JPH08161175A - Language generation device and language generation method - Google Patents

Abstract

(57) [Summary] [Object] An object of the present invention is to output a language expression sequence for explaining the partially fixed content when the content of a certain problem solving act is partially fixed. To provide a possible language generation device and language generation method. According to the present invention, when the content of a certain problem solving action is partially determined, a language expression string for explaining the partially determined content is output, and the content of the problem solving action is further detailed. Is a language generation device that outputs a sequence of language expressions for explaining the detailed contents when the above is determined.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The language generation apparatus and language generation method of the present invention have a problem necessary for converting an initial state into a target state when a problem consisting of a set of an initial state and a target state is given as an input. The present invention relates to a language generation device and a language generation method that generate a sequence of solution actions and output a sequence of language expressions for explaining the content of each problem solution as a voice or a character string.

For example, the problem of an initial state in which a person exists in a certain place and a target state in which the person exists in another place is solved by a character recognition device, a keyboard of a computer terminal, a voice recognition device, or , A sequence of problem-solving actions such as moving from one place to another using a transportation system that is necessary to convert the initial state into a target state via these, and each problem-solving action Is generated gradually while refining the contents of step by step. Further, a computer that gradually outputs a string of language expressions for explaining each problem solving plan as a character string or voice through a display of a computer terminal, a printer, a typewriter, a voice synthesizer, or a combination thereof. The present invention relates to a language generation device and a language generation method in processing technology.

[0003]

2. Description of the Related Art Conventional techniques for realizing a device for outputting a linguistic expression by a character string or voice can be divided into two approaches. The first approach is
This is an approach in which the content to be described in natural language is input, a sequence of linguistic expressions for explaining the content is generated, and the sequence of linguistic expressions is output.

The second approach is to input a problem consisting of a pair of an initial state and a target state, generate a problem solving action for converting the initial state into the target state, and explain the contents of the problem solving action in order. It is an approach to generate a string of language expressions for doing and output the string of language expressions as a character string or a voice, and after determining the content of each problem solving action,
A string of linguistic expressions for transmitting the content is output as a character string or a voice.

[0005]

However, the above-mentioned first conventional approach is based on the assumption that the content to be described is given as an input by a sequence of language expressions.
Since there is no means for generating a problem-solving action for solving the input problem, a language expression for inputting the problem, generating a sequence of problem-solving actions, and explaining the content of the problem-solving action There is a problem that cannot output.

Further, in the above-mentioned second conventional approach, after the content of each problem-solving action is determined, a sequence of language expressions for transmitting the content is output by a character string or a voice, so that the problem-solving action is performed. Even if the content is partially determined, there is a problem that a sequence of language expressions for explaining the partially determined content cannot be output.

Further, in the conventional technique, even after the output of the language expression sequence for explaining the content of one problem-solving action is completed, the content of the problem-solving action to be described next is determined. If the time is longer than a certain specified time, the time interval in which no language expression is output cannot be kept within the specified time. Therefore, there is a problem that it is difficult to judge whether the output of the linguistic expression continues or has ended.

The present invention has been made in view of the above points, and although the contents of a certain problem-solving action are partially fixed as in the prior art, the partially fixed contents are fixed. Even if it is not possible to output the language expression for explaining the, when the content of a certain problem-solving act is partially confirmed, a language expression string for explaining the partially confirmed content is output, An object of the present invention is to provide a language generation device and a language generation method capable of outputting a string of language expressions for explaining the detailed content when the content of the problem solving action is determined in more detail. .

[0009] A further object of the present invention is to keep the time interval in which no language expression is output within a specific prescribed time when the time until the content of the problem solving act is determined is longer than the specific prescribed time. A language generation device and a language generation method capable of outputting a string of language expressions while keeping a time period in which no language expressions are output at a prescribed time even when it is not possible.

[0010]

FIG. 1 is a block diagram showing the principle of the present invention. The present invention, when given as an input a problem expressed as a set of an initial state and a goal state, generates a sequence of problem solving actions necessary for converting the initial state into the goal state, and In a language generation device that outputs a sequence of language expressions for explaining contents as a character string or a voice, a problem of gradually generating a sequence of problem solving actions while gradually detailing each sequence of problem solving actions Utterance planning means for gradually generating the solving means 10 and the sequence of utterance actions necessary for explaining the sequence of each problem solving action transferred from the problem solving means 10 in natural language while gradually generating the details. 20, a linguistic expression output means 30 for sequentially outputting a linguistic expression corresponding to the sequence of utterance actions transferred from the utterance planning means 20 by a character string or voice, the problem solving means 10, the utterance planning means 20 and Language representation output means 30
And a control means 90 for operating in parallel.

Further, when the problem solving means 10 being executed by parallel operation more partially determines a certain problem solving action sequence, the above-mentioned control means 90 executes the partially determined problem solving action. The sequence is controlled to be transferred to the utterance planning unit 20, and the utterance action sequence for explaining the sequence of the problem solving action partially determined by the utterance planning unit 20 is generated and transmitted to the linguistic expression output unit 30. Control.

Further, the above-mentioned control means 90, when the sequence of a certain problem-solving action is decided in more detail by the problem-solving means 10 being executed by parallel operation, the sequence of the problem-solving action decided in more detail. Is controlled to be transferred to the utterance planning means 20, the utterance planning means 20 interrupts the transfer of the utterance act sequence to the linguistic expression output means 30 at that time, and the more detailed confirmed problem action sequence will be explained. A sequence of utterance actions for performing is generated and controlled to be transferred to the language expression output means 30.

Further, the language generating apparatus of the present invention comprises a time measuring means 50 for measuring an elapsed time after the completion of outputting the language expression of the language expression output means, and an elapsed time measured by the time measuring means 50 for a predetermined time. When it exceeds the above, it further has a stagnation expression generating means 40 which generates a stagnation expression and transfers it to the language expression output means 30. FIG. 2 is a diagram for explaining the principle of the present invention.

In the language generation method of the present invention, when a problem consisting of a set of an initial state and a target state is given as an input (step 1), a sequence of problem solving actions necessary for converting the initial state into the target state. Is gradually generated while stepwise refining (step 2), and when the sequence of problem-solving actions is partially determined, it is necessary to explain the sequence of partially-determined problem-solving actions. Sequentially generating a sequence of utterance actions while gradually detailing the contents of each utterance action (step 4), and sequentially providing a sequence of language expressions corresponding to the sequence of utterance actions,
Output as a character string or voice (step 7).

Further, in the language generating method of the present invention, the contents of each problem solving action are detailed step by step (step 2), and when a certain problem solving action string is determined in more detail (step 3, Yes). , The generation of the utterance action sequence is interrupted (step 5), and the utterance action sequence for explaining the more detailed confirmed problem solving action sequence is regenerated (step 6) to correspond to the utterance action sequence. The linguistic expression is output (step 7).

Further, the language generation method of the present invention measures the time after outputting the language expression sequence, and generates the stagnation expression when the language expression sequence is not output even after a predetermined time has elapsed. Output.

[0017]

In the first and fifth inventions, a sequence of problem-solving actions is generated stepwise by parallel operation, and a sequence of utterance actions is also generated stepwise, in the form of a language expression. It becomes possible to output. Furthermore, the second and sixth inventions further solve the problem when a sequence of problem solving actions is gradually generated step by step by operating the respective devices constituting the language generation device in parallel. When the sequence of actions is partially determined, the sequence of partially determined problem solving actions is transferred to the means for generating the sequence of speech actions. With this, even if the content of the problem-solving action is not fixed, it is possible to output a linguistic expression for explaining the content of the problem-solving action when the content is partially fixed.

Further, in the third invention, after the content of the problem-solving action is partially determined, and when the content of the problem-solving action is determined in more detail, the more detailed content will be explained at that time. It is possible to output the language expression of. Therefore, it is possible to start outputting a linguistic expression for explaining the content of the problem-solving action at a time earlier than when the content of each problem-solving action is determined.

Further, the fourth and seventh inventions measure the elapsed time after the output of a certain language expression is finished, and even if the elapsed time exceeds a predetermined time, any language expression can be displayed. If it is not output, the stagnation act is generated and output, so even if no linguistic expression is output, if the linguistic expression output remains, it represents the stagnation act. Since the linguistic representation is output, it can be recognized that the output of the linguistic representation sequence is still continued.

[0020]

EXAMPLES Examples of the present invention will be described in detail below. FIG. 3 shows the configuration of the language generation device of the present invention. The configuration shown in the figure has a gradual problem solving device 10, a gradual utterance planning device 20, a language expression output device 30, and a stagnation action generation device 4.
0 and the time management device 50. The respective devices of the language generation device 100 shown in the figure operate in parallel.

When a problem consisting of a set of an initial state and a goal state is input, the gradual problem solving apparatus 10 displays a sequence of problem solving actions necessary for converting the initial state into the goal state.
Generate the contents of each problem-solving act gradually, while gradually refining them. When the content of each problem solving action is partially determined, the gradual problem solving apparatus 10 gradually develops a sequence of problem solving actions whose content is partially determined to be the utterance utterance planning apparatus 2.
Transfer to 0.

The gradual utterance planning device 20 stores a sequence of utterance actions necessary for explaining the partially confirmed contents of the problem solving act transferred from the gradual problem solving device 10 and the contents of each utterance act. While gradually refining, the sequence of utterance actions is generated and transferred to the linguistic expression output device 30. The language expression output device 30 uses the gradual speech planning device 20.
The sequence of linguistic expressions corresponding to the sequence of speech act transferred from is output as a character string or a voice in order. Also, during this time,
The gradual problem solving device 10 refines the content of a problem solving act step by step, and when the content of a certain problem solving act is decided in more detail, the sequence of problem solving acts whose contents are decided in more detail is gradually made. It is transferred to the speech planning apparatus 20. The gradual utterance planning device 20 interrupts the transfer of the utterance act sequence to the linguistic expression output device 30 at the time when the more detailed problem-solving act sequence is transferred, and the more detailed content is presented. A sequence of utterance actions for explaining the determined problem solving action is regenerated and transferred to the language expression output device 30.

The time management device 50 is a language expression output device 3
When the elapsed time after the output of a certain linguistic expression is 0 is measured and the elapsed time exceeds a specific specified time, the linguistic expression output device 30 does not output any linguistic expression, The time management device 50 transmits a signal to the stagnation behavior generation device 40. Stagnation act generation device 4
When the time management device 50 receives the signal, 0 transfers the stagnation action to the language expression output device 30. Thereby, the linguistic expression output device 30 outputs the stagnation expression as a character string or a voice according to the stagnation act.

The detailed operation will be described below. In the present embodiment, the problem input to the language generation device 100 is expressed as a set of an initial state and a target state in the form shown in the following example. [Example 1] Initial state, target state The initial state and the target state that form the problem are expressed using a first-order predicate logical expression. For example, the problem that consists of the initial state that "a person a exists in a place called NTT Musashino R & D Center" and the goal state that "a person a exists in a place called NTT Atsugi R & D Center" is as follows. [Example 2] of the above.

[Example 2] (at (a, musashino-center), at (a, atsugi-center)) In [Example 2], "a" is a constant representing the person a, and "musashino-center" Is a constant that represents a place called the NTT Musashino R & D Center.
“A” is a constant that represents a place called NTT Atsugi R & D Center, and “at” is a predicate that a certain person exists at a certain place.

FIG. 4 is a diagram for explaining a gradual problem solving device according to an embodiment of the present invention, in which the gradual problem solving device 10 has a traffic route knowledge 61 and a problem solving action creation order knowledge 6.
2 shows a state in which a sequence of problem solving actions is output using 2. FIG. 5 is a diagram for explaining the operation of the embodiment of the present invention. In this embodiment, the gradual problem solving device 10
Is based on the knowledge about the traffic route from “NTT Musashino R & D Center” to “NTT Atsugi Research and Development Center” shown in FIG. [2] The sequence of problem-solving actions for converting the initial state into the target state is gradually generated while gradually detailing the content of each problem-solving action, and finally, "NTT Musashino R & D Center". To go to Kichijoji by bus… ”, a problem-solving act of“ going from Kichijoji to the Inokashira line to Shimokitazawa ”… and a problem-solving action of“ going from Shimokitazawa to Aiko Ishida on the Odakyu line ”… From Aiko Ishida, take a bus to the NTT Atsugi R & D Center. ”... A problem-solving act is created by arranging four problem-solving actions in order.

FIG. 6 is a diagram for explaining a gradual speech planning apparatus according to an embodiment of the present invention. In the present embodiment, the gradual utterance planning device 20 has the utterance action generation order knowledge 70, which is knowledge about the utterance action generation order for explaining the sequence of problem solving actions transferred from the gradual problem solving device 10.
And a vocabulary knowledge 80 for describing the content of each problem solving action in a linguistic expression are used to gradually generate a sequence of utterance actions.

Here, consider a case where the language expression knowledge output device 30 outputs a language expression for explaining the content of the problem solving action when the content of the problem solving action is partially determined. First, when the gradual problem solving apparatus 10 gradually generates the above-described problem solving action sequence (1) to (1) while gradually detailing each problem solving action sequence, the content of the problem solving action is It shows that the language expression for explaining the contents can be output at the time of partial confirmation.

In the present embodiment, when the problem consisting of the set of the initial state and the target state shown in [Example 2] is given, the gradual problem solving apparatus 10 converts the initial state into the target state. The column of is generated gradually while detailing the contents of each problem solving action step by step. First, the expression shown in the following [Example 3] is generated as a sequence of the problem-solving actions necessary for converting the initial state shown in [Example 2] above into the target state.

[Example 3] (PA1, PA2, PA3) where PA1 = move-to-nearest-station-by-bus (musashino-cen
ter, * x), PA2 = move-through-railways (* x, * y), PA3 = move-from-nearest-station (* y, atsugi-cen
ter) In [Example 3], PA1, PA2, and PA3 represent problem solving actions, and (PA1, PA2, PA3) represent a sequence of problem solving actions. In the above [Example 3], symbols starting with "*" represent variables.

The variable * x represents "the nearest station of the NTT Musashino R & D Center", and the variable * y represents "the nearest station of the NTT Atsugi R & D Center".

The problem-solving action PA1 is shown to have the content of "go from the NTT Musashino R & D Center to the nearest station * x of the NTT Musashino R & D Center by bus."
2 indicates that it has the content of "go by train from the nearest station * x to the NTT Musashino R & D Center to the nearest station * y to the NTT Atsugi R & D Center," and the problem-solving action PA
No. 3 has the content of "go to the NTT Atsugi R & D Center by bus from the nearest station * y to the NTT Atsugi R & D Center."

Next, the gradual problem solving apparatus 10 further includes
The problem-solving action column is refined to generate a problem-solving action column shown in [Example 4] below. [Example 4] (PA11, PA21, PA3) However, PA11 = move-to-nearest-station-by-bus (musashino-c
enter, kichijouji), PA21 = move-through-railways (kichijouji, * y), PA3 = move-from-nearest-station (* y, atsugi-cen
ter) In [Example 4], PA11, PA21, and PA3 are problem solving actions. “Kichijouji” is a constant representing “Kichijoji Station”.・ Problem-solving action PA11 is a detailed version of problem-solving action PA1 in [Example 3], and has the content of "going from NTT Musashino R & D Center to Kichijoji by bus."・ Problem-solving action PA21 is a detailed version of problem-solving action PA2 in [Example 3]. It has the content of "go by train from Kichijoji Station to the nearest station * y at the NTT Atsugi R & D Center." .

Next, the gradual problem solving device 10 further refines the content of the problem solving action and generates a sequence of problem solving actions as the expression shown in [Example 5] below. [Example 5] (PA11, PA211, PA212, PA31) However, PA11 = move-to-nearest-station-by-bus (musashino-
center, kichijouji), PA211 = move-through-line (kichijouji, shimokitaza
wa, inokashira-line), PA212 = move-through-line (shimokitazawa, aikouish
ida, odakyu-line), PA31 = move-from-nearest-station (aikouishida, at
sugi-center) In [Example 5], PA11, PA211, PA212, and PA31 are problem solving actions. “Shimokitazawa” is “Shimokitazawa Station”
"Aikouishida" is a constant that represents "Aiko Ishida Station", "inokashira-line" is a constant that represents "Inokashira Line", and "odakyu-line" is "Odakyu Line" Is a constant that represents.

The problem solving actions PA211 and PA212 are described in [Example 4].
It is a detailed version of the content of Problem Solving Act PA21.・ PA211 has the content of "go from Kichijoji to the Inokashira line to Shimokitazawa."・ PA212 has the content of "go from Shimokitazawa to Aiko Ishida Station on the Odakyu Line".・ Problem solving act PA31 is the problem solving act in [Example 4].
It is a detailed version of PA3 and has the content of "going to NTT Atsugi R & D Center from Aiko Ishida Station by bus."

The gradual problem solving device 10 sequentially refines the contents of the problem solving action sequence in order of [Example 3],
The problem solving action sequences shown in [Example 4] and [Example 5] are transferred to the gradual utterance planning device 20. Gradual Speech Planning Device 20
The following is a sequence of utterance actions for explaining the contents of the problem solving action column when the sequence of partially solved problem solving actions shown in [Example 3] is transferred. Generate an expression.

[Example 6] (SA1, SA2, SA3) where SA1 = inform-action (PA1) SA2 = inform-action-successively (PA2, SA1) SA3 = inform-action-successively (PA3, SA2) PA1 = move-to-nearest-station-by-bus (musashino-ce
nter, * x), PA2 = move-through-railways (* x, * y), PA3 = move-from-nearest-station (* y, atsugi-cen
ter) In [Example 6], PA1, PA2, and PA3 are problem solving actions as shown in [Example 3]. SA1, SA2, and SA3 represent utterance actions, and (SA1, SA2, SA3) represent a sequence of utterance actions. Utterance act SA1 is an utterance act of transmitting the content of problem solving act PA1, and utterance act SA2 is an utterance act SA.
2 is the utterance act of transmitting the content of the problem-solving act PA2, and the utterance act SA3 is the utterance act SA2.
This is an utterance act of transmitting the content of the problem-solving act PA3 following the execution of.

Next, the gradual utterance planning device 20 refines the utterance action shown in [Example 6] and generates a sequence of utterance actions represented by the expression shown in [Example 7] below. [Example 7] (UA1, UA2, UA3, UA4, SA2, SA3) However, UA1 = utter (“From NTT Musashino R & D Center”) UA2 = utter (“By bus”) UA3 = utter (“To the nearest station )) UA4 = utter (“go”) SA2 = inform-action-successively (PA2, SA1) SA3 = inform-action-successively (PA3, SA2) PA1 = move-to-nearest-station-by-bus (musashino -ce
nter, * x), PA2 = move-through-railways (* x, * y), PA3 = move-from-nearest-station (* y, atsugi-cente
r) In [Example 7], UA1, UA2, UA3, and UA4 are utterance actions in which the utterance action SA1 shown in [Example 6] is detailed. The utterance act expressed in the form of utter (expr) is ex
Represents the utterance act of uttering the language expression pr.

The gradual utterance planning device 20 uses the utterance act SA1.
However, by using the vocabulary knowledge 80 for describing the content in a linguistic expression by having the content of "go from the Musashino R & D Center to the bus station", the utterance act SA1 is made into the utterance act UA1, UA2. , UA3, UA4. At this point, the speech acts UA1, UA2, UA3, UA4 are sequentially sent to the language expression output device 30.

Each time the utterance act is sent from the utterance utterance planning device 20, the linguistic expression output device 30 corresponds to each utterance act from "from Musashino R & D Center" to "by bus" to "nearest station". The language expressions "to" and "go" are output in sequence as a character string or voice. In this output process, the gradual problem solving device 10 generates a sequence of problem solving actions having more detailed contents as shown in [Example 4] and transfers it to the gradual utterance planning device 20. The gradual utterance planning device 20 suspends sending the utterance action to the linguistic expression output device 30 at the time when the sequence of problem solving actions in which the more detailed content has been determined, which is shown in [Example 4], is sent. Here, the gradual utterance planning device 20 makes the utterance act UA1.
, UA2, UA3, UA4 in sequence, and the language expression output device 30
It is assumed that the gradual problem solving device 10 has transferred to the gradual utterance planning device 20 a sequence of problem solving actions having more detailed contents shown in [Example 4] at the time of transferring to.

The language expression string already output from the language expression output device 30 at this point is as shown in the following [Example 8]. [Example 8] “By bus from Musashino R & D Center” Here, the gradual utterance planning device 20 is for explaining the content of the problem solving action sequence having more detailed content shown in [Example 4]. The sequence of utterance actions is started to be generated, and the utterance action for explaining the content of each problem solving action shown in [Example 4] is generated, and the expression shown in [Example 9] below is generated.

[Example 9] (SA11, SA21, SA3) where SA11 = inform-action (PA11) SA21 = inform-action-successively (PA21, SA11) SA3 = inform-acition-successively (PA3, SA21) PA11 = move-to-nearest-station-by-bus (musashino-c
enter, kichijouji), PA21 = move-through-railways (kichijouji, * y), PA3 = move-from-nearest-station (* y, atsugi-cen
ter) In [Example 9], PA11, PA21, and PA3 are problem solving actions as shown in [Example 4]. SA11, SA21, SA3 are
Represents a speech act. The utterance act SA11 is an utterance act of transmitting the content of the problem solving act PA11.
Is the utterance act of transmitting the content of the problem solving act PA21 after the execution of the utterance act SA11.
Is the utterance act of transmitting the content of the problem solving act PA3 after the execution of the utterance act SA21.

Next, the gradual utterance planning apparatus 20 further refines the utterance action shown in [Example 9], and the following [Example 1]
[0] is generated as a sequence of utterance actions. [Example 10] (UA5, US6, UA7, UA8, SA21, SA3) However, UA5 = utter ("From NTT Musashino R & D Center") UA6 = utter ("By bus") UA7 = utter ("To Kichijoji") UA8 = utter (“go”) SA21 = inform-action-successively (PA21, SA11) SA3 = inform-action-successively (PA3, SA21) PA11 = move-to-nearest-station-by-bus (musashino-c
enter, kichijouji), PA21 = move-through-railways (kichijouji, * y), PA3 = move-from-nearest-station (* y, atsugi-ce
nter) In [Example 10], UA5, UA6, UA7, and UA8 are utterance actions in which the utterance action SA11 shown in [Example 9] is detailed. The gradual utterance planning apparatus 20 uses the fact that the utterance act SA11 has the content of "going from Kusamura R & D center to Kichijoji by bus" and the vocabulary knowledge 80 for describing the content in linguistic expressions. , Utterance act SA11,
Utterance act UA5, UA6, UA7, UA8 are detailed. Of these, utterance act UA5 is the act of uttering a language expression “from NTT Musashino R & D Center”, but since the equivalent action has already been executed as utterance act UA1 shown in [Example 7]. , Utterance act UA5 is a language expression output device 3
Not transferred to 0. Similarly, the utterance act UA6 is an act of uttering the language expression "by bus", but since the equivalent action has already been executed as the utterance act UA2 shown in [Example 7], the utterance act UA6 is , Language expression output device 3
Not transferred to 0. Therefore, the gradual utterance planning device 20
Is the utterance act UA7 and UA8 in order, and the language expression output device 30
Then, the language expression output device 30 sequentially outputs two language expressions “to Kichijoji” and “go” as corresponding language expressions.

Eventually, the language expression sequence shown in the following [Example 11] will be sequentially generated together with the language expression sequence shown in [Example 8]. [Example 11] “Go from Musashino R & D Center to Kichijoji by bus.” As explained above, the gradual problem-solving device 10 refines the sequence of problem-solving actions step by step to detail the content of each problem-solving action. However, when the content of a certain problem-solving action is partially generated and is partially determined, the language expression output device 30 outputs a language expression sequence for explaining the determined content to that part, and the problem When the content of the solution act is further determined in detail, it becomes possible to output a string of language expressions for explaining the detailed content.

As described above, according to the present invention, when the content of the problem-solving action is partially or partially determined, a language expression string for explaining the content is generated, so that the conventional method is realized. In addition, when the content of the problem-solving action is determined, it is possible to start generating the language expression sequence at a faster time than the method of outputting the sequence of language expressions describing the content of the problem-solving action.

Next, a case will be described in which the stagnation expression is output to keep the time interval in which no linguistic expression is output within the specified time. The gradual utterance planning device 20 generates the utterance action sequence shown in [Example 7], and while transferring the utterance action sequence to the linguistic expression output device 30 in sequence, the utterance shown in [Example 7]. When the actions UA1 and UA2 are transferred to the linguistic expression output device 30, the gradual problem solving device 10 interrupts the transfer of the utterance actions UA3 and UA4 to the linguistic expression output device, and in [Example 10]. Consider the process of generating the utterance action sequence shown.

At this time, in the time management device 50, until the language expression output device 30 finishes outputting the language expression corresponding to the speech act UA2 shown in [Example 7] and starts outputting the next language expression. Measure the elapsed time of. If the output of the next linguistic expression does not start even after the elapsed time exceeds the specified time, the time management device 50 transfers the signal to the stagnation action generation device 40. Upon receiving the signal, the stagnation act generation device 40 transfers the stagnation act represented by the expression shown in [Example 12] below to the linguistic expression output device 30.

[Example 12] utter ("Eto") In [Example 12], utter ("Eto") is a stagnation act of uttering the stagnation expression "Eto". The linguistic expression output device 30 receives this naughty act and outputs the corresponding naughty expression.

At this point, the sequence of language expressions output is
This is as shown in [Example 13] below. [Example 13] “From the Musashino R & D Center on the bus ...” Then, while the linguistic expression output device 30 outputs the stagnation expression, the gradual utterance planning device 20 [example 10].
When the generation of the utterance action sequence shown in (3) is finished and the utterance action sequence is transferred to the linguistic expression output device 30, the linguistic expression sequence corresponding to the utterance action sequence is output. If a new utterance action sequence is not generated even after the stagnation expression has been output and the specified time has passed, the language expression output device 30 uses the same method to say "a" or "er". Output linguistic expressions for stagnation.

After all, after the language expression output device 30 outputs a certain language expression, the time management device 50 measures the elapsed time for outputting the next language expression, and the elapsed time exceeds the specified time. If so, a signal is sent to the stagnation act generation device 40, and the stagnation act generation device 40 receives the signal and transfers the stagnation act to the linguistic expression output device 30, whereby no linguistic expression is output. Can be maintained within the specified time.

The present invention is not limited to the above embodiments, and various modifications and applications are possible within the scope of the claims.

[0052]

As described above, according to the language generating apparatus and the language generating method of the present invention, when a problem consisting of the initial state and the target state is given, it is necessary to convert the initial state into the target state. When a sequence of problem-solving actions is generated and a sequence of language expressions for explaining the content of each problem-solving action is output as a character string or a voice, the problem must be confirmed before the content of each problem-solving action is confirmed. Although the content of the problem-solving action is partially confirmed because the sequence of language expressions for explaining the content of the solving action cannot be output, in the conventional method, the partially-determined content is However, according to the present invention, when the content of a certain problem-solving action is partially fixed, the present invention does not output a language expression string for explaining the partially fixed content as a character string or voice To output the contents of the problem solving acts, when finalized in more detail, it is possible to output a language representation for explaining the detailed contents.

In the conventional method, after the sequence of language expressions for explaining the content of one problem-solving action has been output, until the content of the problem-solving action to be explained next is decided. If the time taken to determine the content of the problem-solving action is longer than a certain specified time because no language expression can be output, it is not possible to keep the time interval in which no language expression is output within the specified time, According to the present invention, a string of language expressions is output as a character string or voice while keeping the time interval from the end of outputting a certain language expression to the output of the next string of language expressions within a specified time. It becomes possible.

Further, according to the conventional method, when outputting a linguistic expression for explaining a sequence of problem-solving actions, it may not be possible to keep the length of a time section in which no linguistic expression is output within a prescribed time. Therefore, when the language generation device does not output any language expression at a certain point, it has finished outputting the language expression sequence for explaining the problem solving action sequence, or there is still a language expression to be output. It is difficult to determine which of the
According to the present invention, since the length of the time interval in which no linguistic expression is generated can be kept within the specified time, whether the language generation device has finished outputting the necessary linguistic expression sequence,
It is easy to determine whether or not there are still strings of language expressions to be output.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is a diagram for explaining the principle of the present invention.

FIG. 3 is a configuration diagram of a language generation device according to an embodiment of the present invention.

FIG. 4 is a diagram illustrating a gradual problem solving device according to an embodiment of the present invention.

FIG. 5 is a diagram for explaining the operation of the embodiment of the present invention.

FIG. 6 is a diagram illustrating a gradual speech planning apparatus according to an exemplary embodiment of the present invention.

[Explanation of symbols]

 10 Problem Solving Means, Gradual Problem Solving Device 20 Utterance Planning Means, Gradual Utterance Planning Device 30 Language Representation Output Means, Language Representation Output Devices 40 Naughty Expression Generation Means, Nesting Action Generation Devices 50 Time Measuring Means, Time Management Devices 60 problem solving action creation order knowledge 70 utterance action creation order knowledge 80 vocabulary knowledge 90 control means

Claims (7)

[Claims] 1. When a problem expressed as a set of an initial state and a target state is given as an input, a sequence of problem solving actions necessary for converting the initial state into the target state is generated, and each problem is generated. In a language generation device that outputs a string of language expressions for explaining the content of a solution act as a character string or as a voice, while gradually detailing the string of each problem solution act, the string of problem solution actions is gradually And a sequence of utterance actions necessary for explaining the sequence of each problem-solving action transferred from the problem-solving unit in natural language in a stepwise manner while gradually detailing Utterance planning means, a linguistic expression output means for sequentially outputting a linguistic expression corresponding to the utterance action sequence transferred from the utterance planning means by a character string or voice, the problem solving means, the utterance Language generation apparatus characterized by a control means for operating the image means and said language representation output means in parallel. 2. The control means, when a row of a certain problem solving action is partially decided by the problem solving means in parallel operation, a detailed decided row of the problem solving action is given to the utterance planning means. A control is performed so that the utterance planning means generates a sequence of utterance actions for explaining the partially determined sequence of problem solving actions, and controls the utterance planning means to transfer the utterance action sequence to the linguistic expression output means. Item 1. The language generation device according to item 1. 3. The utterance planning means sets the more detailed set of problem solving actions when the certain problem solving set is set in more detail by the problem solving unit in parallel operation. To the utterance planning means, interrupting the transfer of the utterance act sequence at that point to the linguistic expression output means,
The language generation device according to claim 1 or 2, wherein a sequence of utterance actions for explaining the sequence of the problem solving actions determined in more detail is generated and controlled to be transferred to the language expression output means. 4. A time measuring unit for measuring an elapsed time after the output of the language expression by the language expression outputting unit, and when the elapsed time measured by the time measuring unit exceeds a predetermined time, The language generation apparatus according to claim 1, further comprising a stagnation expression generation unit that generates a stagnation expression and transfers it to the language expression output unit. 5. When a problem consisting of a pair of an initial state and a goal state is given as an input, the sequence of problem solving actions necessary for converting the initial state into the goal state is refined step by step, When the sequence of problem-solving actions is gradually generated and the sequence of the problem-solving actions is partially determined, the sequence of utterance actions necessary for explaining the sequence of partially-determined problem-solving actions is described. A language generation method characterized in that it is generated gradually while being refined step by step, and a sequence of language expressions corresponding to the sequence of utterance actions is sequentially output as a character string or a voice. 6. The operation of claim 5, wherein the content of each problem-solving action is refined in stages, and when the sequence of a certain problem-solving action is determined in more detail, the generation of the sequence of utterance actions is interrupted. 6. A sequence of utterance actions for explaining the sequence of more detailed determined problem solving actions is regenerated, and a linguistic expression corresponding to the sequence of utterance actions is output.
The described language generation method. The language generation method according to claim 3. 7. The stagnation expression is generated and output when the time after output of the linguistic expression string is measured and the linguistic expression string is not output even after a lapse of a predetermined time. Language generation method.