CN109313899A - The control method of answering device and answering device, control program - Google Patents
The control method of answering device and answering device, control program Download PDFInfo
- Publication number
- CN109313899A CN109313899A CN201780035585.2A CN201780035585A CN109313899A CN 109313899 A CN109313899 A CN 109313899A CN 201780035585 A CN201780035585 A CN 201780035585A CN 109313899 A CN109313899 A CN 109313899A
- Authority
- CN
- China
- Prior art keywords
- word
- response message
- content
- input
- concept
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/16—Sound input; Sound output
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L13/00—Speech synthesis; Text to speech systems
- G10L13/08—Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/10—Speech classification or search using distance or distortion measures between unknown speech and reference templates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Human Computer Interaction (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Acoustics & Sound (AREA)
- Machine Translation (AREA)
- Manipulator (AREA)
- Toys (AREA)
Abstract
The present invention provides the control method of a kind of answering device and answering device, control program.Even if in the case where the session scene not being found suitable for also response message can be generated based on the word that input content is included.Answering device (1) has: storage unit (13), is stored with being associated with for the first word and the first related word and the second word and the second related word；And response message generating unit (26) generates the response message that first word and second related word are mutually established to associated content in the case where first word or second word are contained in input content.
Description
Technical field
Identification is carried out the present invention relates to a kind of pair of input content and the answering device of response is carried out based on the content recognized.
Background technique
It has been known that there is a kind of answering devices as the prior art, when receiving the speech of user, according to the speech to defeated
Enter content to be identified, response is carried out according to the result recognized.For example, following patent documents 1 discloses a kind of session feelings
Scape editing device, generation and edit session scene, the device that the session scene is used to that the session with user can be made to set up
I.e. automatic conversational device.
Existing technical literature
Patent document
Patent document 1: Japanese Laid-Open Patent Publication " Japanese Unexamined Patent Publication 2010-73192 " (on April 2nd, 2010 is open)
Summary of the invention
The technical problems to be solved by the invention
However, above-mentioned conventional art is, for recognizing as a result, carrying out response based on pre-prepd session scene
Composition, but the limited amount of scene.Therefore, only right in the case where the session scene for the result for not being found suitable for recognizing
The content for being previously defined as<other>carries out response.It is similar there are response content and become moreover, when the quantity of scene is few
The problem of boring session.
The present invention be completed in view of the problem, it is intended that can be realized even if be not found suitable for it is defeated
In the case where the session scene for entering content, the answering device that the word that input content is included generates response message may be based on.
The means solved the problems, such as
The project solves, the answering device of one embodiment of the present invention, exports for the input content of user's input
Response message, the answering device have: storage unit, are stored with the first word and the first related word is mutually related content
And second word and the second related word be mutually related content；And response message generating unit, it generates relative to the input
The response message of content, first word or second word are contained in the input content, store in the storage unit
Have second word and first related word be mutually related in perhaps described first word and the second word category
In the case where the content of identical concept, the response message generating unit generates first word and second conjunctive word
Language mutually establishes the response message of associated content.
In addition, the project solves, the control method of the answering device of one embodiment of the present invention is inputted for user
Input content export response message, the answering device has storage unit, and storage unit is stored with the first word and is associated with first
Word is mutually related content and the second word and the second related word is mutually related content, the controlling party of the answering device
Method includes: searching for first word that the input content is included or related with second word from the storage unit
The exploration step of storage；And the response message generation step of response message, first word are generated relative to the input content
Language or second word are contained in the input content, are stored with second word and described first in the storage unit and close
Connection word be mutually related in the case that perhaps described first word and second word belong to identical concept, described
In response message generation step, generates first word and second related word mutually establishes the response of associated content
Message.
Invention effect
According to one method of the present invention, realize one kind in the session feelings for not being found suitable for input content even if can play
In the case where scape, the answering device of response message can be also generated based on the word that input content is included.
Detailed description of the invention
Fig. 1 is the exemplary block diagram for indicating the major part of the robot 1 of first embodiment of the invention and constituting.
Fig. 2 is an example of the word information that the robot 1 of first embodiment of the invention has in storage unit.
Fig. 3 is an example of the input and output history that the robot 1 of first embodiment of the invention has in storage unit
Figure.
Fig. 4 indicates the template for the response message for having used the robot 1 of first embodiment of the invention to have in storage unit
An example and template example text.
Fig. 5 is to indicate that the robot 1 of first embodiment of the invention generates the figure of the image of response message by inference.
Fig. 6 is to indicate that the robot 1 of first embodiment of the invention generates the different images of response message by inference
Figure.
Fig. 7 is an exemplary flow chart of the processing for indicating that the robot 1 of first embodiment of the invention executes.
Fig. 8 is to indicate that the robot 1 of second embodiment of the present invention generates the different images of response message by inference
Figure.
Fig. 9 is to indicate that the robot 1 of second embodiment of the present invention generates the different images of response message by inference
Figure.
Specific embodiment
(first embodiment)
Hereinafter, using Fig. 1~7, the embodiment of the present invention will be described in detail.
<structure of robot>
It is illustrated based on summary of the Fig. 1 to the robot 1 of present embodiment.In addition, the robot 1 of present embodiment is needle
Answering device to the input content output response message inputted by user's speech.Fig. 1 is the machine for indicating present embodiment
The exemplary block diagram that the major part of people 1 is constituted.As shown in Figure 1, robot 1 has voice input section 11, voice output
Portion 12, storage unit 13 and control unit 20.Voice input section 11 detects the speech of user.Voice input section 11 it is specific and
As long as saying the audio collections devices such as microphone.The speech for the user that voice input section 11 will test is as voice data
It is sent to aftermentioned speech recognition section 21.It is also desirable to which voice input section 11 (does not issue language during the speech of user
The time of sound) etc. specific, voice number that this is made a speech each time is carried out to primary speech (text or article of combination are made a speech)
According to being sent to speech recognition section 21.Make as the voice data that will be received from aftermentioned speech synthesis portion 27 in voice output portion 12
It is functioned for voice output to external output section.The loudspeaker that voice output portion 12 has in particular by robot 1
Etc. realizing.In addition, voice output portion 12 is built in robot 1 in the example of fig. 1, but voice output portion 12 is also possible to pacify
External device (ED) loaded on robot 1.Storage unit 13 stores the various data handled by robot 1.Present embodiment
Storage unit 13 at least stores the template of aftermentioned word information, input and output history, response message.
<examples for the data that storage unit is stored>
One example of the data stored in present embodiment based on Fig. 2~4 pair storage unit 13 is illustrated.Fig. 2 is this implementation
One example of the word information that the robot 1 of mode stores in storage unit 13, Fig. 3 are that the robot 1 of present embodiment exists
One example of the input and output history stored in storage unit 13.Fig. 4 shows the machine for having used first embodiment of the invention
One example of the template for the response message that device people 1 stores in storage unit 13 and the example text of template.
In the figure 2 example, storage unit 13 store word information to attribute belonging to word and with the pass of other words
Connection property is managed, and has the project of word, concept, related word, and concept is also equipped with part of speech, classification, utilization obstacle.Scheming
In the example shown, word indicates not only to include the nouns such as " apple ", " Osaka ", also comprising the adjectives such as " sweet tea ", " circle ",
Verb as " purchase ", the language that has with monomer the meaning, only do not have the meaning in its word not comprising auxiliary word etc.
Word.Part of speech indicates the type of part of speech belonging to each words such as " noun ", " adjective ", " verb ".Classification expression " fruit ",
Extensive classification belonging to each words such as " state's inner regions ".Utilization obstacle indicates that " What (what) ", " Where (at which) " etc. answer
Use the situation of each word.That is, in the case where the group of concept (part of speech, classification, utilization obstacle) is combined into identical situation, response
Message generating unit 26 can be judged as that these words belong to identical concept for multiple words.Related word is for each word
And it is interrelated with the word, and indicate can in the word and the template of response message at least any one combine and structure
At the different words of response message.That is, storage unit 13 is stored with each word via related word and word closes each other
The content of connection.In the example shown in the series of figures, the related word of " apple " is " sweet tea ", therefore is combined and can constitute to them
Response message " apple is sweet tea ".
In the example of fig. 3, the input and output history pair that storage unit 13 has is related with the input and output of robot 1 to be gone through
History information is managed, and has number, date and time, input and output, content, the project using word.Number is on time
Between the input content of sequence management user and the output content of robot 1, date and time is to being implemented by user or robot 1
The date and time of input and output is recorded.Input and output indicate which in user or robot 1 be the historical information be by
What person implemented, " input " indicates that the input of user, " output " indicate the output of robot in the example shown in the series of figures.Content representation by
The input and output content that user or robot 1 are implemented.The input and output content implemented by user or robot 1 is indicated using word
The word for being included.In addition, being the structure recorded to the word for using word 1~3 until these three in the example shown in the series of figures
At, but it is not necessarily limited to this.
Fig. 4 indicates template, its summary and the example text for having used template for the response message that storage unit 13 has.In addition,
In the example shown in the series of figures, storage unit 13 at least only stores the project of template, can not also store the project of summary and example text.
Template be selected for robot 1 and with word combination and be used to constitute the blank of the article of response message, in the example shown in the series of figures,
The position that group enters word is recorded as [A] and [B].For example, for " [A] be [B] why " as purpose template and
Speech, by the way that [A] and [B] is replaced into word respectively, to constitute response message.Summary indicates response message corresponding to template
Content.In the example shown in the series of figures, " [A] is that [B] why " summary is " joke " to template.Example text expression will be from user's
The example of response message of the word combination that input content obtains in the case where respective template.In addition, in the example of diagram
In, the word for being replaced into [A] is " temple Jin Ge ", and the word for being replaced into [B] is " sweet tea ".At this point, according to the example of diagram it is found that
Template includes to be replaced into the word of [A] and the word for being replaced into [B] with the associated information of which kind of relevance.For example, relevance
It is showed according to word order or auxiliary word etc..In addition, if not having matched sizing text in multiple templates, then robot 1 can also incite somebody to action
Classical Chinese as " my query mode not to right " documented by the end of figure is generated as response message.
Control unit 20 is uniformly controlled each portion of robot 1, and has speech recognition section 21, word/concept acquisition unit 22, feelings
Scape portion 23, concept determination unit 24, related word acquisition unit 25, response message generating unit 26 and speech synthesis portion 27.
Speech recognition section 21, which is directed to from the received voice data once made a speech of voice input section 11, implements speech recognition.Separately
Outside, " speech recognition " indicates to obtain the textual data for indicating speech content (input content) from the voice data of speech in the present invention
According to processing.The method of the speech recognition of speech recognition section 21 is not particularly limited, certain previous arbitrary side also can be used
Method implements speech recognition.
Word/concept acquisition unit 22 is, when speech recognition section 21 receives the text data obtained by speech recognition, by this
Text data divides into word, and the concept of each word after distinguishing is obtained by storage unit 13.In addition, word/concept acquisition unit 22
From the word after differentiation to aftermentioned first word (or second word) and the first related word (or second related word) into
Row setting.Moreover, in the case that each word of the word/concept acquisition unit 22 after differentiation is not existed in the word information of Fig. 2,
It can also be by the word new login in word information.Also, a word after differentiation is not as the different words after distinguishing
The related word of language and in the case where logging in, can also be logged in the word as the related word of different words.
Scene portion 23 obtains word in the context information that storage unit 13 stores, obtaining with word/concept acquisition unit 22
And the corresponding context information of concept of word.
The concept for the word that concept determination unit 24 is obtained for word/concept acquisition unit 22, if taken with by storage unit 13
The concept for the word that the output content just exported obtain, robot 1 is included is identical to be determined.That is, concept determination unit
Whether the input content of 24 couples of users is that content corresponding with the response message exported before will inputting determines.
The concept of word and word that related word acquisition unit 25 is obtained based on word/concept acquisition unit 22, from the word of Fig. 2
Language information obtains property relevant from the word and is the related word of the word different with the word.In addition, related word obtains
The related word of acquirement is set as being used for the word of aftermentioned reasoning by portion 25.Here, related word acquisition unit 25 is for obtaining
The condition of related word is also possible to any condition.For example, can both have been taken in the word information of Fig. 2 for specific word
The word all logged in as related word is obtained, the word for being not included in input content can also be only obtained, can also preferentially take
The concept (part of speech) for obtaining word is the word of " noun ".
Response message generating unit 26 is correspondingly generated conduct based on the instruction content from each portion, with the input content of user
The text data for the response message that robot 1 exports.Specifically, the word that word/concept acquisition unit 22 is obtained and association
Word acquisition unit 25 obtain related word at least any one, storage unit 13 store template, historical information and scene
Any one in the context information that portion 23 obtains combines as needed and generates text data.Here, response message generating unit
26 are also configured to, in the case where the related word that related word acquisition unit 25 obtains is multiple, according to the condition of regulation
Extract specific related word.For example, response message generating unit 26 is also configured to, preferentially extracts and belong to and word/general
The related word of the identical concept of word of the acquirement of acquisition unit 22 is read, generation makes these words mutually establish answering for associated content
Answer message.In addition, being also possible to if response message generating unit 26 is the composition generated as the text data of response message
It is formed arbitrarily.For example, the response message generating unit 26 can also possess following composition as word/concept acquisition unit 22:
Multiple words that input content for user is included, the word based on the front and back comprising concept belonging to each word and each word
The contextual information of language selects the input word.Here, contextual information indicates that word is classified as in input content
Subject, predicate, any other one information.For example, response message generating unit 26 is also configured to, (will divide as concept
Class) word that belongs in the word and input content of inherent noun as predicate is preferably selected with input word.
Speech synthesis portion 27 will be converted to voice number by the text data of the received response message of response message generating unit 26
According to.Speech synthesis portion 27 exports the voice data after conversion to voice output portion 12.
<movement of robot>
It is illustrated based on summary of the Fig. 5 to the movement of the robot 1 of present embodiment.The machine of Fig. 5 expression present embodiment
People 1 generates the image of response message by inference.In addition, in the following description, user implements to issue relative to robot 1
The input of " apple is sweet tea ".
Firstly, robot 1 passes through speech recognition for the hair when user issues " apple is sweet tea " relative to robot 1
Speech is converted to text data, is word by text data separation.That is, if " apple is sweet tea ", is divided into " apple ",
"Yes", and 3 words of " sweet tea ".At this point, robot 1 can also for each word after distinguishing, to the word information of Fig. 2 into
Row is logged in or is updated.
Then, robot 1 is selected as the word of the starting point of reasoning from the word after differentiation, and is set as the first word
(or second word).Here, the first word (or second word) is the input for assigning the word of feature to the input content of user
Word.In addition, the first word (or second word) is identical as word information shown in Fig. 2, has the meaning if it is with monomer
Language, then any word can also be set, preferably be set according to the priority of regulation.For example, comprising making
For predicate word be noun in the case where, set it to the first word (or second word), in the case where not including,
It can be the composition that the noun as subject is set as to the first word (or second word).In addition, either subject still
In the case that predicate does not include noun, it is also possible to for word appropriate to be set as the structure of the first word (or second word)
At.According to above-mentioned example, then " apple ", " be " and " sweet tea " in, as predicate " sweet tea " be adjective, as master
" apple " of language is that " apple " therefore in this case, is set as the first word (or second word) by noun.Moreover, never
It is set as among the word of the first word (or second word), it will be considered that be mutually related word with the first word (or second word)
Language is set as the first related word (or second related word).According to above-mentioned example, " sweet tea " is set as the first conjunctive word
Language (or second related word).
After being set to the first word (or second word) and the first related word (or second related word), machine
The word information of device people 1 according to fig. 2 obtains and sets the related word as the first related word (or second related word)
Second word (or first word)." sweet tea " according to above-mentioned example, as the first related word (or second related word)
Related word be " apple " and " muskmelon ", " apple " is set as the first word (or second word).Thus, herein will
" muskmelon " is set as the second word (or first word).
When setting to the second word (or first word), the word information of robot 1 according to fig. 2 is obtained and is set
It is set for the second related word (or first related word) for the related word of the second word (or first word).In addition, second
Preferably, concept (utilization obstacle) is " How (how) " to related word (or first related word).This is because second
In the case that the concept of related word (or first related word) is not " How (how) ", with the second word (or first word)
The case where natural article will not be become when combination, is more.According to above-mentioned example, as the second word (or first word)
The related word of " muskmelon " is " sweet tea ", " circle " and " purchase ".Here, " sweet tea " is set as the first related word (or
Two related words), therefore remaining " circle " or " purchase " can be set as the second related word (or the first conjunctive word
Language).At this point, when the word information using Fig. 2 is compared " circle " and " purchase ", the concept (utilization obstacle) of " circle "
For " How (how) ", the concept (utilization obstacle) of " purchase " is "-".In this case, " circle " is set as the by robot 1
Two related words (or first related word).
Also, robot 1 (response message generating unit 26) is as the S23 of aftermentioned Fig. 7, according to by so far
It manages the first word (or second word) obtained and the second related word (or first related word) generates response message.
An example of the method for generating response message is illustrated below.In example below, the word of Fig. 2 is believed
Breath is also stored with the word that auxiliary word etc. does not only have the meaning in its word.Specifically, the word information of Fig. 2 is stored with " be "
Word.At this point, " be " and the group of concept (part of speech, classification, utilization obstacle) of word be combined into (auxiliary word ,-, -), related word
For " apple ", " muskmelon ", " sweet tea " and " circle ".In the above example, the first word (or second word) is " apple ", the
Two related words (or first related word) are " circle ".In addition, the second word (or first word) is " muskmelon ".Thus, machine
Device people 1 uses " apple " and " circle ", the auxiliary word for further having " apple " and " muskmelon " this both sides and related word " be "
" apple is round " is generated as response message by combination.In this way, robot 1 can be based on the word that input content is included, root
It it is theorized that generation response message.
In addition, the composition for generating response message is not limited to the content.Below to the composition for generating response message
Different examples be illustrated.Firstly, robot 1 when obtain the second word (or first word) and the second related word (or
First related word) when, the historical information comprising these words is obtained from input and output history shown in Fig. 3.Historical information
Comprising the second word (or first word) and the second related word (or first related word) with the associated information of which kind of relevance.
For example, relevance is showed by word order or auxiliary word etc..According to above-mentioned example, acquirement is " muskmelon is round " as " content "
Historical information.Also, robot 1 (response message generating unit 26) passes through the second word for being included by " content " of historical information
Language (or first word) is replaced into the first word (or second word), to generate response message.That is, in the above example,
" apple is round " is generated as response message.By using " content " of input and output history, suitably determination is answered for robot 1
The word order for answering each word in message can be appropriately combined the word that auxiliary word etc. does not only have the meaning in its word.In addition,
Storage unit 13 can also be by the information of the relevance (word order and/or auxiliary word between multiple words etc.) between multiple words independently of going through
History information and separately store.
In addition, in the case where not finding the second word (or first word), robot 1 (response message generating unit 26)
Response can be generated according to the template of the first related word (or second related word) and Fig. 4 as the S24 of aftermentioned Fig. 7
The composition of message.In the above example, for example, template be " mentioning [A], there are also what? " in the case where, will with as
One related word (or second related word) " sweet tea " combination " mentioning sweet tea, there are also what? " it is generated as response message.This
When, expect in input content of the user relative to response message, comprising being built with the first related word (or second related word)
Found associated second word (or first word).Thus, robot 1 is also possible to be directed to from user relative to response message
Input content obtain word, the conjunctive word as the first related word (or second related word) for response message
Language logs in the composition of the word information of Fig. 2.It, can be in the input from user relative to response message according to the structure
Hold and obtain related word, therefore can be improved the convenience of robot 1.
Moreover, robot 1 is also possible in the case where ignoring the input content of user is the process of dialogue, as aftermentioned
Fig. 7 S26 it is such, the composition of different response messages is generated according to the template of the first word and Fig. 4.At this point, robot 1 is such as
The process that fruit is capable of the input content to user and dialogue is compared, then can also have arbitrary composition.For example, robot
1 (concept determination unit 24) obtains the content that robot 1 exports before user will input according to the input and output history of Fig. 3,
The word that the history is included is set as third word.At this point, robot 1 is using word information shown in Fig. 2 to third word
Affiliated concept with whether concept identical belonging to the first word (or second word) is determined, in different situations may be used
To be judged to implementing the input for the process for ignoring dialogue.At this point, robot 1 for example can also be by the first word (or the second word
Language) be generated as with and third word association the different response messages that combine of template." apple is sweet tea " is issued in robot 1
In the case that user issues " mentioning capital of a country is exactly the temple Jin Ge " afterwards, " temple Jin Ge " as the first word (or second word)
Concept (classification) be " state's inner region ", on the other hand, " apple " as third word is " fruit ", and concept (classification) is no
Together.
At this point, the first word (or second word) " temple Jin Ge " is generated as and and as third word " apple by robot 1
Different the answering of " fruit (food) " associated template " [A] is not food " of fruit " concept (classification) or " [A] cannot be eaten " combination
Answer message.That is, according to example text, by any one in " temple Jin Ge is not food ", " temple Jin Ge cannot be eaten " as response message
Output.
In addition, the example that the method for generating different response messages is not described in restriction.For example, robot 1 can also be with
The words different from the first word (or second word) for being included using the input content of user are as the 4th word, by the first word
Language (or second word), the 4th the word response message different from template combination producing." apple is sweet tea " is issued in robot 1
In the case that user issues " temple Jin Ge is also sweet tea " afterwards, the first word (or second word) " temple Jin Ge " and third word " apple
The concept (classification) of fruit " is different, but shares the word of " sweet tea ".At this point, contained by input content of the robot 1 by user, with
The different word " sweet tea " of first word (or second word) " temple Jin Ge " is used as the 4th word.Also, robot 1 makes the first word
Language (or second word), the 4th word be generated as template " [A] be [B] why? ", " [A] is [B], is please gone next time with me
Eat " and " [A] is [B] " in the different response message of any one combination producing.That is, " temple Jin Ge is for output according to example text
Sweet tea, why? ", any one in " temple Jin Ge is sweet tea, please goes next time to eat with me " and " temple Jin Ge sweet tea ".
In addition, reasoning is not limited to above-mentioned method, in order to the first word (or second word) and template group symphysis
At response message, if it is possible to obtain the second required related word (or first related word), then be also possible to arbitrary side
Method.It is illustrated using different examples of the Fig. 6 to reasoning.Fig. 6 indicates that the robot 1 of present embodiment generates by inference and answers
Answer the different images of message.
(a) of Fig. 6 indicates " apple is sweet tea ", " muskmelon is round " the two articles.A side in two articles is defeated
Enter content, another party is the past input and output content for being stored in storage unit 13.For example, " apple is sweet tea " is in input
Hold, " muskmelon is round " is past input and output content.Specifically, robot 1 receives the input content of user, " apple is
Sweet tea ", " apple " is set as the first word.Later, the first word " apple " and concept (word are retrieved from storage unit by robot 1
Class, classification, utilization obstacle) combination (noun, fruit, What (what)) and include the second word " sweet tea as same concept
The historical information " muskmelon is round " of melon ", to obtain the second word " muskmelon ".First word " apple " and the second word " sweet tea
Melon " is same concept, and in this case, robot 1 is that identical word is generated two using concept as (b) of Fig. 6
Map made of paper sums.Also, in the region for being set with (noun, fruit, What (what)), to " apple " and " muskmelon "
In any one set.
At this point, " apple is sweet tea ", " apple is round ", " muskmelon is sweet tea " and " muskmelon can be obtained from (b) of Fig. 6
It is round " this 4 articles.As a result, robot 1 can will include not establish associated two words mutually in storage unit 13
Combined article " apple is round " and " muskmelon is sweet tea " obtained as the result of reasoning, and generate response message.
Also, above-mentioned reasoning is also possible to combine with the generation of the response message of the session scene based on conventional art
Come the composition used.At this point, it is preferred that storage unit 13 there is also have context information.In this case, robot 1 is also possible to
Established in the case that associated context information is stored in storage unit 13 with the input content of user, as aftermentioned Fig. 7 S25 that
Sample, the composition of the first word for being included according to input content (or second word) and session scenarios generation response message.In addition,
It is also possible to implement to generate constituting for response message by reasoning when context information is not stored in storage unit 13.
<process of processing>
Then, the process of the processing executed based on robot 1 of the Fig. 7 to present embodiment is illustrated.Fig. 7 is to indicate this implementation
One exemplary flow chart of the process for the processing that the robot 1 of mode executes.In addition, in the following description, user's
It include the first word and the first related word, the response message that robot 1 exports before user will input in input content
In include third word.
Firstly, the input for the user that speech recognition section 21 receives for voice input section 11, implements the voice of input content
Identification generates text data (S11).Then, word/concept acquisition unit 22 receives the text data generated in S11, from the text
The word information that data and storage unit 13 store obtains the concept (S12) of the first word, the first related word and the first word.
Also, scene portion 23 is from retrieving corresponding with the concept of the first word or the first word that obtain in S12 scene in storage unit 13
Information (S13), to whether finding that corresponding context information is determined (S14).It is being judged to not finding corresponding context information
The case where (being no in S14) under, concept determination unit 24 to before user will input robot 1 whether replied (speech) into
Row determines (S15).On the other hand, in the case where being determined as the case where having found corresponding context information (being yes in S14), processing enters
Aftermentioned S25.
In the processing of S15, in the case where being determined as the case where robot 1 has carried out response (being yes in S15), concept determination unit
24 further whether identical to the concept of the first word and third word are determined (S16).Also, it is being determined as that concept is identical
The case where (being yes in S16) under, the first related word of the word information that related word acquisition unit 25 is stored from storage unit 13
Related word retrieves the second word (S17: exploring step).On the other hand, it is being determined as the case where robot 1 does not carry out response
Under (being no in S15), processing is directly entered S17.
After S17, whether related word acquisition unit 25 is to finding the second word (whether the related word of the first related word
It is word information) determined (S18).In the case where being determined as the case where not finding the second word (being no in S18), related word is taken
Obtain the word information retrieval that portion 25 is stored from storage unit 13 and the second word (S19) that the first word is same concept.Also, it closes
Join word acquisition unit 25 in the processing of S19 to whether finding that the second word is determined (S20).It is being judged to finding the second word
It is determined as under the case where having found the second word (being yes in S18) under the case where language (being yes in S20) or in S18, conjunctive word
Language acquisition unit 25 retrieves the second related word (S21) from the related word of the second word in the word information of storage unit 13.
Later, whether related word acquisition unit 25 is to finding (the association of the second word of the second related word in the processing of S21
Whether word is word information) determined (S22).It is being determined as the case where having found the second related word (being yes in S22)
Under, response message generating unit 26 generates response message (S23: response message generation step according to the first word and the second related word
Suddenly), processing enters S27.In this way, response message generating unit 26 generate the first word mutually established with the second related word it is associated
The response message of content.On the other hand, under the case where not finding the second word in the processing of S20 (being no in S20) or S22
Processing under the case where not finding the second related word (being no in S22), response message generating unit 26 is according to the first conjunctive word
Language and template generate response message (S24).Later, processing enters S27.In addition, having found corresponding scene in being determined as S14
Information the case where (being yes in S14) under, response message generating unit 26 generates response message according to the first word and context information
(S25), processing enters S27.Also, the different situation (S16 of concept of the first word and third word in being determined as S16
In be no) under, response message generating unit 26 generates response message (S26) according to the first word and template, and processing enters S27.
In S27, response message is converted to voice data by speech synthesis portion 27, and carries out language via voice output portion 12
Sound exports (S27).
By above processing, the robot 1 of present embodiment also can in the case where the session scene not being found suitable for
Enough words for being included based on input content generate response message.
(second embodiment)
Other embodiments of the invention are illustrated based on Fig. 7~Fig. 9.The robot 1 of present embodiment is including user
Input content be three or more words in the case where, generate response message this point and the institute of the word comprising three or more
State embodiment difference.In addition, for ease of description, to the component function having the same that illustrates in the embodiment
Component marks identical symbol, and the description thereof will be omitted.The robot 1 of present embodiment included for the input content of user
Three or more words, in the same manner as the embodiment to the first word (or second word) and the first related word (or
Second related word) it is set.In addition, word/concept acquisition unit 22 is directed to as the first related word (or the second conjunctive word
Language) related word and different terms different from the first word (or second word), be set as the first additional word (or second
Additional word).Here, being also possible to arbitrarily according to the benchmark that input content sets the first additional word (or second additional word)
's.For example, it is also possible to after setting the first word in the same manner as the embodiment, by remaining word, concept (word
Class) it is " part of speech " and concept (classification) is that word other than " other " is set as the first additional word (or second additional word).
In addition, being set to the first related word (or second related word) and the first additional word (or second additional word) suitable
Sequence is not particularly limited, but the word that the concept of word (part of speech) is " noun " is preferentially preferably set as the first additional word
(or second additional word).Also, it is the first word (or second word) and the first additional word (or second additional word) is raw
Associated response message is mutually established as with the second related word (or first related word) obtained by inference.
<movement of robot>
It is illustrated based on summary of the Fig. 8 and Fig. 9 to the movement of the robot 1 of present embodiment.Fig. 8 and Fig. 9 is to indicate this reality
Apply mode robot 1 generate by inference response message image figure.
(a) of Fig. 8 is to indicate for " famous product of Osaka is okonomiyaki " and " oyster in Hiroshima is nice " the two articles
Using the image of reasoning.For example, " oyster in Hiroshima is nice " is pre-stored within storage unit 13 as past input and output, it is " big
The famous product of slope is okonomiyaki " it is inputted by user.At this point, robot 1 (word/concept acquisition unit 22) sets " okonomiyaki "
For the first word (or second word), " famous product " is set as the first related word (or second related word), and by " Osaka "
It is set as the first additional word (or second additional word).Also, robot 1 (related word acquisition unit 25) is from the input of Fig. 3
It exports history and obtains the group comprising the first word (or second word) " okonomiyaki " and concept (part of speech, classification, utilization obstacle)
It is combined into identical word " oyster " and includes the word with first additional word (or second additional word) " Osaka " same concept
The historical information " oyster in Hiroshima is nice " in " Hiroshima ".Here, " okonomiyaki " and " oyster " is being classified as " food/Japanese burning
On this point cake/Takoyaki " is with " food/seafood " is different, but the part of " food " is consistent.Therefore, it is determined as " okonomiyaki "
It is the word (subordinate concept different but the consistent word of upperseat concept) that can be used in an identical manner from " oyster ".At this point,
Robot 1 (response message generating unit 26) is in the same manner as (b) of Fig. 6, raw using the identical word of concept as (b) of Fig. 8
At map made of two paper sums.Also, " Osaka " is set for the region of (noun, state's inner region, Where (at which))
And any one in " Hiroshima ", in the region setting " okonomiyaki " and " oyster " of (noun, food, What (what))
Any one.But the case where not having the meaning for the combination of set word, is more, it is therefore desirable to maintain to utilize phase
It makes a speech together the combination of the word used, and is set.For example, the first additional word (or second additional word) is " big
Slope " needs set in combination with the first word (or second word) " okonomiyaki ", similarly, " oyster " be also required to and
" Hiroshima " is set in combination.
Thus, according to (b) of Fig. 8, it can obtain that " famous product of Osaka is okonomiyaki ", " okonomiyaki of Osaka is good
Eat ", " famous product in Hiroshima is oyster " and the oyster of " Hiroshima nice " this 4 articles.It as a result is that robot 1 will the " day of Osaka
Formula sesame seed cake is nice " or " famous product in Hiroshima is oyster " result acquirement as reasoning, and response message can be generated.
(a) of Fig. 9 indicate for " famous product of Osaka is okonomiyaki " and " local and special products in Hiroshima are red autumnal leaves steamed buns " this two
The image of a article application reasoning.For example, " local and special products in Hiroshima are red autumnal leaves steamed buns " is stored in advance as past input and output
In storage unit 13, " famous product of Osaka is okonomiyaki " is inputted by user.At this point, robot 1 (word/concept acquisition unit 22)
In the same manner as with Fig. 8 the case where, " okonomiyaki " is set as the first word (or second word), " famous product " is set as the first pass
Join word (or second related word) and " Osaka " is set as the first additional word (or second additional word).Also, machine
It includes the first word (or second word) " Japanese burning that people 1 (related word acquisition unit 25) obtains from the input and output history of Fig. 3
The group of cake " and concept (part of speech, classification, utilization obstacle) is combined into identical word " red autumnal leaves steamed bun " and includes and the first additional word
The historical information " local and special products in Hiroshima are red autumnal leaves steamed buns " of the word " Hiroshima " of (or second additional word) " Osaka " same concept.
Return at this point, robot 1 (response message generating unit 26) generates two articles using the identical word of concept as (b) of Fig. 9
Map made of receiving.Here, " famous product " and " local and special products " is consistent for the combination (noun, other) of concept (part of speech, classification).No
It crosses, the word of the combination (noun, other) as concept (part of speech, classification), other words of aspect ratio shown in the word are few, because
This, which does not have to conclude, uses.In this case, according to (b) of Fig. 9, with (b) of Fig. 8 similarly, for (noun, state's inner region,
Where (at which)) region setting " Osaka " and " Hiroshima " in any one.Also, for (noun, food, What are (assorted
)) region setting " okonomiyaki " and " red autumnal leaves steamed bun " in any one.
As a result, robot 1 can obtain " famous product of Osaka is okonomiyaki ", " local and special products of Osaka are okonomiyaki ",
" famous product in Hiroshima is oyster " and " the local and special products oyster in Hiroshima " this 4 articles.As a result, robot 1 is by " the soil of Osaka
Specialty is okonomiyaki " and " famous product in Hiroshima is oyster " as reasoning result obtain, response message can be generated.
<process of processing>
Then, the process of the processing executed based on robot 1 of the Fig. 7 to present embodiment is illustrated.Fig. 7 is to indicate this implementation
One exemplary flow chart of the process for the processing that the robot 1 of mode executes.In addition, in the following description, user's is defeated
Enter comprising the first word and the first related word in content, the response message exported before user will input by robot 1
In include third word.Here, in the case where the input content of user is " famous product of Osaka is okonomiyaki ", the first word
And first related word be respectively " okonomiyaki " and " famous product ", " Osaka " be the first additional word.
The processing of S11~S22 is identical with first embodiment.It is response message generating unit 26 in the case where being in S22
Response message (S23) is generated according to the first word, the first additional word and the second related word, processing enters S27.It is specific and
First word, the first additional word and the second related word are generated as being associated with mutually foundation by speech, response message generating unit 26
Response message " local and special products of Osaka are okonomiyaki " or " famous product in Hiroshima is oyster ".In addition, not sent out in the processing of S20
The case where not finding the second related word under the case where existing second word (being no in S20) or in the processing of S22 (be in S22
It is no) under, response message generating unit 26 generates response message (S24) according to the first related word and template.Later, processing enters
S27.In addition, being determined as under the case where having found corresponding context information (being yes in S14) in S14, response message generating unit
26 generate response message (S25) according to the first word, the first additional word and context information, and processing enters S27.Also,
It is determined as in the different situation of concept (being no in S16) of the first word and third word in S16, response message generating unit 26
Response message (S26) is generated according to the first word, the first additional word and template, processing enters S27.The processing of S27 and first
Embodiment is identical.
By above processing, it is even if the robot 1 of present embodiment is not found suitable for the input content comprising user
In the case where the session scene of the case where more than three words word, it can also be answered based on the word generation that input content is included
Answer message.
(variation)
In each embodiment, robot 1 is illustrated by the composition of voice and user session, but is not limited to
This.Such as control unit 20 can also be installed on the answering device using text and user session.In addition, by user and answering device
Between voice data or text implement dialogue can also be carried out via network.
(by the example of software realization)
The control unit 20 (especially response message generating unit 26 and concept determination unit 24) of robot 1 can also be by being formed in collection
It is realized at the logic circuit (hardware) of circuit (IC chip) etc., CPU (Central Processing Unit) also can be used
It is realized using software.
In the latter case, control unit 20 have the order for the program for executing the software as each function of realization CPU,
Record the ROM (Read Only Memory) or storage dress of the described program and various data that there is computer (or CPU) can be read
It sets and (they is referred to as " recording medium "), the RAM (Random Access Memory) that described program is unfolded etc..And
And described program is read and executed from the recording medium by computer (or CPU), to reach the purpose of the present invention.Make
For the recording medium, be able to use " the tangible medium of non-transitory ", for example, tape, disk, card, semiconductor memory,
Programmable logic circuit etc..In addition, described program can also be via the arbitrary transmission medium (communication network that can transmit the program
Network, broadcast wave etc.) it supplies to the computer.Even if in addition, the transmission instantiated that the present invention passes through electronics with described program
, the mode of the data-signal for being embedded in carrier wave can also realize.
(summary)
The answering device (1) of mode 1 of the invention exports response message, the response dress for the input content of user's input
Set (1) to have: storage unit (13) is stored with the first word and the first related word is mutually related content and the second word
It is mutually related content with the second related word；Response message generating unit (26) generates answering relative to the input content
Message is answered, first word or second word are contained in the input content, store in the storage unit (13)
State that the second word and first related word are mutually related content or first word and second word belongs to phase
In the case where the content of same concept, the response message generating unit (26) generates first word and second conjunctive word
Language mutually establishes the response message of associated content.
According to the structure, in first word or second of the answering device based on the input content for being contained in user's input
It word and is mutually related with the first word of storage unit storage and the first related word and with the second word is mutually related second
Related word, is stored with that the second word and the first related word are mutually related content or the first word belongs to the second word
In the case where identical concept, the response that the first word mutually establishes associated content with the second related word can be generated and disappeared
Breath.Thus, even if realization can also be played in the case where the session scene not being found suitable for, based on the word for being contained in input content
Language generates the effect of the answering device of response message.
The answering device (1) of mode 2 of the invention is on the basis of mode 1, described in storage unit (13) storage
Second word is associated with second related word with which kind of relevance, described in response message generating unit (26) generation
First word and second related word establish the response message of associated content with the relevance.
According to the structure, answering device is based on the relevance between the second word and the second related word, Neng Gousheng
The response message of associated content is mutually established at the first word and the second related word.Thus, exist even if realization can also be played
The word and the second word and the second related word for being included based on input content in the case where the session scene not being found suitable for
Between relevance generate response message answering device effect.
The answering device (1) of mode 3 of the invention is on the basis of any way of the mode 1 or 2, in the storage
Portion (13) storage second word and first related word are mutually related in the case where content, and the response message is raw
First word is generated at portion (26) and second related word mutually establishes the response message of associated content.
According to the structure, the second word is stored in answering device storage unit and is mutually related with the first related word
In the case where content, the first word can be generated and the second related word mutually establishes the response message of associated content.Thus,
Even if realization can be played in the case where the session scene not being found suitable for, mutually closed based on the second word and the first related word
The content of connection, the first word of generation and the second related word mutually establish the effect of the answering device of the response message of associated content
Fruit.
The answering device (1) of mode 4 of the invention is on the basis of any way of the mode 1 or 2, in the storage
Portion (13) is stored with first word and in the case that second word belongs to identical concept, and the response message generates
Portion (26) generates first word and second related word mutually establishes the response message of associated content.
According to the structure, answering device is stored with the first word in storage unit and the second word belongs to identical concept
In the case where, the first word can be generated and the second related word mutually establishes the response message of associated content.Thus, it can also
Even if playing realization in the case where the session scene not being found suitable for, belonged to based on the first word and the second word identical general
The content of thought, the first word of generation and the second related word mutually establish the effect of the answering device of the response message of associated content
Fruit.
The answering device (1) of mode 5 of the invention is on the basis of any way of the mode 1~4, in the storage
In the case where content that portion (13) is stored with second word and multiple related words are mutually related, the response message is generated
Portion (26) by the related word for belonging to concept identical with first related word in the multiple related word preferentially
It is extracted as second related word, first word is generated and second related word mutually establishes associated content
The response message.
According to the structure, it can play that answering device is stored with the second word in storage unit and multiple related words are mutual
It is in the case where associated content, the related word for belonging to concept identical with the first related word in multiple related words is excellent
It first is extracted as the second related word, the first word is generated and the second related word mutually establishes the response message of associated content
Effect.Thus, even if realization can be played in the case where the session scene not being found suitable for, it is being stored with the second word and more
A related word is mutually related in the case where content, and belonging in multiple related words is identical with the first related word general
The related word of thought is preferentially extracted as the second related word, and the first word of generation is mutually established associated with the second related word
The effect of the answering device of the response message of content.
The answering device (1) of mode 6 of the invention has: defeated on the basis of the arbitrary mode of the mode 1~5
Portion (12) out, the response message is exported to outside；And concept determination unit (24), to will input in the input
Concept and input content institute belonging to the third word for being included from the response message that the output section (12) export before appearance
Whether concept is identical belonging to the input word for including is determined, the storage unit (13) is stored with the mould of the response message
Version, response message generating unit (26) concept belonging to the third word are identical as concept belonging to the input word
In the case where, it generates first word and second related word mutually establishes the response message of associated content,
In the case that the concept belonging to the third word is different from concept belonging to the input word, generated using template different
Response message.
According to the structure, response message that answering device exports before the answering device will input input content
In the situation identical with concept belonging to the input word that the input content is included of concept belonging to the third word for being included,
Generate the first word and the second related word mutually establish the response message of associated content, the concept belonging to third word with
In the case where inputting the difference of concept belonging to word, it is able to use template and generates response message.Thus, even if realization can also be played
In the case where the session scene not being found suitable for, generated based on concept belonging to concept belonging to third word and input word
The effect of the answering device of response message.
The answering device (1) of mode 7 of the invention is general belonging to the third word on the basis of mode 6
Read it is different from concept belonging to the input word in the case where, the response message generating unit (26) by the input word and
It is combined with the template of the third word association and generates the different response message.
According to the structure, answering device concept belonging to third word is different from concept belonging to input word
In the case of, can by the input word with and the template of third word association combine and generate different response messages.Thus,
Even if realization can be played in the case where the session scene not being found suitable for, the concept belonging to third word and input word institute
In the case that the concept of category is different, by the input word from and third word association the different response message of template combination producing
Answering device effect.
The answering device (1) of mode 8 of the invention is general belonging to the third word on the basis of mode 6
Read it is different from concept belonging to the input word in the case where, the response message generating unit (26) make the input word,
The 4th word that the input content the is included response message different from described in template combination producing.
According to the structure, answering device concept belonging to third word is different from concept belonging to input word
In the case where, the 4th word that can included by input word and the input content is different from template combination producing to answer
Answer message.Thus, it is general belonging to the third word even if realization can also be played in the case where the session scene not being found suitable for
Read it is different with concept belonging to word is inputted in the case where, the 4th word that is included by input word and the input content
The effect of the answering device of the response message different from template combination producing.
The answering device (1) of mode 9 of the invention is on the basis of mode 7 or 8, the response message generating unit
(26) the multiple words for being included for the input content, based on concept belonging to each word and the front and back comprising each word
The contextual information of word selects the input word.
According to the structure, answering device can be based on attribute belonging to each word and the front and back comprising each word
The contextual information selection input word of word.Thus, even if the feelings realized in the session scene not being found suitable for can also be played
Under condition, input word is selected according to the contextual information of attribute belonging to each word and the word of the front and back comprising each word, and
The effect of the answering device of the response message different from the template combination producing of third word association.
The control method of the answering device (1) of mode 10 of the invention exports response for the input content of user's input
Message, the answering device (1) have storage unit, and the storage unit is stored with the first word and the first related word is interrelated
Content and the second word and the second related word be mutually related content, the control method of the answering device includes: from institute
State the exploration of storage unit search storage related with first word that the input content is included or second word
Step (S17)；And the response message generation step (S23) of the response message relative to the input content is generated, described first
Word or second word are contained in the input content, are stored with second word and described first in the storage unit
Related word be mutually related in perhaps described first word and second word the case where belonging to the content of identical concept
Under, in the response message generation step (S23), generates first word and second related word mutually establishes pass
The response message of the content of connection.According to the structure, function and effect same as mode 1 can be played.
The answering device (1) of each mode of the invention can also be realized by computer, in this case, be counted by making
Calculation machine is acted as each portion (software elements) that the answering device (1) has and computer is made to realize the answering device
(1) the control program of answering device (1) and the computer-readable recording medium recorded to it are also contained in this hair
Bright scope.
The present invention is not limited to above-mentioned each embodiment, various changes are able to carry out in the range shown in claim
More, to different embodiments disclosed technological means is appropriately combined respectively and the embodiment that obtains also be contained in it is of the invention
Technical scope.Moreover, being capable of forming the spy of new technology by the way that disclosed technological means is combined respectively to each embodiment
Sign.
Description of symbols
1: robot (answering device)；11: voice input section；12: voice output portion (output section)；13: storage unit；20:
Control unit；21: speech recognition section；22: word/concept acquisition unit；23: scene portion；24: concept determination unit；25: related word takes
The portion of obtaining；26: response message generating unit；27: speech synthesis portion.
Claims (11)
1. a kind of answering device exports response message for the input content of user's input, which is characterized in that
The answering device has:
Storage unit, is stored with the first word and the first related word is mutually related content and the second word and the second conjunctive word
Language is mutually related content；And
Response message generating unit generates the response message for the input content,
When first word or second word are contained in the input content, described is stored in the storage unit
Two words and first related word be mutually related in perhaps described first word belong to second word it is identical
In the case where the content of concept,
The response message generating unit, which is generated, mutually establishes associated content for first word and second related word
The response message.
2. answering device according to claim 1, which is characterized in that
The storage unit stores second word is associated with second related word with which kind of relevance,
The response message generating unit is generated to be established first word and second related word with the relevance and be closed
The response message of the content of connection.
3. answering device according to claim 1 or 2, which is characterized in that
The storage unit be stored with second word and first related word be mutually related content in the case where,
The response message generating unit generates first word and second related word mutually establishes associated content
The response message.
4. answering device according to claim 1 or 2, which is characterized in that
In the case where the storage unit is stored with first word and second word belongs to the content of identical concept,
The response message generating unit generates first word and second related word mutually establishes associated content
The response message.
5. answering device according to any one of claim 1 to 4, which is characterized in that
The storage unit be stored with second word and multiple related words be mutually related content in the case where,
The response message generating unit will belong to concept identical with first related word in the multiple related word
Related word be preferentially extracted as second related word, and generate first word and second related word
Mutually establish the response message of associated content.
6. answering device according to any one of claim 1 to 5 characterized by comprising
Output section exports the response message to outside；And
Concept determination unit, to before it will input the input content from the output section export response message included
Third word belonging to concept, included with the input content input word belonging to whether concept identical is sentenced
It is fixed,
The storage unit is stored with the template of the response message,
Response message generating unit concept belonging to the third word is identical with concept belonging to the input word
In the case of, the response message that first word and second related word are mutually established to associated content is generated,
In the case that the concept belonging to the third word is different from concept belonging to the input word, generated using template
Different response messages.
7. answering device according to claim 6, which is characterized in that
In the case that the concept belonging to the third word is different from concept belonging to the input word, the response message
Generating unit is by the input word and combines with the template of the third word association and generates the different response message.
8. answering device according to claim 6, which is characterized in that
In the case that the concept belonging to the third word is different from concept belonging to the input word, the response message
Generating unit combine the input word and the 4th word that is contained in the input content with template and described in generating not
Same response message.
9. answering device according to claim 7 or 8, which is characterized in that
The response message generating unit is directed to the input content multiple words for being included, based on concept belonging to each word and
The contextual information of the word of front and back comprising each word selects the input word.
10. a kind of control method of answering device exports response message for the input content of user's input, which is characterized in that
The answering device has a storage unit, in the storage unit is stored with the first word and the first related word is mutually related
Hold and the second word and the second related word be mutually related content,
The control method of the answering device includes:
First word that is included with the input content from storage unit search or second word is related deposits
The exploration step of storage；And
The response message generation step of response message is generated for the input content,
First word or second word are contained in the input content, are stored with described second in the storage unit
Word and first related word be mutually related in perhaps described first word and second word belong to it is identical general
In the case where the content of thought,
In the response message generation step, generation, which mutually establishes first word with second related word, to be associated with
Content response message.
11. a kind of control program, for functioning computer as answering device described in claim 1, feature exists
In functioning computer as the response message generating unit.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016-114838 | 2016-06-08 | ||
| JP2016114838 | 2016-06-08 | ||
| PCT/JP2017/006333 WO2017212689A1 (en) | 2016-06-08 | 2017-02-21 | Responding device, method for controlling responding device, and control program |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109313899A true CN109313899A (en) | 2019-02-05 |
Family
ID=60578456
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201780035585.2A Pending CN109313899A (en) | 2016-06-08 | 2017-02-21 | The control method of answering device and answering device, control program |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP6644141B2 (en) |
| CN (1) | CN109313899A (en) |
| WO (1) | WO2017212689A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI725577B (en) * | 2019-10-17 | 2021-04-21 | 中國信託商業銀行股份有限公司 | Intelligent voice information quality inspection method and system |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP7166350B2 (en) * | 2018-09-13 | 2022-11-07 | 株式会社Ｎｔｔドコモ | dialogue device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101536455A (en) * | 2006-11-03 | 2009-09-16 | 朗讯科技公司 | Methods and apparatus for delivering control messages during a malicious attack in one or more packet networks |
| JP2010273192A (en) * | 2009-05-22 | 2010-12-02 | Kyocera Mita Corp | Image reading apparatus and image forming apparatus |
| CN101950288A (en) * | 2009-07-08 | 2011-01-19 | 索尼公司 | Signal conditioning package, information processing method and program |
| JP2013072887A (en) * | 2011-09-26 | 2013-04-22 | Toshiba Corp | Interactive device |
| CN104834704A (en) * | 2015-04-29 | 2015-08-12 | 深圳市梦网科技股份有限公司 | Response method and response system |
| CN105393248A (en) * | 2013-06-27 | 2016-03-09 | 国立研究开发法人情报通信研究机构 | Non-factoid question-and-answer system and method |
| CN112527975A (en) * | 2019-09-02 | 2021-03-19 | 科沃斯商用机器人有限公司 | Human-computer interaction method and device, intelligent robot and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007328283A (en) * | 2006-06-09 | 2007-12-20 | Kenwood Corp | Interaction system, program and interactive method |
-
2017
- 2017-02-21 CN CN201780035585.2A patent/CN109313899A/en active Pending
- 2017-02-21 JP JP2018522313A patent/JP6644141B2/en active Active
- 2017-02-21 WO PCT/JP2017/006333 patent/WO2017212689A1/en not_active Ceased
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101536455A (en) * | 2006-11-03 | 2009-09-16 | 朗讯科技公司 | Methods and apparatus for delivering control messages during a malicious attack in one or more packet networks |
| JP2010273192A (en) * | 2009-05-22 | 2010-12-02 | Kyocera Mita Corp | Image reading apparatus and image forming apparatus |
| CN101950288A (en) * | 2009-07-08 | 2011-01-19 | 索尼公司 | Signal conditioning package, information processing method and program |
| JP2013072887A (en) * | 2011-09-26 | 2013-04-22 | Toshiba Corp | Interactive device |
| CN105393248A (en) * | 2013-06-27 | 2016-03-09 | 国立研究开发法人情报通信研究机构 | Non-factoid question-and-answer system and method |
| CN104834704A (en) * | 2015-04-29 | 2015-08-12 | 深圳市梦网科技股份有限公司 | Response method and response system |
| CN112527975A (en) * | 2019-09-02 | 2021-03-19 | 科沃斯商用机器人有限公司 | Human-computer interaction method and device, intelligent robot and storage medium |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI725577B (en) * | 2019-10-17 | 2021-04-21 | 中國信託商業銀行股份有限公司 | Intelligent voice information quality inspection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| JPWO2017212689A1 (en) | 2019-02-14 |
| WO2017212689A1 (en) | 2017-12-14 |
| JP6644141B2 (en) | 2020-02-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jane | Systemic misogyny exposed: Translating rapeglish from the manosphere with a random rape threat generator | |
| KR102119868B1 (en) | System and method for producting promotional media contents | |
| McQuarrie et al. | The megaphone effect: Taste and audience in fashion blogging | |
| US11646026B2 (en) | Information processing system, and information processing method | |
| US11159462B2 (en) | Communication system and communication control method | |
| CN107799119A (en) | Audio preparation method, apparatus and system | |
| CN102508554A (en) | Input method with communication association, personal repertoire and system | |
| CN106874355A (en) | The collaborative filtering method of social networks and user's similarity is incorporated simultaneously | |
| Decter-Frain et al. | Why and how groups create moral heroes | |
| CN109791571A (en) | Information processing system, information processing device, information processing method, and storage medium | |
| JP6927318B2 (en) | Information processing equipment, information processing methods, and programs | |
| CN118115208A (en) | AIGC-based advertisement landing page generation system, method, medium and equipment | |
| CN109791766A (en) | Interface, the control method of Interface and control program | |
| Choi et al. | Proxona: Supporting Creators' Sensemaking and Ideation with LLM-Powered Audience Personas | |
| CN109313899A (en) | The control method of answering device and answering device, control program | |
| Dykema | Embodied Knowledges: Synesthesia and the Archive in Monique Truong’s Bitter in the Mouth | |
| Eagle et al. | Common sense conversations: understanding casual conversation using a common sense database | |
| KR20210108565A (en) | Virtual contents creation method | |
| Abbas | Through the Lens of Television: Progression in Portrayal of Pakistani Trans* Community (A study of TV shows from 2010 to 2018) | |
| KR101804679B1 (en) | Apparatus and method of developing multimedia contents based on story | |
| Allotey-Pappoe | Genre or Hypernym: Deconstructing Afrobeats in the Global Music Industry | |
| CN107680598A (en) | Information interacting method, device and its equipment based on good friend's vocal print address list | |
| CN109960442A (en) | Transmission method, device, storage medium and the electronic device of prompt information | |
| Harris | Hearing God’s Voice: Towards a Theology of Contemporary Pentecostal Revelatory Experience | |
| Labotka | Language and women’s place in drag: Power, femininity, and gay speech |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190205 |