JP2569415B2 - Interactive processing controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dialogue processing control device used in a dialogue processing system for performing question answering, information retrieval, and the like, and assisting a user in solving a problem in an interactive manner with a computer.

[0002]

2. Description of the Related Art FIG.
1 shows a configuration diagram of a conventional interactive processing control device described in Japanese Patent Publication No. 31. In the figure, reference numeral 241 denotes a system for providing information according to a user request. 242 is a natural language interpreting means for interpreting the meaning of the input sentence from the user; 243 is a command generating means for generating a command to the information providing system 241 in accordance with the meaning of the input sentence by the natural language interpreting means 242; Topic management means 245 for managing a topic is a gaze target storage means for storing an object that the user is currently watching and its transition. In this interactive processing control device, the user's gazing point is captured by extracting the gazing object of the user, storing it in the gazing object storage unit 245, and referencing the input sentence when analyzing it thereafter. We try to have a smooth dialogue.

[0003]

However, in the conventional interactive processing control device, since the output from the information providing system is presented to the user as it is, it is difficult to solve the problem of the user interactively. In addition, there is a problem that a natural language sentence cannot be generated according to a situation that changes as the dialogue progresses.

[0004] The present invention has been made to solve such a problem, and includes a goal for expressing a target state of a user and a plan for expressing a sequence of actions necessary to achieve the goal. By considering information about such intentions, a question sentence requesting the presentation of information necessary for solving the problem of the system, an explanation that communicates the action that the user should perform in solving the problem and the reason why the action is necessary, It is an object of the present invention to provide a dialogue processing control device for generating a response corresponding to a situation that changes with the progress of a dialogue, with a confirmation sentence for checking the state of the user estimated by the system.

[0005]

SUMMARY OF THE INVENTION An interactive processing control device according to the present invention comprises a user goal estimating means for estimating a user's goal, that is, a user's target state, based on a semantic expression of a natural language sentence, System goal generating means for generating a system goal necessary to support the achievement of a goal, and plan goal structure generating means for generating a data structure expressing a sequence of actions to be performed to achieve the generated system goal A plan goal structure generating means for generating a plan goal structure expressing an action necessary to achieve the generated system goal, and a user or a system acquiring by a speech from the next system based on the generated plan goal structure Utterance goal generation means for generating goals that express knowledge that can be Utterance plan generating means for generating a sequence of actions required to generate the utterance plan, utterance plan executing means for generating a natural language sentence to be transmitted to the user, and next utterance predicting means for predicting a Japanese sentence to be input next by the user Things.

[0006]

In the dialogue processing control device according to the present invention, a question sentence, an explanation sentence, and a confirmation sentence according to a situation that changes with the progress of the dialogue are correctly generated by adding a plan goal structure generating means and an utterance goal generating means. be able to.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is an overall configuration diagram of an embodiment of a dialog processing control device according to the present invention. In FIG. 1, reference numeral 1 denotes a natural language text storage unit that stores a character string input using a keyboard or the like, 2 denotes a semantic expression generation unit that generates a semantic expression of the natural language text input by the natural language text storage unit 1,
Reference numeral 3 denotes a user goal estimating unit that estimates a user's goal, that is, a user's target state, based on the semantic expression generated by the semantic expression generating unit 2.

Reference numeral 4 denotes a belief base for storing knowledge about system users and system goals. Reference numeral 5 denotes a semantic expression and a user goal based on the semantic expressions and user goals obtained by the semantic expression generating means 2 and the user goal estimating means 3. Is a system goal generating means for generating a goal of the system necessary to support the user's achievement of the goal based on the belief base updated by the belief base updating means 5.

Reference numeral 7 denotes a problem domain knowledge base for storing knowledge about a problem domain to be solved by dialogue, and reference numeral 8 denotes a system goal generation unit 6 which uses the knowledge of the problem domain knowledge base 7 and the knowledge of the belief base 4 to generate. A plan goal structure generating means for generating a plan goal structure which is a data structure expressing knowledge assumed when executing a plan and a pop which are a sequence of actions performed to achieve the system goal.

9 generates an utterance goal based on the plan goal structure generated by the plan goal structure generating means 8 and the knowledge of the belief base 4, that is, a goal expressing knowledge that can be acquired by the user or the system by uttering from the next system. Utterance goal generating means. 10 is a belief in the utterance goal generated by the utterance goal generation means 9
This is an utterance plan generation unit that generates an utterance plan, that is, an action sequence necessary to achieve an utterance goal, based on the knowledge of the base 4 . What You Must Do ".

11 is an utterance plan executing means for executing the utterance plan generated by the utterance plan generating means 10 to generate a natural language sentence to be transmitted to the user;
Reference numeral 12 denotes a natural language sentence display means for displaying a natural language sentence generated by the utterance plan executing means 11 on a display, and reference numeral 13 denotes a Japanese sentence to be next input by the user based on the utterance plan generated by the utterance plan generation means 10. Is the next utterance prediction means for predicting the next utterance.

The operation will be described below with reference to FIGS. FIG. 2 shows an example of interaction between a user and the system in a travel guide system according to one embodiment of the present invention. 21,
23, 25, and 27 indicate input from the user;
4, 26 and 28 indicate responses from the system.

FIG. 3 shows an example 3 of a generation result of the meaning expression generating means 2.
1 represents the meaning of the first user input “travelers check dropped” 21 in the dialogue example.

FIG. 4 shows an example 41 of the estimation result of the user goal estimating means 3, which expresses that "the user has a goal to have a traveler's check", and the state of the user "the user drops the traveler's check". And the goal of the user, "the user has the goal of having a traveler's check".

FIG. 5 shows a generation result example 51 of the system goal generation means 6, which expresses a goal that the system is about to solve, that is, "a user having a traveler's check is a system goal".

FIG. 6 shows an example of a plan goal structure 601 generated by the plan goal structure generating means 8. The plan is a sequence of actions to be performed to achieve the system goal, and the knowledge premised in executing the plan is shown. Express. Specifically, the data 601 is represented by terms of nine arguments, and the meaning of each is as follows. First argument 602
Indicates the name of the selected plan in the system,
Expresses that the system has selected the plan "The user will reissue traveler checks at the bank".

The second argument 603 indicates a desire condition of the plan, that is, a condition necessary to prompt the user to execute the plan indicated by the first argument 602. The user knows that it is useful to reissue the check. "

The third argument 604 indicates a condition for applying the plan, that is, a condition that must be satisfied in advance when the plan indicated by the first argument 602 is adopted. In order to adopt a plan to do so, the condition that the user has a copy of the traveler's check must be satisfied. "

The fourth argument 605 indicates a precondition of the plan, that is, a condition that must be satisfied before the plan indicated by the first argument 602 is executed. “The user performs a procedure for reissuing a traveler's check. In the past, the user must be at the bank, and in order for the user to be at the bank, the user has to adopt a plan to go to the bank. "

The fifth argument 606 is the schema condition of the plan, that is, the plan indicated by the first argument 602 is the ninth argument 6
The condition necessary for having the goal indicated by 10 as an effect is shown. “In order for the user to have a traveler's check as a result of performing the traveler's check reissue procedure, You must have a lost certificate for the check, and to have a lost certificate for the travelers check, you should adopt a plan to apply for the lost certificate at the police station. "

The sixth argument 607 indicates an action sequence executed when the plan is decomposed, that is, when the plan indicated by the first argument 602 is realized. "The user will go to the counter", "Apply for reissuing the traveler's check", and "Receive the traveler's check".

The seventh argument 608 indicates a deletion proposition, that is, a proposition expressing a state that is not satisfied after execution of the plan indicated by the first argument 602. In this example, "[], that is, there is no particular To express that.

The eighth argument 609 indicates an additional proposition, that is, a proposition expressing a state that is satisfied after the execution of the plan indicated by the first argument 602. “If the user performs an application procedure for a traveler's check, it has a traveler's check. Is obtained. "

The ninth argument 610 indicates the goal of the plan, that is, the goal to be achieved by adopting the plan shown in the first argument 602. Is to achieve the goal of having a traveler's check. "

FIG. 7 shows an example 71 of knowledge about a problem area stored in the problem area knowledge base. Specifically, it is expressed in terms of eight arguments , and the meaning of each is as follows. The first argument 72 indicates the name of the act, and expresses that the knowledge 71 is knowledge relating to the act of “the user goes to the bank”. No.
The two arguments 73 indicate the desire condition of the action, and the user
If you want to go to the line,
Users have to go to the bank to have
That the user needs to know that
Express. The third argument 74 indicates an application condition of the action, and expresses "[], that is, there is no particular condition that must be satisfied before the user performs the action of going to the bank."

The fourth argument 75 indicates a precondition of the action, and expresses that "in order for the user to go to the bank, the condition that the user knows the location of the bank must be satisfied". .

The fifth argument 76 indicates the schema condition of the act. [[], Ie, the condition that the user needs to go to the bank, as a result of executing the act of going to the bank, the condition that the user is in the bank appears. There is nothing in particular.
The sixth argument 77 indicates the disassembly of the act, and expresses "[], ie, the act of the user going to the bank is an act that cannot be further divided in this area plan".

The seventh argument 78 indicates a deletion proposition of the action, and expresses that "the user goes to the bank and, as a result, the user is not in the place where he / she was before going to the bank". 8th draw
Equation 79 shows the additional proposition of the action, and expresses that “the user is at the bank as a result of the user going to the bank”.

FIG. 8 is a flow chart showing the processing contents of the plan goal structure generating means 8. When the processing of the plan goal structure generating means 8 is started, in step 801, the system goal generated by the system goal generating means 6 is received as an input.

In step 802, step 80
1 is searched from the problem domain knowledge base for an action having the system goal input as an additional proposition. In step 803, it is determined whether or not the search in step 802 is successful.
If the process fails in step 804, the process ends. In step 804, the knowledge about the action obtained in step 802 is converted into a plan goal structure.

More specifically, arguments indicating application conditions, preconditions, schema conditions, decomposition, deletion propositions, and addition propositions in the plan goal structure correspond to the corresponding ones in the knowledge on the actions obtained from the problem domain knowledge base. The condition is copied, and [] is given for the argument indicating the desire condition, and step 8 is given for the argument indicating the goal.
Substitute the system goal obtained by 01.

In step 805, step 80
It is determined whether or not the application condition of the plan goal structure obtained in step 4 is satisfied. If so, the process proceeds to step 806. If not, the process proceeds to step 802.
Branch to In step 806, step 80
It is determined whether the prerequisites of the plan goal structure obtained in step 4 are satisfied, and if it is satisfied, the process proceeds to step 810. If not, step 807 is performed.
Branch to

In step 807, step 80
The process of the plan goal structure generation means is recursively called, with the precondition determined in step 6 as to whether the condition has been satisfied is set as the system goal. In step 808, step 80
It is determined whether the recursive call of No. 7 is successful. If the recursive call is successful, the process branches to step 809; otherwise, the process branches to step 802. In step 809, the plan goal structure generated in step 808 is substituted into the precondition of the plan goal structure obtained in step 804.

In step 810, step 80
Schema conditions resulting plan goal structure by 4 judges whether satisfied, if satisfied, stearyl
If not established in step 814, the flow branches to step 811 . In step 812, step 811
Step of determining whether the recursive call is successful, in step 813, if successful, if not successful
Branch to 802 .

In step 813, step 81
The plan goal structure obtained in step 1 is substituted into the argument indicating the schema condition of the plan goal structure obtained in step 804 . In step 814, a desire condition having a plan goal structure is added. More specifically, a proposition indicating that the system plan indicated by the plan name of the plan goal structure is useful for the goal is substituted into the argument indicating the desired condition of the plan goal structure.

Next, in the case where the belief base 4 includes only the proposition “having (user, purchase of traveler's check)” as a proposition to express the situation, that is, “the user has a purchase of traveler's check”, An example of the processing flow of the system generating means when the system goal indicated by 51 in FIG. 5 is input will be described with reference to FIGS.

In step 801, the system goal 51 is input as a system goal. Next, in the processing of step 802, the plan 91 shown in FIG. 9 is searched from the problem area knowledge base, and the search is successful, so the flow branches to step 804.

At step 804, the plan goal structure 101 shown in FIG. 10 is generated, and at step 805, the application condition of the plan goal structure 101 is "having (user, user,
Since the purchase of travelers checks) 102 is established in the belief base 4, the process branches to step 806.

In step 806, since the prerequisite "is (user, bank)" 103 of the plan goal structure 101 is not satisfied in the belief base 4, step 807 is executed.
Branch to In step 807, the plan goal structure 111 shown in FIG. 11 is generated by calling the plan goal structure generation means 8 with “is (user, bank)” as the system goal, and the recursive call is successful. Branch.

At step 809, the plan goal structure 121 shown in FIG. 12 is generated. Next, step 81
At 0, since the schema condition “having (user, lost certificate)” 104 of the plan goal structure 101 is not satisfied in the belief base 4, the process branches to step 811.

In step 811, the plan goal structure 131 shown in FIG. 13 is generated. By the processing in step 812, the process branches to step 813, and data 141 indicating the plan goal structure shown in FIG. 14 is generated. Finally,
In step 814, a plan goal structure 601 is generated.

FIG. 15 shows an example utterance goal 151 generated by the utterance goal generation means 9 from the example plan goal structure 181 and expresses a proposition expressing the knowledge of the system or user trying to achieve the utterance of the next system. . Specifically, "the user knows the location of the bank near the user's location" indicates that the utterance of the next system is the goal to be achieved.

FIG. 16 is a flowchart showing the operation of the utterance goal generating means 9. When the processing of speech goal generating means 9 is activated, at step 161, Purango
Receive Lumpur structure generation means 8 thus generated flop cormorants Ngoru structure as input. In step 162, propositions relating to the knowledge of the user or the system that cannot be confirmed in the belief base 4 are extracted from the plan goal structure, and a set of the extracted propositions is output.

In step 163, the evaluation value of each of the extracted speech goal candidates is calculated using the evaluation means shown in FIG. In step 164, step 1
The speech goal candidate having the highest evaluation value among the speech goal candidates calculated in 63 is extracted, and the speech goal is set as the output of the speech goal generating means 9.

FIG. 17 is a flowchart showing the processing contents of step 163. In step 171, a set of speech goal candidates obtained as the output of step 162 is input. In step 172, step 17
It is determined whether or not there is any set of utterance goal candidates obtained in 1 for which the evaluation value has not yet been calculated. Branches to step 173.

In step 173, step 171
From the set of utterance goal candidates obtained in (1), one utterance goal candidate whose evaluation value has not yet been calculated is selected. Step 1
At 74, a value V1 for evaluating the depth of the relationship with the utterance goal corresponding to the previous utterance is determined. Specifically, the plan goal structure generated by the plan goal structure generating means 8 is taken as a graph as shown at 182 in FIG. 18, and obtained in step 173 from a node representing the utterance goal corresponding to the previous utterance. The reciprocal of the number of arcs on the graph required to reach the node representing the utterance goal is calculated, and its value is set to V1.

In step 175, step 173
To calculate the value V2 for evaluating the magnitude of the effect of achieving the utterance goal obtained by the above on the achievement of the upper goal of the user. Specifically, using a function that gives a type of condition and returns a value indicating the importance of the condition, a value is obtained by applying the type of condition in the plan goal structure of the utterance goal candidate to the function, The value is set to V2. The conditions in the plan goal structure are assumed to have higher importance in the order of application conditions, schema conditions, preconditions, and desire conditions, and as a function for associating a condition type with a value, for example,
Assume that 10, 7, 5, and 3 correspond in the order of the conditions listed above.

In step 176, step 173
A value V3 for evaluating the validity of the utterance goal obtained from the user in view of the problem solving ability is obtained. Specifically, using the function shown in FIG. 19 that returns a value indicating the importance of the utterance goal from the user's comprehension and the type of the condition, the condition in the plan goal structure of the utterance goal candidate is applied to the function. Thus, a value is obtained, and the value is set to V3.

In step 177, step 173
A value V4 for evaluating whether or not the utterance goal obtained by the above is established is obtained. For example, as the value of V4, a proposition expressing the utterance goal candidate is included in the belief base 4 as 0, otherwise, 1 and a proposition expressing the utterance goal is negated. In this case, it is 2.

In step 178, step 17
Evaluation value V obtained from 4 to step 177
The product of the values of 1, V2, V3, and V4 is calculated, and that value is used as the evaluation value of the utterance goal obtained in step 173.

FIG. 18 shows the correspondence between the expression 181 using the term of the plan goal structure and the expression 182 using the graph.

FIG. 19 is an example of a function that returns a value indicating the importance of a condition from the amount of understanding of the user and the type of the condition. In the row of FIG. The right of → indicates the value of evaluating the importance of the utterance goal from the viewpoint of the user's problem solving ability. When the user's comprehension is large, the desire condition is emphasized. When the user's comprehension is small, the desire condition is not emphasized, and the application condition is emphasized regardless of the user's comprehension. I have.

Next, when the belief base 4 stores a proposition that the user's comprehension is small, the utterance of FIG.
An example of the processing flow of the utterance goal generating means performed when the system generates the utterance “Where are you?” 26 will be described with reference to FIGS. 20 and 21.

At step 161, the plan goal structure 181 shown in FIG. 18 is input, and at step 162, a set of speech goal candidates shown in FIG.
In step 163, a row of pairs of speech goal candidates and evaluation values shown in FIG. 21 is obtained.

This evaluation value is calculated according to the flowchart shown in FIG.
In the case of 1, the value of V1 is the value of the previous utterance "Do you have a lost certificate for travelers checks" 24, "Yes"
25 on the graph 182 corresponding to “true (has (us
er, lost certificate))] 183 and the node “know-ref (user, X,
The number of arcs that pass before reaching the "location, bank, [])" 184 is 3 because it is 3, and the value of V2 is node 1
Since 83 is a precondition, the value is 5 and the value of V3 is 2 because the user's comprehension is low and the precondition is low.
And the value of V4 is 1 because the proposition indicating that this goal is established is not stored in the belief system.

As a result, in step 178, the evaluation value of the utterance goal candidate 151 becomes 3.33. Calculation of evaluation values of other goal candidates is performed in the same manner. Step 164
Utterance goal candidate "know-r" with the largest evaluation value
ef (user, X, location, bank, []) ”211 is extracted, and the output of the utterance goal generating means 9 is set as the utterance goal 151.

FIG. 20 shows an example of a set of utterance goal candidates obtained as the output of step 162.
"Users know that it is useful to have a traveler's check to apply for the application for reissuing traveler's checks", "Users know that going to a bank is useful for having a traveler's check", "User's The user knows the location of the bank close to his / her location. "The user knows that it is helpful to have a police check for issuing a lost certificate." The user knows what is useful for the user "and" the user knows the location of the police station near the user's location ".

FIG. 22 shows an example of an utterance plan 221 generated by the utterance plan generation means 10, which represents an action sequence necessary to achieve the utterance goal generated by the utterance goal generation means 9. Specifically, the utterance goal shown in “User knows where bank is” 151 is “The system asks the user where the user is” 222,
"User communicates location to system" 223, "System searches for user's nearest bank location" 22
4. “The system communicates the location of the bank to the user” 22
5 is achieved by four actions. Further, the utterance plan executing means 11 executes “the system asks for the location of the user” 222 to generate the utterance 26.

FIG. 23 shows a user's action 231 predicted by the next utterance predicting means 13, and "the user informs the system where the user is."
Express that. This action is based on the actions that remain after removing the executed actions of the system from the utterance plan.
It is obtained by the user selecting the next action to be performed.

In the above example, an example of the dialogue when the user's goal does not conflict with the system goal has been described. However, when the user's goal does not conflict with the system's goal, the difference between the user's goal and the system's goal is determined. Can be realized by adding a process for checking the user's misunderstanding.

[0061]

As described above, according to the present invention, a joint problem between a system and a user is obtained by using knowledge obtained from an utterance made before and knowledge about a problem area to be solved by dialogue. Based on the data structure expressing the solution process, generate a goal that expresses the knowledge that can be acquired by the user or the system by uttering from the next system, so that questions and explanations according to the situation that changes as the dialogue progresses A sentence and a confirmation sentence can be generated, and a more natural and smooth conversation can be realized.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of an embodiment of a dialog processing control device according to the present invention.

FIG. 2 is an explanatory diagram showing an example of a dialog in one embodiment of the present invention.

FIG. 3 is an explanatory diagram illustrating an example of a generation result of a meaning expression generation unit.

FIG. 4 is an explanatory diagram showing an example of an estimation result of a user goal estimation unit.

FIG. 5 is an explanatory diagram illustrating an example of a generation result of a system goal generation unit.

FIG. 6 is an explanatory diagram illustrating an example of a generation result of a plan goal structure generation unit.

FIG. 7 is an explanatory diagram showing an example of knowledge stored in a problem area knowledge base.

FIG. 8 is a flowchart showing the operation of a plan goal structure generating means.

FIG. 9 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generating unit.

FIG. 10 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generation unit.

FIG. 11 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generation unit.

FIG. 12 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generation unit.

FIG. 13 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generation unit.

FIG. 14 is an explanatory diagram showing an example of data generated during the operation of a plan goal structure generation unit.

FIG. 15 is an explanatory diagram illustrating an example of an utterance goal generated by an utterance goal generation unit.

FIG. 16 is a flowchart showing the operation of an utterance goal generation unit.

FIG. 17 is a flowchart illustrating a method of calculating an evaluation value of an utterance goal appearing during the operation of the utterance goal generation unit.

FIG. 18 is an explanatory diagram showing a method of expressing a plan goal structure.

FIG. 19 is an explanatory diagram showing functions used when calculating an evaluation value of an utterance goal candidate.

FIG. 20 is an explanatory diagram showing a set of speech goal candidates.

FIG. 21 is an explanatory diagram showing pairs of utterance goal candidates and their evaluation values.

FIG. 22 is an example of an utterance plan generated by an utterance plan generation unit.

FIG. 23 is an explanatory diagram showing an action of the user predicted by the next utterance prediction means.

FIG. 24 is an overall configuration diagram of a conventional interactive processing control device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Natural language sentence storage means 2 Meaning expression generation means 3 User goal estimation means 4 Belief base 5 Belief base update means 6 System goal generation means 7 Problem domain knowledge base 8 Plan goal structure generation means 9 Utterance goal generation means 10 Utterance plan generation means 11 utterance plan execution means 12 natural language sentence display means 13 next utterance prediction means 241 information providing system 242 natural language interpretation means 243 command generation means 244 topic management means 245 watched object storage means

Claims (1)

(57) [Claims] An interactive processing system for interpreting the contents of an input sentence, solving a problem, and generating a response sentence, comprising: a natural language sentence storage means for storing an input character string;
Semantic representation generator means for generating a semantic representation of natural language text, which is stored in the natural language sentence storage means, this means representation students
User goal estimation means for estimating the state (goal) to the user of the target based on the semantic representation generated by the formation means, belief base for storing the goal of knowledge and system itself about the user of the system, to the semantic representation generator means Yo
The generated semantic expression and the user goal estimating means
Ri estimated belief based updating means for updating the semantic representation and user goal based on user goal was, this belief base
System goal generating means for generating a goal of a system necessary to support achievement of a user goal based on a belief base updated by an updating means, a problem area for storing necessary knowledge in a problem area to be solved by dialogue A knowledge base, a plan that is a sequence of actions performed to achieve the generated system goal using the knowledge of this problem domain knowledge base, and a plan that is a data structure that expresses the knowledge assumed when executing the plan Plan goal structure generating means for generating a goal structure;
Speech goal generating means for generating a goal to represent the knowledge that the user or the system can be acquired by the utterance (speech goals) from the following system based on the plan goals structures generated by formation means, raw this speech goal generating means < br /> speech plan generating means that generates a speech plan is an act columns needed in order to achieve the utterance goal on the basis of the speech goal has been made, issued, which is generated by the speech plan generating means <br / > by executing the talk plan, speech plan execution means for generating a natural language sentence to be communicated to the user, this speech-flops
Natural language sentence display means for displaying a natural language sentence generated by the run execution means, the next utterance prediction means for predicting a Japanese sentence which the user then inputs based on the utterance plan generated by the speech plan execution means An interactive processing control device, comprising: