WO2001038962A1

WO2001038962A1 - Voice activated hyperlinks

Info

Publication number: WO2001038962A1
Application number: PCT/US1999/028004
Authority: WO
Inventors: Michael J. Polcyn
Original assignee: Intervoice Limited Partnership
Priority date: 1999-11-23
Filing date: 1999-11-23
Publication date: 2001-05-31
Also published as: AU2474300A; EP1240579A1

Abstract

The inventive system provides a user with a plurality of function choices based upon the communication device of the user. The system includes a device specific variable state transition model that formats the system functions into the function choices based upon a device of the user. Thus, if the user has a personal computer or other device with a large screen, the functions are formatted into a single level so that all of the functions are displayed simultaneously. If the user has a TDD or PDA or other device with a small screen, the functions are formatted into two or more levels, depending upon the screen size, so that only a manageable portion of the functions are displayed simultaneously. If the system lacks a screen, then the functions are formatted into a plurality of levels, each level having only a small number of function choices. The system uses a data base having key words that describe each function, and permutation lists that have expanded lists of phrases that are synonymous with the key words. The system uses a dialogue engine to compare the commands from the user with the key words and permutation lists to identify the requested function.

Description

VOICE ACTIVATED HYPERLINKS

TECHNICAL FIELD OF THE INVENTION

This invention relates in general to automated telephony systems, and in specific to

voice activated menu command systems.

BACKGROUND OF THE INVENTION

The existing prior art uses a scripting tool and an execution environment for interactive

voice mail audio text that defines scripts in a very linear fashion. For instance, a script

definition could be "wait for a telephone ring." When the phone the rings, then answer;

prompt the caller with a predefined set of choices; receive the selected choice via either DTMF

or voice recognition, and then perform the selected function.

FIGURE 7 is a diagram of flow of typical prior art interactive voice recognition/DTMF

program 700, i.e., "for information regarding ... speak or press 1 now". Typical programs

follow a sequence of events: telephone ring and answer 701, prompt and greet the caller 702,

present some finite number of choices or functions 704 in a predetermined fashion 703, and

receive the user input 705 either through a DTMF input or a voice command. Then perform

the selected function 706, and so forth.

Whether the system uses a voice recognition choice or a numeric DTMF input, some

function is invoked, and again in a very predetermined manner, the caller is returned to the

prompt 702 at the beginning of the program. The prior art system is very inflexible in that the

flow of the choices or functions, and the presentation of those choices, are tied very tightly to

the functions in the function definition.

The limitation of the prior art is that it is very rigid and only able to present a finite

number of options to the caller in a predetermined fashion. Moreover, the logic of the

presentation to the caller is preset, and thus is limited by the creativity of the system

programmer. SUMMARY OF THE INVENTION

The invention is to decouple the invocation and execution of functions from the

manner in which they are selected. For instance, in a listing of twenty-five functions in a voice

menu, via DTMF, no more than three or four choices are presented to the caller using DTMF

at any given menu step in the call. This is because of ergonomic or human limitations.

However, if the same functions are displayed on a screen to the caller, for example via an

internet session, there is no reason that all twenty-five of the functions could not be presented

at the same time. By decoupling the functions from the method by which those functions are

invoked, the functions can be leveraged across many different access devices and technologies.

For example, a twenty-five function system can be invoked from voice menus by

grouping certain sets of the functions through a choice engine or choice selection mechanism.

However, a graphical interface allows the presentation of many more options, if not all

options, to the caller.

Another advantage is that both natural language voice recognition and screen based

interfaces allow the user to exit to the top level choices in the choice engine once the user

completes a function. While in the prior art, the user is very restricted as to allowable

navigation through functions because of how functions are nested. Specifically, a user may

only move up one level at a time. Thus, all possible navigation paths must be pre-defined by

the programmer. The invention allows for dynamic addition and deletion of both function and

access methods. Decoupling of the functions allows for a fully webbed navigation of the

system. The user can go from function 1, directly to function 2, then to function 3 without

necessarily having to return to the top level menu or choice engine. This provides more

efficient navigation to desired functions. A technical advantage of the present invention is to promote re-use of defined

functions.

Another technical advantage of the present invention is to allow the addition of new

access methods independent of functions.

A further technical advantage of the present invention is to allow the addition of new

functions independent of defined functions.

A further technical advantage of the present invention is to have a device specific

variable state transition model that formats the different functions into function choices based

upon a device of the user.

A further technical advantage of the present invention is to have a key word data base

comprising key words used to describe each function in the system, and to have a plurality of

permutation lists comprising expanded lists of phrases that are synonymous with the key

words.

A further technical advantage of the present invention is to have a dialogue engine for

parsing voice input from the user into segments and examining each segment for key words

and expanded phrases for identifying the requested function.

A further technical advantage of the present invention is to have a learning operation

for ascertaining the meaning and the function association for an unknown phrase received

from the user, and amending the key word data base and permutation lists to include the

unknown phrase.

A further technical advantage of the present invention is to have the permutation lists

predefined by applying thesaurus and lexicon applications to the key word data base. A further technical advantage of the present invention is to use a format controller for

determining characteristics of the device, and arranging the functions into function choices

based upon limitations of the device.

The foregoing has outlined rather broadly the features and technical advantages of the

present invention in order that the detailed description of the invention that follows may be

better understood. Additional features and advantages of the invention will be described

hereinafter which form the subject of the claims of the invention. It should be appreciated by

those skilled in the art that the conception and the specific embodiment disclosed may be

readily utilized as a basis for modifying or designing other structures for carrying out the same

purposes of the present invention. It should also be realized by those skilled in the art that

such equivalent constructions do not depart from the spirit and scope of the invention as set

forth in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of the present invention, and the advantages

thereof, reference is now made to the following descriptions taken in conjunction with the

accompanying drawings, in which:

FIGURE 1 depicts the inventive hyperlink system;

FIGURE 2 depicts a telephone access decision tree;

FIGURE 3 depicts a screen phone or PDA decision tree;

FIGURE 4 depicts a personal computer decision tree;

FIGURE 5 depicts the logical groupings of related functions;

FIGURE 6 depicts the device specific variable state transition

model; and FIGURE 7 depicts a flow diagram of a prior art program.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

FIGURE 1 depicts the inventive hyperlink system 100. The system 100 interacts with

various types of communications equipment, including a personal computer or PC 101. The

interface with the PC 101 may be some proprietary PC interface, a graphical interface or

possibly an internet browser interface. Another type of communication equipment may be a

TDD (telecommunication device for the deaf) screen phone 102, which is a telephone device

with a VDT screen for displaying text or images. The screen phone may be a PDA (personal

data assistant). A further type of communication equipment that interacts with the system is a

telephone interface 102. This interface 102 may be to a standard DTMF telephone, but this

interface may also be to many other types of telephonic devices including faxes and other

devices that allow for dialogue. The device specific variable state transition model 104 acts as

an interpretive assembler in that the model selects and arranges the various functions 10, 11,

12, 99 based upon the requesting device, i.e., the computer 101, the screen phone 102, and the

telephone 103. The functions 10, 11, 12, 99 represent the various options and choices that a

system user has when operating the system. The system 100 of FIGURE 1 can be divided into

three sections, the devices 101, 102, 103, the device specific variable state transition model

104, and the functions 10, 11, 12, 99. This system 100 that will present choices and options,

or functions 10, 11, 12, 99 to the user in a variable manner based on the devices type 101, 102,

103 that the user is using to access the system, without forcing the system programmer to

anticipate all the different devices that may attempt to access the system when the programmer

develops the code for the system.

An example of the system 100 is a banking system. Function 10, for instance, in the

banking application, may be a savings account menu. Function 11 may be a checking account menu; function 12 may be a credit card account menu; and another function may be a loan

application function. On a touch tone telephone for instance, the user would be offered these

different functions as voice choices with DTMF input. Specifically, push one for savings,

push two for checking and so on. If the user entered the system with a PC, however, the

different functions may be presented in a series of text choices or graphical icons. The

inventive system 100 relieves the programmer of having to anticipate the navigation portion,

i.e., device specific portion for determining the different functions and how these functions

access the data and format the data for output.

FIGURE 2 depicts the output menu or tree structure 200 from the device specific

variable state transition model when a telephone 103 is used to access the system 100. As an

example, the tree structure 200 shown in FIGURE 2 defines 20 functions. These functions are

the same functions shown in FIGURE 1. The caller first receives a greeting 201 and a

message defining the top level of input choices, shown as functions FI 10, F2 11, and F3 12.

Selection of one of these choices reveals a second layer of functions. For example, if FI 10 is

selected, the caller is presented with further options of F4, F5 and F6. However, the caller is

not presented all of the different functions F1-F20 because of ergonomic limitations.

The device specific variable state transition model 104 has recognized the limitations

of the callers telephone device 103 and has appropriately formatted the presentation of the

functions. However, this has limited the users movement around the tree in that a caller

cannot move directly from F9 203 to FI 1 204 without moving back up the tree to FI 10. This

works similarly for a move from F9 203 to F19 205; except in this instance, the caller must

return to the greeting 201. FIGURE 3 depicts the output menu or tree structure 300 from the device specific

variable state transition model when a screen telephone or PDA 102 is used to access the

system 100. As an example, the tree structure 300 shown in FIGURE 3 defines 20 functions;

these functions are the same functions shown in FIGURES 1 and 2. First, the caller is shown a

greeting 301 and a message defining the top level of input choices, shown as functions FI 10,

F2 11, and F3 12. Selection of one of these choices reveals a second layer of functions.

As compared to FIGURE 2, however, the device specific variable state transition

model 104 has recognized that the caller is using a screen telephone device or PDA 102, which

has a small screen that is capable of displaying several lines of text or a few graphical icons,

and thus has appropriately formatted the presentation of the functions so that some, but not all,

are displayed at the same time. A small or diminutively sized screen is capable of displaying

some of the functions on the screen simultaneously, but not all of the functions. In this

instance, the user can freely move between F9 203 and FI 1 204 without moving back up the

tree to FI 10. But the caller cannot move from F9 203 to F19 205 without returning to the

greeting 201 because of the small screen size.

FIGURE 4 depicts the output menu or tree structure 400 from the device specific

variable state transition model when a personal computer 101 or other large screen device is

used to access the system 100. A large or substantially sized screen is capable of displaying

all or nearly all of the functions on the screen simultaneously. As an example, the tree

structure 400 shown in FIGURE 4 defines 20 functions, these functions are the same functions

shown in FIGURES 1-3. First, the caller is shown a greeting 401 and then a text message or

graphical icons that define all of the functions FI to F20. As compared to FIGURES 2 and 3, however, the device specific variable state

transition model 104 has recognized that the caller is using personal computer 101, which has

a large screen that is capable of displaying many lines of text or many graphical icons, and

thus has appropriately formatted the presentation of the functions so that all of the functions

are displayed at the same time. In this instance, the user can freely move between F9 203 and

FI 1 204, as well as, from F9 203 to F19 205 without returning to the greeting 401.

FIGURE 4 also depicts the tree structure 400 when using a natural language voice

recognition system which means that if, for example, the caller using a telephone 103, screen

phone 102, or PC base phone 101 utters a phrase such as "I want my savings account balance,"

the system would automatically invoke function FI 10. Using such as system would put all of

the functions FI to F20 at the disposal of the user at the same time.

FIGURE 5 depicts the logical groupings of related functions. This shows the functions

and their relationship to each other and describes the hierarchy or priority of the particular

functions. For instance, there are logical groupings in the banking system example previously

discussed, such as types of accounts, e.g. savings, checking, credit cards, and loans, or types of

transactions, e.g. balance inquiries, withdrawals, deposits, and transfers.

These groupings can be arranged into particular sets of functions that are roughly

equivalent in precedence like withdrawals and deposits. Moreover, the different elements of

the different groups can be combined to form various associated groups, such as all of the

types of transactions can be combined with each of the account types. For example, savings

can be combined with balance inquiries, withdrawals, deposits, and transfers to form a savings

account transaction grouping. FIGURE 5 is a diagrammatic representation of these types of groups, but this could be

described physically, graphically or in some text definition, which would show some logical

grouping of these functions. FIGURE 5 function FI 10 is associated with F2 and F3 in logical

group B, but is also associated with F3, F4 and F9 in logical group A. This grouping of

functions would support a program that generates the actual navigation structures of the access

methods in determining how the choices are presented according to the different types of

devices. In the case of touch-tone telephone input, it would use the information provided here

in FIGURE 5 to set a hierarchial menu structure shown in FIGURE 2. In the case of access by

a screen phone device, the system would provide some logical grouping choices in a graphical

representation as shown in FIGURE 3, either as a logical grouping of icons or text menu

choices. Thus, the grouping diagram of FIGURE 5 provides the necessary information to

express the association between the different functions.

FIGURE 6 depicts the device specific variable state transition model 104 that uses the

groups of functions shown in FIGURE 5 and arranges the functions according to the device

type, as shown in FIGURES 2-4.

FIGURE 6 includes a series of functions 601. These functions are the same functions

depicted in previous figures. In keeping with the example of banking functions used

previously, an example of a discrete function would be, for instance, a cleared checks function

602. This function is account specific, i.e., it is only applicable to the checking account.

However, other functions, such as a balance inquiry function, would not be account specific

and could be re-used or called by other functions, i.e, checking account, savings account, or

credit card menu functions. The cleared check function 602 involves looking/accessing a host system for a given

account, then retrieving the details of the last several checks that have cleared through that

account, and then formatting this information in a manner that is presentable to the caller, i.e.,

for a voice activated dialogue transaction as shown in FIGURE 2, then speaking that

information to the caller.

The cleared check function 602 has a certain number of required inputs 604; for

example, there may be an account number or a security key like a pin number. Other inputs

include parameters that are passed to the function from the user; for instance, the number of

checks to be examined, such as the last five checks or the last ten checks. The cleared check

function 602 also has a certain number of methods 605 for performing its function with the

input 604. For example, within the function there are data access methods, data manipulation

methods and data formatting methods. The function 602 has an output section 606 for sending

the output back to the caller in a format acceptable to the device being used by the caller.

The cleared check function 602 has a logical group section 603 that defines links and

associations to other functions as depicted in FIGURE 5. The function 602 also has the ability

to invoke other functions 607 when necessary. For example, cleared check function 602 can

invoke a security function if the pin number is incorrectly entered.

The cleared check function 602 can be invoked by a user 608 in a number of different

ways, depending upon the user's device 614. For instance, in a non-voice manner or graphical

format 609, the function may appear as an icon to be activated by the caller. Another way is

via voice dialogue 610, although this avenue is problematic because of the number of ways

spoken language can describe the desired function. A user 608 could say "give me the history of the checking account", or "look at (or examine, retrieve, check, verify etc.) the last ten

checks", and so on.

Thus, as a function 602 is developed for voice input action 610, a certain number of

key action words or phrases will be described for that function. The key action words or

phrases will be a starting point to describe that functionality. A further step, taken non-run

time or during the development of the function, is to run the key action phrases through a

lexicon-grammar engine that constructs the various permutations of the possible requests. If

enough of the key word phrases are defined up front, and given a sufficiently intelligent

grammar engine, then the dialogue engine 611 could determine the proper function that would

satisfy the user's 608 needs and match the user's input request.

The permutation lists 613, and the key word data base 612 feeds the dialogue engine

611. The dialogue engine 611 is the reactive mechanism that allows the user 608 to call or

otherwise verbally communicate with the system 100. The engine 611 parses a command

sentence or expression into words or phrases. Theses words and phrases are examined by the

engine for various key action phrases that are defined in the key word data base 612 and the

expanded permutations of those key words stored in the permutation lists 613. The dialogue

engine 611 would then determine which function is being invoked by the user from the data in

the key word data base 612 and the lists 613. In the example shown in FIGURE 6, the user

608 may have said, "what are the last ten checks that have cleared my account."

The format controller 615 organizes the functions into a hierarchical arrangement

depending upon the type of device 614 of the user 608. The information about the user's

device is obtained several ways. The most direct is that the format controller solicits the

device type from the user, i.e., push one for a touch tone phone, two for a screen phone, etc. Another way is for the format controller to compare the signal characters of the command with

a stored data base of such information to determine the type of device being used by the user.

Another way is for the format controller to send interrogatory signals to the device and have

the device respond as to its type. Once the user's device type is ascertained, the format

controller will display the functions as shown in FIGURES 2-4. The format controller works

with the dialogue engine to control the responses sent by the dialogue engine back to the user,

such that if full natural language voice recognition is being utilized by the user, the format of

the functions will appear as shown in FIGURE 4. However, if only a voice response unit is

being used (i.e., say one for savings), then the format of the functions as shown in FIGURE 2

would be used.

Each function 601 has other inputs, either from voice or non- voice. In the example of

FIGURE 6, one of the required inputs is an account number and pin. Now, if these inputs are

not provided, the dialogue engine 611 would not attempt to perform a function without having

required fields filled. The engine 611 would either terminate the function request or it may

invoke another function 607; for example, a security function would prompt the user to supply

the missing account number and pin number. Then the security function would verify these

numbers and fill in the data set of the invoking function 602. The security function, if needed,

could invoke other functions.

Beginning with a limited number of key action words and phrases, followed by a

process through grammar analysis, thesaurus support, lexicons, and dictionaries (generating

the various permutations of this relatively limited set of definitions), would allow for a more

flexible system and make a natural language voice recognizer much more useful. Thus, it would open up a much broader range of dialogue and more useful approach to natural

language dialogues between the user and the system.

The approach of this system is not independently defining a transaction engine here

with all of the permutations and then providing a limited finite number of them for use. But

rather, it describes the function and the general access methods, the data elements required,

and the key action words and phrases that a user would typically use and building all of the

permutations of that access method, automatically feeding the dialogue engine.

The use of key words and phrases has the same application in non-voice application as

in, for example, a screen phone. For a list selection of functions to be displayed, the choices

may be narrowed to a few; or even a single item would then be added dynamically to a menu

on a screen phone.

The dialogue engine is capable of learning new word/phrase associations for existing

functions. For example, if the dialogue engine receives a command or request, and the engine

is unable to determine the action or the function in the command because the choice of words

used by the user to convey the command is not located in either the key word data base or the

permutation lists, then the dialogue engine will ascertain the meaning of the command. The

engine will do this by either asking the user to re-define the command in some other terms or

consult built-in dictionaries, thesauruses, and the like to determine the most likely meaning of

the command, and then proceed accordingly. Then engine will then update its permutation

lists to contain the new reference from the command.

Because the functions are decoupled from the manner that the functions are invoked,

additional functions may be added by the system manager without requiring revision of the

existing functions. Only the key word data base and the permutation lists would have to be updated when a new function is added to the system. However, if the new function is going to

be invoked by existing functions or will be associated with particular logical groups, then

those particular, existing functions will have to be updated. Additionally, it may be possible

for the user to define user specific functions characterizing a scripting preference for the user.

These user defined functions would be formed from combining or modifying existing

functions.

Although the present invention and its advantages have been described in detail, it

should be understood that various changes, substitutions and alterations can be made herein

without departing from the spirit and scope of the invention as defined by the appended

claims.

Claims

WHAT IS CLAIMED IS:

1. A system for providing a user with a plurality of function choices, the system

comprising:

a device specific variable state transition model; and

a plurality of functions;

wherein the state transition model formats the functions into function choices based

upon a device of the user.

2. The system of claim 1, wherein:

the functions are associated into a plurality of logical groupings, wherein each logical

group comprises functions which have a related functionality.

3. The system of claim 2, wherein:

at least one function is associated with more than one logical group.

4. The system of claim 1, wherein:

the device is selected from the group consisting of: a personal computer, a telephone, a

screen telephone, a personal data assistant, and a TDD.

5. The system of claim 1, wherein:

the device has a substantially sized display screen, and the data is formatted into a

single level comprising all of the functions as function choices.

6. The system of claim 1, wherein:

the device has a diminutively sized display screen and the data is formatted into at least

two levels, each level comprising a portion of the functions as function choices.

7. The system of claim 1, wherein:

the device lacks a display screen and the data is formatted into a plurality of levels,

each level comprising a portion of the functions as function choices.

8. The system of claim 7, wherein:

the device is a telephone using a voice response system.

9. The system of claim 7, wherein:

the device is a telephone using a touch tone response system.

10. The system of claim 1, wherein:

the device lacks a display screen and the data is formatted into a single level

comprising all of the functions as function choices.

11. The system of claim 10, wherein:

the device is a telephone using a natural language voice recognition system.

12. The system of claim 1, wherein the transition model comprises:

a key word data base comprising key words used to describe each function; a plurality of permutation lists comprising expanded lists of phrases that are

synonymous the key words; and

a dialogue engine for receiving voice input action from the user, parsing the input

action into segments, searching the segments for key words and expanded lists of phrases,

identifying a requested function in the voice input action, and activating the requested

function.

13. The system of claim 12, wherein the dialogue engine comprises:

a learning operation for ascertaining a meaning and a function association for an

unknown phrase received from the user, and amending the key word data base and

permutation lists to include the unknown phrase.

14. The system of claim 12, wherein:

a format controller for determining characteristics of the device, arranging the

functions into function choices based upon limitations of the device.

15. The system of claim 14, wherein:

the format controller determines the characteristics of the device by querying the user.

16. The system of claim 14, wherein:

the format controller determines the characteristics of the device by interrogating the

device.

17. The system of claim 14, wherein:

the format controller determines the characteristics of the device by analyzing signal

characteristics of the input action.

18. The system of claim 12, wherein:

the permutation lists are predefined.

19. The system of claim 18, wherein:

the permutation lists are formed by applying a thesaurus application to the key word

data base.

20. The system of claim 18, wherein:

the permutation lists are formed by applying a lexicon application to the key word data

base.

21. The system of claim 18, wherein:

the permutation lists are formed by applying a lexicon application and a thesaurus

application to the key word data base.

22. The system of claim 1, wherein the transition model comprises:

a format controller for determining characteristics of the device, and arranging the

functions into function choices based upon limitations of the device.

23. The system of claim 22, wherein:

24. The system of claim 22, wherein:

device.

25. The system of claim 22, wherein:

characteristics of a command issued by the device from the user.

26. A variable state transition system for arranging a plurality of functions into a

presentation format based upon a device of a user of the system, the system comprising:

a key word data base comprising key words used to describe each function;

a plurality of permutation lists comprising expanded lists of phrases that are

synonymous with the key words;

function; and

functions into function choices based upon limitations of the device.

27. The system of claim 26, wherein the dialogue engine comprises:

unknown phrase received from the user, and amending the key word data base and

permutation lists to include the unknown phrase.

28. The system of claim 26, wherein:

the format controller determines the characteristics of the device by performing one of

the group of methods consisting of querying the user, interrogating the device, and analyzing

signal characteristics of the input action.

29. The system of claim 26, wherein:

the permutation lists are predefined.

30. A variable state transition system for arranging a plurality of functions into a

means for determining characteristics of the device;

means for arranging the functions into the presentation format based upon limitations

of the device to allow the user to select a requested function;

means for storing key words used to describe each function;

means for storing expanded lists of phrases that are synonymous with the key words;

means for receiving voice input action from the user;

means for parsing the input action into segments;

means for searching the segments for key words and expanded lists of phrases; and means for identifying the requested function in the voice input action.

31. The system of claim 30, wherein the means for identifying comprises:

means for ascertaining a meaning and a function association for an unknown phrase

received from the user, and amending the key word data base and permutation lists to include

the unknown phrase.

32. The system of claim 30, the means for determining comprises:

means for determining the characteristics of the device by querying the user.

33. The system of claim 30, the means for determining comprises:

means for determining the characteristics of the device by interrogating the device.

34. The system of claim 30, the means for determining comprises:

means for determining the characteristics of the device by analyzing signal

characteristics of a command issued by the device from the user.

35. The system of claim 30, wherein:

the expanded lists of phrases are predefined.