JP2007079397A - Interaction method, interaction device, interaction program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an interaction method and an interaction device that hold continuity of an answer format at the time of scenario transition from a focus interaction scenario to an auxiliary interaction scenario. <P>SOLUTION: In the interaction device which can communicate with a plurality of interaction systems storing interaction scenarios and answer generation models and stores auxiliary interaction scenarios in a storage means, an input processing means generates indication information indicating an input conversion result and a scenario to be used for interaction processing from an interaction input and outputs them, and a focus interaction system setting means receives the interaction scenario indicated by the indication information and its answer generation model from an interaction system and sets them as a focus interaction model and a focus answer generation model; and an interaction scenario execution means generates answer contents for the input conversion result by using the focus interaction scenario or auxiliary interaction scenario and outputs them, and an answer generating means generates an interaction output from the answer contents by using the focus answer generation model and outputs the interaction output. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an interactive technique for outputting an appropriate response to an input based on an interactive scenario. More specifically, the present invention relates to a dialog technology that realizes a dialog corresponding to a wide range of topics by automatically switching between a plurality of dialog systems that handle different topics.

  A traditional dialogue system that accepts input such as voice or text and outputs an appropriate response can handle dialogue limited to a specific topic, but responds appropriately to input related to a wide range of topics. It was difficult to output. This is because the description of the dialogue scenario for determining the response to the input becomes complicated as the topic range increases. In addition, in a dialogue system that can handle only a narrow range of topics, there are problems such as an inadequate response to the user's request, and the user's ability to understand the capabilities of the system in advance. Therefore, constructing a dialogue system that handles a wide range of topics in a pseudo manner by constructing a plurality of dialogue systems that handle a narrow range of topics that can be constructed relatively easily, and performing dialogue while switching them appropriately. Has been tried.

  In such an attempt, for example, as shown in Non-Patent Document 1, in addition to a plurality of dialogue systems, a neutral dialogue state (a dialogue state that does not correspond to any of the dialogue states in the plurality of dialogue systems) is prepared. There is a multi dialogue system. When such a multi-dialog system is used, an auxiliary dialog scenario corresponding to a neutral dialog state is prepared, and when the currently operating dialog system cannot process input, a dialog for switching to another dialog system is prepared. It is conceivable to use a supplementary dialogue scenario. In addition, the above-mentioned “when input cannot be processed” means, for example, when the user performs input outside the design assumption range of the currently operating interactive system, or when the input processing procedure involves pattern recognition processing. If the pattern recognition fails, the subsequent processing cannot be executed.

In the future, a dialogue system that is currently operating (or operated) among a plurality of dialogue systems will be referred to as a “focal dialogue system”, and a dialogue scenario that determines the response content of the focal dialogue system will be referred to as a “focal dialogue scenario”. write.
Toshihiro Isobe, 5 people, “Examination of domain switching scale in spoken dialogue system using multiple models selectively”, 47th SIG-SLP Research Group, Information Processing Society of Japan, 2003 July 19, p. 41-46

  In the multi dialogue system described above, one of a plurality of dialogue systems is operated as a focal dialogue system, and the dialogue is performed while transitioning between the focal dialogue scenario and the auxiliary dialogue scenario corresponding to the focal dialogue system. The dialogue is realized by switching the focal dialogue system according to the progress. In such a multi-dialog system, there is a technical difficulty in that a dialogue is naturally performed when a scenario used between a focal dialogue scenario and an auxiliary dialogue scenario transitions.

  In the multi-dialog system, for example, when the user intends to input some request to the focus dialog system, but the input cannot be processed by the focus dialog system, the scenario to be used is changed to the auxiliary dialog scenario. In this case, in the auxiliary dialogue scenario, it may be possible to perform processing to prevent failure of the dialogue such as telling the user that the current focal dialogue system cannot handle input and confirming whether to switch to another dialogue system. .

  Here, let us consider a case where the response system of the dialog system is different in each dialog system. For example, in the case of responding by voice, the “response mode” is the speaker nature such as the volume of the voice, the speaking speed, the voice quality, and the like. The agent character is displayed on the screen, and the gesture and the text or the voice are combined. In the case of responding, it is the type of agent character. In such a case, when the scenario used for the auxiliary dialogue scenario changes from the focal dialogue scenario, the agent character suddenly changes or the voice speaker nature changes. That is, from the user's standpoint, the user feels that the response style suddenly changes despite the intention to input to the focal dialogue system. In general, in the same dialog system, the user proceeds with the dialog assuming the same response mode. Therefore, a dialog system that seems to suddenly change the response mode of the same dialog system is confusing to the user. In other words, the conventional multi-dialog system has a problem in that discontinuity of the response style occurs when the scenario used from the individual focus dialog scenario to the auxiliary dialog scenario transitions.

  Accordingly, in view of the above problems, the present invention provides a dialog method, a dialog device, a dialog program, and a recording medium that maintain continuity of response style when a scenario transition occurs from a focused dialog scenario to an auxiliary dialog scenario. Objective.

  In order to solve the above-mentioned problem, in the present invention, at least a dialogue scenario and a response generation model can be stored to communicate with a plurality of dialogue systems capable of executing dialogue processing, and the storage means includes at least the dialogue system. In an interactive device in which an auxiliary interactive scenario capable of performing interactive processing in the switching of the dialog is stored, the input processing means of the interactive device converts the user's interactive input into a format capable of interactive processing from the user's interactive input The input conversion result and the instruction information for instructing the dialog scenario or the auxiliary dialog scenario for executing the dialog processing are generated and output, and the dialog system instructed by the focus dialog system setting means of the dialog device specifies the dialog Receive a dialogue scenario and response generation model of a dialogue system having a scenario from the dialogue system , Each of which is set as a focal dialogue model and a focal response generation model for executing dialogue processing, and the dialogue scenario execution means of the dialogue device uses the focal dialogue scenario or the auxiliary dialogue scenario to respond to the input conversion result. The content is generated and output, and the response generation means of the dialog device generates and outputs the dialog output presented to the user from the response content using the focus response generation model.

  Further, the computer can be operated as an interactive device by an interactive program that causes the interactive device to function on the computer. A computer-readable program recording medium that records this interactive program makes it possible for another computer to function as an interactive device, or to distribute the interactive program.

  According to the present invention, even when a scenario transition occurs from a focal dialogue scenario to an auxiliary dialogue scenario, a dialogue output is generated and output from the response contents using the focal response generation model in the current dialogue system. The continuity of response style when a scenario transition occurs from a scenario to an auxiliary dialogue scenario is maintained.

In the following, two embodiments will be described.
In the first embodiment, a dialogue system to be used is determined for each input from a user, and dialogue processing is performed according to the decided dialogue system.
The second embodiment is a case where the technique of the present invention is used in the dialogue system disclosed in Reference Document 1.
(Reference Document 1) Japanese Patent Application No. 2005-232215 << First Embodiment >>
Below, 1st Embodiment of this invention is described, referring FIGS.

One of the best embodiments of the present invention is a form in which a computer functions as an interactive device mainly by causing the computer to execute the interactive program of the present invention.
FIG. 1 is a configuration block diagram illustrating a hardware configuration of the interactive apparatus (A) according to the first embodiment.

  As illustrated in FIG. 1, the interactive device (A) includes an input unit (11) to which a microphone, a keyboard, and the like can be connected, an output unit (12) to which a speaker, a liquid crystal display, and the like can be connected, and an external device of the interactive device (A). A communication unit (13) to which a communication device (for example, a modem) that can communicate with the communication unit (13) can be connected, MPU [Micro Processing Unit] (14) [cache memory, etc. may be provided. RAM (Random Access Memory) (15), ROM (Read Only Memory) (16), an external storage device (17) such as a hard disk, etc., and their input unit (11), output unit (12), A communication unit (13), an MPU (14), a RAM (15), a ROM (16), a bus (18) connected so as to exchange data between the external storage devices (17), and the like are provided. If necessary, the interactive device (A) may be provided with a device (drive) that can read and write a storage medium such as a CD-ROM.

  In the external storage device (17) of the dialog device (A), an auxiliary dialog scenario and an auxiliary response generation model corresponding to at least a neutral dialog state are stored and stored.

  The ROM (16) of the interactive device (A) stores and stores a program for enabling interactive processing, data necessary for processing of this program, and the like. Further, data obtained by the processing of these programs is appropriately stored and stored in the RAM (15) or the like.

  More specifically, in the ROM (16), the response contents etc. depending on the program for performing information processing so that the dialogue processing can be executed on the information input to the dialogue device (A), the focal dialogue scenario, etc. , A program for executing setting / changing of the focal dialog system, and a program for generating and outputting response information from the response contents are stored and stored. In addition, a control program for controlling processing based on these programs is also stored as appropriate.

  In the interactive apparatus (A) according to the first embodiment, each program stored in the ROM (16) and data necessary for processing of each program are read into the RAM (15) as necessary, and the MPU (14 ) Is interpreted and executed. As a result, the MPU (14) implements predetermined functions (input processing unit, dialogue scenario execution unit, focal dialogue system setting unit, response generation unit, control unit), thereby realizing dialogue processing.

  Subsequently, the dialogue processing in the first embodiment will be described descriptively with specific examples, with reference to FIGS.

The interactive device (A) and a plurality (n) of interactive systems, that is, the first interactive system (1041), the second interactive system (1042),..., The nth interactive system (104n) 1) and are communicably connected to each other (see FIG. 18). Each interactive system is assumed to be a known interactive system capable of executing an interactive process. By using an existing dialogue system in this way, a multi-dialog system that can deal with various topics at a reduced cost is constructed.
Each dialogue system stores at least a dialogue scenario and a response generation model in its storage means. That is, the first dialog system (1041) stores the first dialog scenario (1041b) and the first response generation model (1041a) in its storage means, and the second dialog system (1042) stores the second dialog scenario (1042b). ) And the second response generation model (1042a) are stored in the storage means, and the nth interactive system (104n) stores the nth interaction scenario (104nb) and the nth response generation model (104na) in the storage means. ing. The dialogue scenario is data in which a processing command for constructing a dialogue, a response to an input, and the like are described for a predetermined topic (for example, written in a program format). The response generation model is data in which information for determining a response mode is described. Here, for convenience, the first dialogue system (1041) is a Tokyo sightseeing guidance system, and the nth dialogue system (104n) is a Tokyo administrative service guidance system.
Note that the dialog device (A) and the plurality of dialog systems are not necessarily connected to be able to communicate via the network (1), but at least the dialog device (A) and the dialog system are important. Any configuration may be employed as long as at least a dialogue scenario and a response generation model of the dialogue system can be transmitted to the dialogue apparatus (A) by being connected to each other.

  The auxiliary dialogue scenario (1012) and the auxiliary response generation model (1032) stored and stored in the external storage device (17) of the dialogue device (A) are stored in the RAM (15) under the control of the control unit (80). Stored in the storage area.

Furthermore, the control unit (80) generates processing scenario information (1003) and stores it in a predetermined storage area of the RAM (15). Note that the processing scenario information (1033) generated here is, for example, a null value as initial information.
Hereinafter, when it is described that “read XX from RAM (15)”, it means “read XX from a predetermined storage area in which XX is stored in RAM (15)”.

  The input to the interactive device (A) includes, for example, voice uttered by a user who is a user of the interactive device (A), text input using a keyboard, mouse input, touch panel input, button operation, gesture input, or A combination of some of these may be considered. In the first embodiment, it is assumed that dialogue processing by voice is assumed as an example, and the input is voice.

  The voice uttered by the user is picked up by the microphone (30) of the dialogue apparatus (A) (step S1). The user's voice collected by the microphone (30) is input to the input processing unit (100) of the interactive apparatus (A) as a collected sound signal.

The input processing unit (100) of the dialog device (A) performs information processing that enables the dialog device (A) to execute the dialog processing on the collected sound signal.
As a specific example, the input processing unit (100) performs A / D conversion or the like on the collected sound signal to convert it into a discrete signal, and detects and detects a speech section for the discrete signal. Perform voice analysis processing such as frequency domain conversion of the voice segment. Further, the input processing unit (100) performs an appropriate acoustic model (for example, a probability model that gives a relationship between the pronunciation of a word and a speech feature amount as a probability) and a language model for the discrete signals subjected to these processes. (For example, a probabilistic model that gives a co-occurrence relationship between words as a probability.) Or the like is used to obtain text (speech recognition result) corresponding to a sound collection signal (speech made by the user). In addition, the input processing unit (100) extracts characteristic keywords and text types (for example, a question format and a response format) from the obtained text, and uses a keyword attribute value correspondence table or the like to extract them. Output as an input conversion result (1001) in the form of attribute-value pairs. The input processing unit (100) also generates and outputs processing scenario instruction information (1002) for specifying a processing scenario from an extracted keyword or the like using, for example, the technique disclosed in Non-Patent Document 1. Step S2).
Since the input processing unit (100) of the interactive device (A) is achieved by a known technique (for example, see Reference Document 2, Non-Patent Document 1 above), the detailed configuration of the input processing unit (100).・ Description of functions is omitted.
(Reference 2) “NTT Technical Journal”, Telecommunications Association, January 2004

  For example, when the user utters “I want to know the nearest station of Roppongi Hills”, the input processing unit (100) of the dialogue device (A) displays the input conversion result (1001) in the attribute-value pair format and the attribute is “ The value of “Intent type” is output as “Question”, the value of “Subject” is “Nearest station”, and the value of “Area” is “Roppongi Hills” (see FIG. 3). Further, the input processing unit (100) of the dialogue apparatus (A) outputs the processing scenario instruction information (1002) as scenario = “first dialogue scenario” based on the keywords “nearest station” and “Roppongi Hills”. (See FIG. 3). The input conversion result (1001) and the processing scenario instruction information (1002) output by the input processing unit (100) of the interactive device (A) are stored in a predetermined storage area of the RAM (15).

The dialogue scenario execution unit (101) of the dialogue device (A) reads the processing scenario information (1003) and the processing scenario instruction information (1002) from the RAM (15), and determines whether or not each information matches (step). S3).
The control unit (80) performs control so that the process of step S9 is executed when the pieces of information match, and the process of step S4 is executed when the pieces of information do not match.
At this stage, the processing scenario information (1003) is a Null value, and the processing scenario instruction information (1002) is “first dialog scenario”, so they do not match, so the processing of step S4 is executed.

Under the control of the control unit (80), the dialogue scenario execution unit (101) of the dialogue apparatus (A) determines whether or not the processing scenario instruction information (1002) read from the RAM (15) is an “auxiliary dialogue scenario”. Is determined (step S4).
The control unit (80) determines that the process scenario instruction information (1002) is not “auxiliary dialogue scenario” when the determination result is that the processing scenario instruction information (1002) is “auxiliary dialogue scenario”. In that case, control is performed to execute the process of step S5.
At this stage, since the process scenario instruction information (1002) is “first dialog scenario”, the process of step S5 is executed.

Under the control of the control unit (80), the dialogue scenario execution unit (101) of the dialogue device (A) determines whether or not the processing scenario information (1003) read from the RAM (15) is a Null value ( Step S5).
The control unit (80) performs the process of step S6 when the determination result is that the process scenario information (1003) is a Null value, and the process of step S24 when the process scenario information (1003) is not a Null value. Control to run.
At this stage, since the process scenario information (1003) is a Null value, the process of step S6 is executed.

  The control unit (80) stores information having the same contents as the processing scenario instruction information (1002) as processing scenario information (1003) in a predetermined storage area of the RAM (15) (step S6). That is, the processing scenario instruction information (1002) is copied to the processing scenario information (1003). At this stage, the processing scenario information (1003) is changed from the Null value to “first dialogue scenario” which is the processing scenario instruction information (1002).

  Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) generates the focal dialogue system setting information (1021) from the processing scenario information (1003), and stores the predetermined information in the RAM (15). Store in the area (step S7). Here, the focal dialogue system setting information (1021) is information for setting / changing the dialogue system necessary for setting a dialogue scenario corresponding to the processing scenario information (1003).

Subsequently, under the control of the control unit (80), the focal dialogue system setting unit (102) of the dialogue device (A) reads the focal dialogue system setting information (1021) from the RAM (15). Then, the focal dialogue system setting unit (102) interprets the focal dialogue system setting information (1021) and selects a dialogue system corresponding to the dialogue scenario indicated by the processing scenario information (1003). Further, the focal dialogue system setting unit (102) reads the dialogue scenario and the response generation model from the storage unit of the selected dialogue system via the network (1), and generates the focal dialogue scenario (1011) and the focal response generation. The model (1031) is stored in a predetermined storage area of the RAM (15) (step S8).
At this stage, since the processing scenario information (1003) is the “first dialog scenario”, the focus dialog system setting unit (102) is connected to the storage unit of the first dialog system (1041) via the network (1). The first dialogue scenario (1041b) and the first response generation model (1041a) are read, respectively, and stored as a focal dialogue scenario (1011) and a focus response generation model (1031) in a predetermined storage area of the RAM (15).

  Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) of the dialogue device (A) reads the focused dialogue scenario (1011) and the input conversion result (1001) from the RAM (15), and responds. Contents (1033) are generated, and the response contents (1033) are stored in a predetermined storage area of the RAM (15) (step S9). As this response content (1033), for example, a text or an agent character drawing command can be considered. In the first embodiment, since it is assumed that voice interactive processing is executed, it is assumed that the response content (1033) is in text format. As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "the nearest station is Roppongi" (see FIG. 4).

  Subsequently, under the control of the control unit (80), the response generation unit (103) of the dialogue apparatus (A) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and receives response information. (Interactive output) is generated and output (step S10). Here, as the response to the user in the interactive processing, for example, text displayed on a display, an image, an animation such as an agent character gesture, a synthesized voice output from a speaker, or a combination of these may be considered. It is done. In the first embodiment, since it is assumed that dialogue processing by voice is executed, it is assumed that the response is synthesized speech. Therefore, it is assumed that the response generation unit (103) is a waveform connection type text-to-speech synthesis unit, and the response information output from the response generation unit (103) is a synthesized speech signal in which speech waveform data is connected. However, the response generation unit (103) is not limited to the waveform connection type text-to-speech synthesizer, and may be another type of speech synthesizer.

  A specific example of the response generation unit (103) will be described. The response generation unit (103) includes a text analysis unit, a prosody generation unit, a speech waveform selection unit, and a speech synthesis unit. The response generation unit (103) receives the focus response generation model and the text as the response content and outputs a synthesized speech signal. The external storage device (17) stores a voice waveform database and a voice information database (not shown). The speech waveform database performs well-known A / D conversion on speech data read out from words and sentences, and is cut out by an appropriate synthesis unit (for example, phonemes) for assembling synthesized speech (as speech waveform segments) Audio waveform data).

The speech information database has a data structure (table) composed of entries in which various units are associated with phonemes, which are units suitable for assembling synthesized speech. Each entry in the speech information database normalizes the speech waveform segment number, which is the serial number of the speech waveform segment, the phoneme label information indicating the utterance content, the phoneme duration information indicating the phoneme duration, and the average power of the phoneme segment. Power information obtained, F ₀ pattern information representing the time transition of phoneme pitches, information indicating the position of speech waveform data in the speech waveform database, for example, an index indicating the speaker nature such as gender distinction ( Hereinafter, it is referred to as voice waveform data position information).
The speech information database entry and each speech waveform data (as a speech waveform segment) in the speech waveform database are associated with each other by speech waveform data position information in the speech information database.

  The text analysis unit performs morphological analysis on the input text and outputs a phoneme string and an accent type corresponding to the input text.

Prosody generation unit, as input information and focus response generation model text analyzer is output, F ₀ pattern (fundamental frequency pattern) of phonemes each speech phoneme duration length (the length of the phoneme uttered), power information (Sound volume) is estimated and output. The focus response generation model (i-th response generation model) describes text-to-speech synthesis parameters that specify the speech speed, voice pitch, speaker characteristics, and the like of synthesized speech. At this stage, the first response generation model (1041a) is used. For example, when pitch = "200Hz", the average voice height is 200Hz, speed = "fast" is used to increase the tone speed, and power = "normal" The normal voice volume is designated as (see FIG. 5).

Speech waveform selector, in accordance with the arrangement of the sequence of phonemes focus response generation model and text analysis unit has output, and output by the prosody generation part, F ₀ pattern of phonemes each speech phoneme duration length, power information, response generation model (For example, the first response generation model (1041a) and the speaker is specified as female by voicetype = "female" at this stage). A combination of speech waveform segments (optimum speech waveform segment sequence) that minimizes connection distortion between speech waveform segments when the speech waveform segments are connected from the speech information database. The selected speech waveform segment number of the optimal speech waveform segment sequence (corresponding to the sequence of phoneme sequences output by the text analysis unit) is output. Dynamic programming or the like is used to determine the optimum speech waveform segment sequence.

  The speech synthesizer receives each speech waveform segment number of the optimum speech waveform segment sequence selected by the speech waveform selection unit as input, and speech waveform data corresponding to each speech waveform segment number of this optimum speech waveform segment sequence Are read from the speech waveform database (with reference to the speech waveform data position information), and the speech waveform data are sequentially connected and output as a continuous synthesized speech signal.

Since the response generation unit (103) of the interactive device (A) is achieved by a known technique (see, for example, Reference 3), the detailed configuration / function of the response generation unit (103) is described. Abbreviated.
(Reference 3) Japanese Patent No. 2761552

  The synthesized speech signal (response information) output by the response generation unit (103) is output as synthesized speech from the speaker (40) of the interactive device (A) (step S11). In other words, from the speaker (40), the average voice pitch of a female voice is about 200 Hz, with a slightly early tone and a normal synthesized voice, “The nearest station is Roppongi.” Is output. The user perceives this synthesized speech as a response of the dialogue process.

The user is satisfied with listening to the synthesized speech (in this example, the user is satisfied only by knowing the nearest station), may end the dialog processing, or may be interactive for further information. Processing may continue. When continuing, the user may utter words with content related to the previous conversation, or utter words with content not related to the previous conversation. In addition, a previous user may suddenly interrupt or change from a previous user and utter words of content related to the previous dialog, or utter words of content not related to the previous dialog. As described above, various cases can be considered in the dialogue processing.
However, since the present invention can deal with such various cases, further explanation will be given by taking as an example a case where the user continues the dialogue process for further information.

  It is assumed that the user receives a synthesized speech output and utters a certain word for further information. This word (speech) is picked up by the microphone (30), and the processes of steps S1 and S2 are executed. As a result, the processing scenario instruction information (1002) may be “first dialog scenario” or may not be “first dialog scenario”. For example, when the user utters “I want you to tell me a famous park around Roppongi Hills”, the processing scenario instruction information (1002) becomes the “first dialog scenario” of the first dialog system, which is the Tokyo sightseeing information system. When the user utters “Please tell me the ward office near Roppongi Station”, the processing scenario instruction information (1002) is “nth dialogue scenario” of the nth dialogue system which is the Tokyo administrative service guidance system. (See FIG. 6).

When the process scenario instruction information (1002) is “first dialog scenario”, in the process of step S3, the dialog scenario execution unit (101) reads the process scenario information (1003) and process scenario instruction information (from the RAM (15)). 1002) is read and it is determined whether or not the pieces of information match. At this stage, the processing scenario information (1003) is the “first dialog scenario” and the processing scenario instruction information (1002) is the “first dialog scenario”, so the information matches.
Therefore, the control unit (80) controls to execute the process of the next step S30.

The dialogue scenario execution unit (101) reads the processing scenario information (1003) and the focal dialogue system setting information (1021) from the RAM (15), and the focal dialogue system setting information (1021) becomes the current processing scenario information (1003). It is determined whether or not the information is information for setting / changing a dialogue system necessary for setting a dialogue scenario or the like corresponding to (Step S30).
The control unit (80) determines whether the determination result indicates that the focus dialogue system setting information (1021) is set or changed for a dialogue system necessary for setting a dialogue scenario corresponding to the current processing scenario information (1003). Is not information for setting / changing a dialog system necessary for setting a dialog scenario or the like corresponding to the current processing scenario information (1003). Is controlled to execute the processing of step S31.
At this stage, the focus dialogue system setting information (1021) is information for setting / changing the dialogue system necessary for setting the “first dialogue scenario” and the like, and the current processing scenario information (1003) is Since it is the “first dialog scenario”, the determination is established, and the processing of steps S9 to S11 is executed.

  Here, in the conventional dialogue processing, the dialogue scenario (1041b) and the response generation model (1041a) in the first dialogue system (1041) are respectively referred to as the focus dialogue scenario (1011) and the focus response generation model (1031). Since it is stored in the predetermined storage area of 15), it should be noted that the processing of the subsequent steps S9 to S11 is executed based on the dialogue scenario (1041b) and the response generation model (1041a).

  Assuming that the process scenario instruction information (1002) is not the “first dialog scenario”, consider the case where the process scenario instruction information (1002) is the “nth dialog scenario” as described above. In this case, in the process of step S3, the dialogue scenario execution unit (101) reads the processing scenario information (1003) and the processing scenario instruction information (1002) from the RAM (15), and determines whether or not each information matches. To do. At this stage, the processing scenario information (1003) is the “first dialog scenario” and the processing scenario instruction information (1002) is the “nth dialog scenario”, so the information does not match.

Therefore, in the process of step S4, the dialogue scenario execution unit (101) determines whether or not the processing scenario instruction information (1002) read from the RAM (15) is an “auxiliary dialogue scenario”.
At this stage, since the process scenario instruction information (1002) is “first dialog scenario”, the process of step S5 is executed.

In the process of step S5, the dialogue scenario execution unit (101) determines whether or not the process scenario information (1003) read from the RAM (15) is a Null value.
At this stage, since the process scenario information (1003) is not a Null value, the process of step S24 is executed under the control of the control unit (80).

  The dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) read in advance from the RAM (15) from the RAM (15), generates response contents (1033), The response content (1033) is stored in a predetermined storage area of the RAM (15) (step S24). This response content (1033) is assumed to be in text format as described above. As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "Are you sure the administrative service guidance system will guide you now?" (See FIG. 7). .

Subsequently, under the control of the control unit (80), the response generation unit (103) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and generates and outputs response information. (Step S25). Since the process of step S25 is the same as the process of step S10, description thereof is omitted.
Note that at this stage, the focus response generation model (1031) is the first response generation model (1041a).

Subsequently, the control unit (80) stores information having the same contents as the processing scenario instruction information (1002) as processing scenario information (1003) in a predetermined storage area of the RAM (15) (step S26). The process of step S26 is the same as the process of step S6.
At this stage, the processing system information (1003) is changed from the “first dialogue scenario” to the “nth dialogue scenario” which is the processing scenario instruction information (1002).

The synthesized speech signal (response information) output by the response generation unit (103) in step S25 is output as synthesized speech from the speaker (40) (step S27). The process in step S27 is the same as the process in step S11.
As described above, since the focus response generation model (1031) used in the process of step S25 is the same as the previous first response generation model (1041a), an average of female voices from the speaker (40). A normal voice with a slightly early tone, with a high voice pitch of 200 Hz, is output asking "Are you sure the government service guidance system will guide you?" Therefore, even if the dialogue scenario is changed to the auxiliary dialogue scenario, the response mode becomes the same as the previous response mode (in this case, the response mode in the first dialog system). Must not.

  It is assumed that the user perceives this synthesized speech and utters “Yes”, which is a reply of approval, for example. This sound is picked up by the microphone (30), and the processes in steps S1 and S2 are executed. As a result, in the attribute-value pair format input conversion result (1001), the value of the “intention type” attribute is “reply” and the value of the “subject” attribute is “acknowledgement”, and the processing scenario instruction information (1002) becomes an “auxiliary dialogue scenario” (see FIG. 8).

  In this case, in the process of step S3, the dialogue scenario execution unit (101) reads the processing scenario information (1003) and the processing scenario instruction information (1002) from the RAM (15), and determines whether or not each information matches. To do. At this stage, the processing scenario information (1003) is the “nth dialogue scenario” and the processing scenario instruction information (1002) is the “auxiliary dialogue scenario”.

Therefore, in the process of step S4, the dialogue scenario execution unit (101) determines whether or not the processing scenario instruction information (1002) read from the RAM (15) is an “auxiliary dialogue scenario”.
At this stage, since the process scenario instruction information (1002) is an “auxiliary dialogue scenario”, the process of step S12 is executed under the control of the control unit (80).

The dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) previously read into the RAM (15), and commands the auxiliary dialogue scenario (1012) for the input conversion result (1001). Is a focal dialogue system change command (step S12).
Here, the “command” is a command that specifies the contents of the execution process described in the auxiliary dialogue scenario (1012). The “focused dialogue system change command” is a command having a content for executing a change of the focused dialogue system.
When the command of the auxiliary dialogue scenario (1012) with respect to the input conversion result (1001) is the focal dialogue system change command, the control unit (80) performs the process of step S19 and the input conversion result (1001). When the command of the auxiliary dialogue scenario (1012) is not the focal dialogue system change command, control is performed to execute the process of step S13. The process of step S13 will be described later.
At this stage, the input conversion result (1001) has an “intent type” attribute value “reply” and an attribute “subject” value “acknowledgement”. The command of the dialogue scenario (1012) is a focal dialogue system change command. Therefore, the control unit (80) performs control so as to execute the process of step S19.

  The dialogue scenario execution unit (101) generates the focal dialogue system setting information (1021) from the processing scenario information (1003) and stores it in a predetermined storage area of the RAM (15) (step S19). The process in step S19 is the same as the process in step S7.

Subsequently, under the control of the control unit (80), the focal dialog system setting unit (102) reads the focal dialog system setting information (1021) from the RAM (15). Then, the focal dialogue system setting unit (102) interprets the focal dialogue system setting information (1021) and selects a dialogue system corresponding to the dialogue scenario indicated by the processing scenario information (1003). Further, the focal dialogue system setting unit (102) reads the dialogue scenario and the response generation model from the storage unit of the selected dialogue system via the network (1), and generates the focal dialogue scenario (1011) and the focal response generation. The model (1031) is stored in a predetermined storage area of the RAM (15) (step S20). The process in step S20 is the same as the process in step S8.
At this stage, since the processing scenario information (1003) is the “nth dialogue scenario”, the focal dialogue system setting unit (102) is stored in the nth dialogue system (104n) via the network (1). The nth dialogue scenario (104nb) and the nth response generation model (104na) are read, respectively, and stored in a predetermined storage area of the RAM (15) as the focus dialogue scenario (1011) and the focus response generation model (1031).
Note that the n-th response generation model (104na) is, for example, voice type = "male" when the speaker is male voice, pitch = "95Hz" when the average voice pitch is 95Hz, and speed = "slow" The normal voice level is designated by power = "normal" (see FIG. 9).

  Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) reads the focused dialogue scenario (1011) from the RAM (15) and generates a response content (1033) that is an initial message of the dialogue system. The response content (1033) is stored in a predetermined storage area of the RAM (15) (step S21). The process in step S21 is the same as the process in step S9. As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "I will guide you through the administrative service guidance system from now on" (see FIG. 10).

Subsequently, under the control of the control unit (80), the response generation unit (103) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and generates and outputs response information. (Step S22). The process in step S22 is the same as the process in step S10.
It should be noted that at this stage, the focus response generation model (1031) is the nth response generation model (104na).

  The synthesized speech signal (response information) output by the response generation unit (103) in step S22 is output as synthesized speech from the speaker (40) (step S23). The process in step S23 is the same as the process in step S11. As described above, since the focus response generation model (1031) used in the process of step S25 is the nth response generation model (104na), the average voice level of the male voice is higher than the speaker (40). The message will be "I will guide you through the administrative service guidance system from now on" with a synthesized voice of a normal volume with a slightly delayed tone of about 95 Hz.

  As described above, the response mode is the response mode in the first dialog system in spite of the transition from the first dialog scenario (1041b) → the auxiliary dialog scenario (1012) → the nth dialog scenario (104nb) → This is a transition to the response mode in the nth interactive system. In other words, since the response of the response style corresponding to the auxiliary dialogue scenario does not intervene, the user perceives that the user has shifted from the first dialogue system to the n-th dialogue system, so that unnecessary confusion and confusion are not caused to the user.

<Supplementary explanation 1>
Next, a supplementary explanation of the dialogue processing in the first embodiment will be given. In this supplementary explanation, when the processing scenario information (1003) is a Null value and the processing scenario instruction information (1002) of the input speech is “auxiliary dialogue scenario”, or after the processing of step S27, step S12 is executed. In this process, when the instruction of the auxiliary dialogue scenario (1012) for the input conversion result (1001) is not the focal dialogue system change command, the dialogue processing is prevented from failing. is there.
Here, supplementary explanation will be given by taking as an example a case where the processing scenario information (1002) of the input voice is “auxiliary dialogue scenario” in a state where the processing scenario information (1003) is a Null value.

  It is assumed that a user utters a certain word to start a dialogue process with the dialogue device (A). This word (speech) is picked up by the microphone (30), and the processes of steps S1 and S2 are executed. For example, when the user performs utterance of "Hello", the attribute - input conversion result value pairs form (1001), the value of which is the attribute "intention type" is the value of "greeting", which is an attribute "subjects" The processing scenario instruction information (1002) becomes “auxiliary dialogue scenario” (see FIG. 11).

  In this case, in the process of step S3, the dialogue scenario execution unit (101) reads the processing scenario information (1003) and the processing scenario instruction information (1002) from the RAM (15), and determines whether or not each information matches. To do. At this stage, the processing scenario information (1003) is a Null value, and the processing scenario instruction information (1002) is an “auxiliary dialogue scenario”.

Therefore, in the process of step S4, the dialogue scenario execution unit (101) determines whether or not the processing scenario instruction information (1002) read from the RAM (15) is an “auxiliary dialogue scenario”.
At this stage, since the process scenario instruction information (1002) is an “auxiliary dialogue scenario”, the process of step S12 is executed under the control of the control unit (80).

In the subsequent process of step S12, the dialogue scenario executing unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) read in advance in the RAM (15), and assists the input conversion result (1001). It is determined whether the command of the dialogue scenario (1012) is a focal dialogue system change command.
At this stage, since the dialogue process is started, the command of the auxiliary dialogue scenario (1012) for the input conversion result (1001) is not usually the focal dialogue system change command. Accordingly, the process of step S13 is executed under the control of the control unit (80).

  The dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) read in advance from the RAM (15) into the RAM (15), generates response content (1033) that is an initial message of the dialogue system, The response content (1033) is stored in a predetermined storage area of the RAM (15) (step S13). This response content (1033) is assumed to be in text format as described above. As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "What do you want to know?" The process of step S13 is the same as the process of step S24.

Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) determines whether or not the processing scenario information (1003) read from the RAM (15) is a Null value (step S14). .
The control unit (80) executes the process of step S17 when the process scenario information (1003) is a Null value, and executes the process of step S15 when the process scenario information (1003) is not a Null value. Control.
At this stage, since the process scenario information (1003) is a Null value, the process of step S17 is executed.

  Under the control of the control unit (80), the response generation unit (103) reads the auxiliary response generation model (1032) read in advance from the RAM (15) into the RAM (15) and the response content obtained in step S13 ( 1033) is read and response information is generated and output (step S17). At this stage, since the focus response generation model (1031) is not read into the RAM (15), the auxiliary response generation model (1032) is used for generating response information.

  The synthesized speech signal (response information) output by the response generation unit (103) in step S17 is output as synthesized speech from the speaker (40) (step S18). The process in step S18 is the same as the process in step S11.

  If the processing scenario information (1003) is not a Null value in the process of step S14 (in this case, for example, in the above example, the user returns a response other than “Yes” after the process of step S27). Thus, the case where the processing scenario information becomes “auxiliary scenario information” can be considered.), The process of step S15 is executed.

  In this case, since some focus response generation model (1031) is read into the RAM (15), the response generation unit (103) generates the focus response from the RAM (15) under the control of the control unit (80). The model (1031) and the response content (1033) obtained in step S13 are read, and response information is generated and output (step S15). The process in step S15 is similar to the process in step S10 and the like.

  The synthesized speech signal (response information) output by the response generation unit (103) in step S15 is output as synthesized speech from the speaker (40) (step S16). The process in step S16 is similar to the process in step S11 and the like.

  In this way, the failure of the dialogue process is prevented by performing the processes of step S13, step S14, step S15, step S16, step S17, and step S18. In particular, when the processes of step S15 and step S16 are performed, the response format is the same as the previous response format, so that the user is not confused.

<Supplementary explanation 2>
In the above example, consider a case where the user does not reply “yes” after the process of step S27. For example, after the process of step S27, it is assumed that the user repeats the same word “Please tell me the ward office near Roppongi Station”. In this case, the processing scenario instruction information (1002) becomes the “nth dialogue scenario”. However, in the processing of step S26, the processing scenario information (1003) is changed to “nth dialogue scenario”. In the determination process, it is determined that the processing scenario instruction information (1002) matches the processing scenario information (1003). Next, the determination process of step S30 is executed. At this stage, the focus dialog system setting information (1021) is used for setting / changing the dialog system necessary for setting the “first dialog scenario” and the like. Since the current processing scenario information (1003) is the “nth dialog scenario”, the determination is not satisfied, and the control unit (80) performs control so as to execute the process of step S31.

The dialogue scenario execution unit (101) generates the focal dialogue system setting information (1021) from the current processing scenario information (1003) and stores it in a predetermined storage area of the RAM (15) (step S31). The process in step S31 is the same as the process in step S19.
That is, at this stage, the focus dialogue system setting information (1021) is information for setting / changing the dialogue system necessary for setting the “nth dialogue scenario” or the like.

Subsequently, under the control of the control unit (80), the focal dialog system setting unit (102) reads the focal dialog system setting information (1021) from the RAM (15). Then, the focal dialogue system setting unit (102) interprets the focal dialogue system setting information (1021) and selects a dialogue system corresponding to the dialogue scenario indicated by the processing scenario information (1003). Further, the focal dialogue system setting unit (102) reads the dialogue scenario and the response generation model from the storage unit of the selected dialogue system via the network (1), and generates the focal dialogue scenario (1011) and the focal response generation. The model (1031) is stored in a predetermined storage area of the RAM (15) (step S32). The process in step S32 is the same as the process in step S20.
At this stage, since the processing scenario information (1003) is the “nth dialogue scenario”, the focal dialogue system setting unit (102) is stored in the nth dialogue system (104n) via the network (1). The nth dialogue scenario (104nb) and the nth response generation model (104na) are read, respectively, and stored in a predetermined storage area of the RAM (15) as the focus dialogue scenario (1011) and the focus response generation model (1031).

  Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) reads the focused dialogue scenario (1011) from the RAM (15) and generates a response content (1033) that is an initial message of the dialogue system. The response content (1033) is stored in a predetermined storage area of the RAM (15) (step S33). The process in step S33 is the same as the process in step S21.

  Subsequently, under the control of the control unit (80), the response generation unit (103) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and generates and outputs response information. (Step S34). The process in step S34 is the same as the process in step S22.

  The synthesized speech signal (response information) output by the response generation unit (103) in step S34 is output as synthesized speech from the speaker (40) (step S35). The process in step S35 is the same as the process in step S23.

  As described above, the same processing scenario instruction information can be obtained by performing the processing of step S30, step S31, step S32, step S33, step S34, and step S35, even though the dialog system to be used is not switched. In this case, the conventional dialogue system is prevented from being used.

  The first embodiment described so far is merely an example for explaining the application of the method of the present invention, and can be applied to other embodiments of the interactive method and apparatus.

<< Second Embodiment >>
The second embodiment of the present invention will be described below.
As described above, the second embodiment is a case where the technique of the present invention is used in the dialogue system of the above-mentioned reference 1.
Hereinafter, a second embodiment of the present invention will be described with reference to FIGS. 16 and 17, and corresponding portions in the first embodiment and the respective drawings are denoted by the same reference numerals, and redundant description will be omitted. Also, the hardware configuration example of the interactive device (B) in the second embodiment and the network configuration example with the interactive system are the same as those in the first embodiment, and the description thereof will be omitted. Note that the second embodiment assumes voice interactive processing as in the first embodiment.

In the first embodiment, the dialog system used for each user input is determined, and the dialog processing is performed according to the determined dialog system.
On the other hand, in the second embodiment, the input processing unit (100B) corresponding to the input processing unit (100) in the first embodiment determines whether the user input relates to a focal dialogue scenario corresponding to the focal dialogue system. This is a case of having a function of determining whether or not it is related to an auxiliary dialogue scenario that handles a dialogue that occurs when the system is switched.
The difference between the two is mainly that the input processing unit (100) in the first embodiment designates, for example, “first dialogue scenario” and “nth dialogue scenario” as the processing scenario instruction information. The input processing unit (100B) in the embodiment is to specify, for example, “focus dialogue scenario” and “auxiliary dialogue scenario” as the processing scenario instruction information.
In addition, about the dialog system used as the premise of 2nd Embodiment, it shall refer to the said reference document 1, and the detailed description is abbreviate | omitted.

  Also, for convenience of explanation, it is assumed that the current focal dialogue system is the first dialogue system (1041). From another point of view, this is the same as setting the initial interactive system as the first interactive system (1041).

  For example, in the same way as in the first embodiment, when the current focal dialogue system is the first dialogue system (1041), when the user utters "Please tell me the ward office near Roppongi Station" The utterance is picked up by the microphone (30) (step S1).

  Next, in response to the user input, the input processing unit (100B) switches the dialogue system from the first dialogue system (1041), which is a focal dialogue system, to the nth dialogue system (104n), which is a Tokyo administrative service guidance system. Is determined to be appropriate, and “auxiliary dialogue scenario” is designated as the processing scenario instruction information (1002) (step S2B). The input processing unit (100B) outputs the input conversion result (1001) together with the processing scenario instruction information (1002), and these are stored in a predetermined storage area of the RAM (15).

Next, under the control of the control unit (80), the dialogue scenario execution unit (101) reads the processing scenario instruction information (1002) from the RAM (15), and the processing scenario instruction information (1002) is “auxiliary dialogue scenario”. It is determined whether or not (step S3B).
When the determination result is that the processing scenario instruction information (1002) is not “auxiliary dialogue scenario” (that is, “focus dialogue scenario”), the control unit (80) performs steps S9 to S11. When the process scenario instruction information (1002) is “auxiliary dialogue scenario”, control is performed so as to execute the process of step S12.
At this stage, since the process scenario instruction information (1002) is “auxiliary dialogue scenario”, the process of step S12 is executed under the control of the control unit (80).

The dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) previously read into the RAM (15), and commands the auxiliary dialogue scenario (1012) for the input conversion result (1001). Is a focal dialogue system change command (step S12).
When the command of the auxiliary dialogue scenario (1012) with respect to the input conversion result (1001) is the focal dialogue system change command, the control unit (80) performs the process of step S19B and the input conversion result (1001). If the command of the auxiliary dialogue scenario (1012) is not the focal dialogue system change command, control is performed so as to execute the process of step S24.
At this stage, the input conversion result (1001) in the attribute-value pair format has a value of “question” for the attribute “intent type”, a value “city office” for the attribute “subject”, and a value “area” for the attribute. Is “Roppongi” (see FIG. 6). Therefore, the command of the auxiliary dialogue scenario (1012) for the input conversion result (1001) is not a focal dialogue system change command. Therefore, the control unit (80) performs control so as to execute the process of step S24.

  The dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) read in advance from the RAM (15) from the RAM (15), generates response contents (1033), The response content (1033) is stored in a predetermined storage area of the RAM (15) (step S24). This response content (1033) is assumed to be in text format as in the first embodiment. As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "Are you sure the administrative service guidance system will guide you now?" (See FIG. 7). .

Subsequently, under the control of the control unit (80), the response generation unit (103) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and receives response information (synthesized speech signal). Generate and output (step S10).
Note that at this stage, the focus response generation model (1031) is the first response generation model (1041a).

  The synthesized speech signal (response information) output by the response generation unit (103) in step S10 is output as synthesized speech from the speaker (40) (step S11).

  Since the focus response generation model (1031) used in the process of step S10 is the same as the previous first response generation model (1041a), the average voice pitch of the female voice is obtained from the speaker (40). "Sorry, the government service guidance system will guide you in the future." Is output with a normal synthesized speech with a slightly early tone of about 200Hz. Therefore, even if the dialogue scenario is changed to the auxiliary dialogue scenario, the response mode becomes the same as the previous response mode (in this case, the response mode in the first dialog system). Must not.

  It is assumed that the user perceives the above synthesized speech and utters “Yes”, which is a reply of approval, for example. This sound is picked up by the microphone (30), and the processes in steps S1 and S2B are executed. As a result, in the input conversion result (1001) in the attribute-value pair format, the value of the “intention type” that is the attribute is “reply”, and the value of the “subject” that is the attribute is “accepted”. Further, the processing scenario instruction information (1002) becomes an “auxiliary dialogue scenario” to handle the dialogue processing at the time of switching the dialogue system from the dialogue history so far (see Reference 1) (see FIG. 8).

  Next, in the process of step S3B, the dialogue scenario execution unit (101) reads the processing scenario instruction information (1002) from the RAM (15), and whether or not the processing scenario instruction information (1002) is “auxiliary dialogue scenario”. Determine whether. At this stage, since the process scenario instruction information (1002) is “auxiliary dialogue scenario”, the process of step S12 is executed under the control of the control unit (80).

Next, in the process of step S12, the dialogue scenario execution unit (101) reads the auxiliary dialogue scenario (1012) and the input conversion result (1001) read in the RAM (15) in advance, and the input conversion result (1001). It is determined whether or not the command of the auxiliary dialogue scenario (1012) is a focal dialogue system change command.
At this stage, the attribute-value pair format input conversion result (1001) is such that the value of the attribute “intention type” is “reply” and the value of the attribute “subject” is “accepted”. Therefore, the command of the auxiliary dialogue scenario (1012) for the input conversion result (1001) is a focal dialogue system change command. Therefore, the control unit (80) performs control so as to execute the process of step S19B.

  The dialogue scenario execution unit (101) generates the focal dialogue system setting information (1021) from the focal dialogue system change command, and stores it in a predetermined storage area of the RAM (15) (step S19B).

Subsequently, under the control of the control unit (80), the focal dialog system setting unit (102) reads the focal dialog system setting information (1021) from the RAM (15). Then, the focal dialogue system setting unit (102) interprets the focal dialogue system setting information (1021) and selects the dialogue system designated by the focal dialogue system change command. Further, the focal dialogue system setting unit (102) reads the dialogue scenario and the response generation model from the storage unit of the selected dialogue system via the network (1), and generates the focal dialogue scenario (1011) and the focal response generation. The model (1031) is stored in a predetermined storage area of the RAM (15) (step S20B).
At this stage, since the focus dialog system change command is the nth dialog system (104n), the focus dialog system setting unit (102) performs the nth dialog scenario (104nb) and the nth response in the nth dialog system (104n). Each of the generation models (104na) is read and stored in a predetermined storage area of the RAM (15) as a focus dialogue scenario (1011) and a focus response generation model (1031).

  Subsequently, under the control of the control unit (80), the dialogue scenario execution unit (101) reads the focal dialogue scenario (1011) from the RAM (15) (also reads the input conversion result (1001) as necessary). The response content (1033) is generated, and the response content (1033) is stored in a predetermined storage area of the RAM (15) (step S9). As a specific example, the dialogue scenario execution unit (101) outputs the response content (1033) as text = "I will guide you through the administrative service guidance system from now on" (see FIG. 10).

Subsequently, under the control of the control unit (80), the response generation unit (103) reads the focus response generation model (1031) and the response content (1033) from the RAM (15), and receives response information (synthesized speech signal). Generate and output (step S10).
It should be noted that at this stage, the focus response generation model (1031) is the nth response generation model (104na).

  The synthesized speech signal (response information) output by the response generation unit (103) in step S10 is output as synthesized speech from the speaker (40) (step S11). As described above, since the focus response generation model (1031) used in the process of step S10 is the nth response generation model (104na), the average voice level of the male voice is higher than the speaker (40). The message will be "I will guide you through the administrative service guidance system from now on" with a synthesized voice of a normal volume with a slightly delayed tone of about 95 Hz.

  As described above, the response mode is the response mode in the first dialog system in spite of the transition from the first dialog scenario (1041b) → the auxiliary dialog scenario (1012) → the nth dialog scenario (104nb) → This is a transition to the response mode in the nth interactive system. That is, also in the second embodiment, since the response in the response mode corresponding to the auxiliary dialogue scenario does not intervene, the user perceives that the user has shifted from the first dialogue system to the n-th dialogue system. It will not cause embarrassment.

  The interactive apparatus / method according to the present invention is not limited to the above-described embodiment, and can be appropriately changed without departing from the gist of the present invention. In addition, the processing described in the above embodiment may be executed not only in time series according to the order of description but also in parallel or individually as required by the processing capability of the apparatus that executes the processing. .

  Further, when the processing functions in the interactive device described in the above embodiment are realized by a computer, the processing contents of the functions that the interactive device should have are described by a program. Then, by executing this program on a computer, the processing functions in the above-described dialog device are realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. The computer-readable recording medium may be any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only). Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto-Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its storage device. When executing the process, the computer reads the program stored in its own recording medium and executes the process according to the read program. As another execution form of the program, the computer may directly read the program from the portable recording medium and execute processing according to the program, and the program is transferred from the server computer to the computer. Each time, the processing according to the received program may be executed sequentially. Also, the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition. It is good. Note that the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

  In this embodiment, the interactive apparatus is configured by executing a predetermined program on the computer. However, at least a part of the processing contents may be realized by hardware.

  The present invention is useful for a multi-dialog system that handles a wider range of topics by combining a plurality of dialog systems.

The block diagram which illustrated the hardware constitutions of the dialogue apparatus (A) concerning 1st Embodiment. The figure which shows the function structural example of a dialogue apparatus (A). The figure which shows the example of an input conversion result and process scenario instruction information. The figure which shows the example of a response content. The figure which shows the example of a response production | generation model. The figure which shows the example of an input conversion result and process scenario instruction information. The figure which shows the example of a response content. The figure which shows the example of an input conversion result and process scenario instruction information. The figure which shows the example of a response production | generation model. The figure which shows the example of a response content. The figure which shows the example of an input conversion result and process scenario instruction information. The figure which shows the processing flow in a dialogue apparatus (A) (the 1). The figure which shows the processing flow in a dialogue apparatus (A) (the 2). The figure which shows the processing flow in a dialogue apparatus (A) (the 3). The figure which shows the processing flow in a dialogue apparatus (A) (the 4). The figure which shows the function structural example of the dialogue apparatus (B) concerning 2nd Embodiment. The figure which shows the processing flow in a dialogue apparatus (B). The figure which shows the whole structure of a dialogue apparatus and a dialogue system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Network 30 Microphone 40 Speaker 100 Input processing part 101 Dialog scenario execution part 102 Focus dialog system setting part 103 Response generation part 1001 Input conversion result 1002 Process scenario instruction information 1003 Process scenario information 1011 Focus dialog scenario 1012 Auxiliary dialog scenario 1021 Focus dialog system Setting information 1031 Focus response generation model 1032 Auxiliary response generation model 1033 Response content 1041 First interaction scenario 1041a First response generation model 1041b First interaction scenario 104n nth interaction system 104na nth response generation model 104nb nth interaction scenario

Claims (4)

Each of them can communicate with a plurality of interactive systems capable of executing interactive processing by storing at least an interactive scenario and a response generation model. A dialogue method in a dialogue device in which a dialogue scenario is stored,
The input processing means of the dialog device converts the user's dialog input into a format capable of dialog processing from the user's dialog input, and instruction information indicating the dialog scenario or auxiliary dialog scenario for executing the dialog processing An input processing step that generates and outputs these,
A focused dialogue system setting means for receiving a dialogue scenario and a response generation model of a dialogue system having a dialogue scenario indicated by the indication information from the dialogue system, and executing a dialogue process for each of them. And a focus dialogue system setting step to set as a focus response generation model,
A dialog scenario executing means for generating a response content to the input conversion result using the focal dialog scenario or the auxiliary dialog scenario, and outputting the response content using the focus dialog scenario or the auxiliary dialog scenario;
And a response generation step of generating and outputting a dialog output presented to a user from the response content using the focus response generation model. A dialogue device capable of mutually communicating with a plurality of dialogue systems each capable of executing dialogue processing by storing at least a dialogue scenario and a response generation model;
Storage means for storing an auxiliary dialogue scenario capable of performing dialogue processing at least in dialogue system switching;
Generates the input conversion result obtained by converting the user's dialog input into a format that can be processed from the user's dialog input, and the instruction information that indicates the dialog scenario or auxiliary dialog scenario for executing the dialog processing, and outputs these Input processing means to
A focal dialogue that receives a dialogue scenario and a response generation model of a dialogue system having the dialogue scenario indicated by the instruction information from the dialogue system and sets them as a focal dialogue model and a focal response generation model for executing the dialogue processing, respectively. System setting means;
A dialogue scenario executing means for generating a response content for the input conversion result using the focal dialogue scenario or the auxiliary dialogue scenario, and outputting the response content;
And a response generation unit configured to generate and output a dialog output presented to the user from the response content using the focus response generation model.       A dialogue program for causing a computer to function as the dialogue device according to claim 2.       A computer-readable program recording medium on which the interactive program according to claim 3 is recorded.

Images