method of claim 8, further comprising the step of automatically identifying at least one of the following:

states where there is no way to get operator assistance;

menus that have more than a preset number of choices;

tasks that allow more than a preset number of retries;

tasks that involve too many inputs to complete;

hang-ups;

transfers;

infinite call loops, application behavior change, and prompts that are longer than a preset length of time.

14. A method for dynamically detecting a call flow path for a voice application, the method comprising the steps of:

using an application to access the voice application and to work through each dialog state in the voice application, wherein working through each dialog state is accomplished by;

identifying a current dialog state with a unique identifier (ID);

identifying for said current dialog state, based on grammars associated with said current dialog state, all possible inputs that will cause a transition from said current dialog state;

wherein said all possible inputs includes any input error conditions that will cause a transition from said current dialog state;

sending a signal representative of one of said all possible inputs and causing a transition from said current dialog state, wherein other non-sent possible inputs are saved for future exploration; and

repeating said identifying and sending steps for successive dialog states until one of the following two situations occur;

call termination, and call flow returns to a state that has already been visited; and

upon occurrence of one of those two situations, stopping that exploration and pursuing another exploration based on a different one of said all possible inputs not yet investigated;

recording information relative to states visited and transitions from one state to another; and

determining a complete call flow path based on the connection of each dialog state to other dialog states in said voice application.

15. The method of claim 14, wherein said unique ID is formed by forming, for said each dialog state, a fingerprint based on attributes associated with that dialog state, wherein said attributes comprise at least one of:

grammars, audio files, text, tags, and commands;

16. The method of claim 14, wherein said unique ID is based on a self identifying attribute comprising an application property set equal to a name for that state.

17. The method of claim 14, wherein said voice application is in one of a VXML document, a CCXML document, and a SALT document.

18. The method of claim 14, further comprising the step of automatically identifying at least one of the following:

states where there is no way to get operator assistance;

menus that have more than a preset number of choices;

tasks that allow more than a preset number of retries;

tasks that involve too many inputs to complete;

hang-ups;

transfers;

infinite call loops, application behavior change, and prompts that are longer than a preset length of time.

19. A system for analyzing a voice application, the system comprising:

a voice application and a first web server configured to run said voice application;

an IVR in communication with said first web server; and

an exploration engine, wherein said exploration engine comprises an exploration server running a program configured to communicate with said first web server by sending and receiving information to and from said first web server as if said exploration engine were said IVR, wherein said exploration engine is further configured to automatically detect a call flow of said voice application running on said first web server.

21. The system of claim 19, wherein said voice application is one of a VXML application, a CCXML application, or a SALT application.

22. The system of claim 19, wherein said exploration engine is configured to create unique identifiers (IDs) for each dialog state in said voice application based on at least one of:

attributes associated with a dialog state, and one of self identifying VXML, CCXML, and SALT properties.

23. The system of claim 19, wherein said exploration engine is configured to:

identify for a current dialog state, based on grammars associated with said current dialog state, all possible inputs that will cause a transition from said current dialog state;

wherein said all possible inputs includes any input error conditions that will cause a transition from said current dialog state;

send a signal representative of one of said all possible inputs and cause a transition from said current dialog state, wherein other non-sent possible inputs are saved for future exploration; and

repeat said identifying and sending for successive dialog states until one of the following two situations occur;

call termination, and call flow returns to a state that has already been visited; and

upon occurrence of one of those two situations, stop that exploration and pursue another exploration based on a different one of said all possible inputs not yet investigated.

24. The system of claim 19, wherein said exploration engine is configured to provide output that facilitates creating a graphical illustration of said call flow of said voice application.