CN109002501A - For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue - Google Patents
For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue Download PDFInfo
- Publication number
- CN109002501A CN109002501A CN201810696649.XA CN201810696649A CN109002501A CN 109002501 A CN109002501 A CN 109002501A CN 201810696649 A CN201810696649 A CN 201810696649A CN 109002501 A CN109002501 A CN 109002501A
- Authority
- CN
- China
- Prior art keywords
- message
- user
- reply
- institute
- intention type
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L51/00—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
- H04L51/02—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J11/00—Manipulators not otherwise provided for
- B25J11/0005—Manipulators having means for high-level communication with users, e.g. speech generator, face recognition means
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/1822—Parsing for meaning understanding
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/221—Announcement of recognition results
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
Abstract
According to an example embodiment of the present disclosure, a kind of method, apparatus for handling natural language dialogue, electronic equipment and computer readable storage medium are provided.Method includes providing and replying for the first of first message in response to receiving first message from the user.Method further includes determining the confidence level for being directed to the parsing result of second message in response to receiving second message from the user, and wherein second message is the feedback replied first message or first.In addition, method further includes that confidence level based on analysis result is replied to provide for the second of second message.In accordance with an embodiment of the present disclosure, user can correct the mistake in dialogue in a manner of natural language interaction, and chat robots can understand that the confidence level of result acts to provide corresponding dialogue according to dialogue, which thereby enhance chatting service quality and the user experience is improved.
Description
Technical field
Embodiment of the disclosure relates generally to artificial intelligence field, and more particularly relates to processing natural language
Method, apparatus, electronic equipment and the computer readable storage medium of dialogue.
Background technique
In recent years, the theory of " dialogue is platform (Conversation as a Platform) " is increasingly rooted in the hearts of the people, more
Begin to use conversational man-machine interaction mode come more networking products and application.Chat robots, which refer to, can pass through text
Word, voice or picture etc. realize the computer program or software of human-computer interaction, are understood that the content that user issues, and certainly
It is dynamic to respond.Chat robots can replace true man to engage in the dialogue to a certain extent, can be integrated into conversational system
It is middle to be used as automatic on-line assistant, for the scene such as intelligence chat, customer service, information query.
Natural language is the mankind language used in everyday, such as Chinese, English etc..Natural language processing refers at computer
A kind of technology of the natural language of the mankind is managed, natural language dialogue refers to that the mode of chat robots simulation people and the mankind carry out pair
Words.Due to the diversity and complexity of natural language, in chat conversations, the message that chat robots input user may
There are certain misinterpretations, and this misinterpretation once occurs, and the reply of chat robots will not meet user to be expected
's.
Summary of the invention
According to an example embodiment of the present disclosure, a kind of method, apparatus for handling natural language dialogue, electronics are provided
Equipment and computer readable storage medium.
In the first aspect of the disclosure, a kind of method for handling natural language dialogue is provided.This method comprises:
In response to receiving first message from the user, provides and replied for the first of first message;It uses by oneself in response to receiving
The second message at family determines the confidence level of the parsing result for second message, and wherein second message is to first message or the
One feedback replied;And confidence level based on analysis result, it provides and is replied for the second of second message.
In the second aspect of the disclosure, provide a kind of for handling the device of natural language dialogue.The device includes:
First replys and provides module, is configured to respond to receive first message from the user, provides for first message
One replys;Parsing result determining module is configured to respond to receive second message from the user, determines and disappear for second
The confidence level of the parsing result of breath, wherein second message is the feedback replied first message or first;And second reply mention
For module, it is configured as confidence level based on analysis result, provides and is replied for the second of second message.
In the third aspect of the disclosure, a kind of electronic equipment is provided comprising one or more processors and deposit
Storage device, storage device is for storing one or more programs.One or more programs, which are worked as, to be executed by one or more processors,
So that electronic equipment realizes method or process according to an embodiment of the present disclosure.
In the fourth aspect of the disclosure, a kind of computer-readable medium is provided, computer program is stored thereon with, it should
Method or process according to an embodiment of the present disclosure are realized when program is executed by processor.
It should be appreciated that content described in this part of the disclosure is not intended to limit the key of embodiment of the disclosure
Feature or important feature, without in limiting the scope of the present disclosure.The other feature of the disclosure will be become by description below
It must be readily appreciated that.
Detailed description of the invention
It refers to the following detailed description in conjunction with the accompanying drawings, the above and other feature, advantage and aspect of each embodiment of the disclosure
It will be apparent.In the accompanying drawings, the same or similar appended drawing reference indicates the same or similar element, in which:
Fig. 1, which shows embodiment of the disclosure, can be realized schematic diagram in example context wherein;
Fig. 2 shows the graphical users of the example dialogue between user according to an embodiment of the present disclosure and chat robots
The diagram at interface (GUI);
Fig. 3 shows according to an embodiment of the present disclosure for handling the flow chart of the method for natural language dialogue;
Fig. 4 shows according to an embodiment of the present disclosure for improving the side of dialogue understanding effect by natural language interaction
The flow chart of method;
Fig. 5 shows according to an embodiment of the present disclosure for parsing the schematic diagram of user message;
Fig. 6 shows the flow chart according to an embodiment of the present disclosure for understanding the method for quality for analyzing dialogue;
Fig. 7 shows according to an embodiment of the present disclosure for handling the block diagram of the device of natural language dialogue;And
Fig. 8 shows the block diagram that can implement the electronic equipment of multiple embodiments of the disclosure.
Specific embodiment
Embodiment of the disclosure is more fully described below with reference to accompanying drawings.Although shown in the drawings of the certain of the disclosure
Embodiment, it should be understood that, the disclosure can be realized by various forms, and should not be construed as being limited to this
In the embodiment that illustrates, providing these embodiments on the contrary is in order to more thorough and be fully understood by the disclosure.It should be understood that
It is that being given for example only property of the accompanying drawings and embodiments effect of the disclosure is not intended to limit the protection scope of the disclosure.
In the description of embodiment of the disclosure, term " includes " and its similar term should be understood as that opening includes,
I.e. " including but not limited to ".Term "based" should be understood as " being based at least partially on ".Term " one embodiment " or " reality
Apply example " it should be understood as " at least one embodiment ".Hereafter it is also possible that other specific and implicit definition.
Due to the diversity and complexity of Human Natural Language, during robot chat conversations, inevitably
There can be the situation of chat robots misinterpretation, and this misinterpretation once occurs, the reply of chat robots will be not
Meet expected from user.In order to avoid or reduce and do not meet appearance expected from user, traditional improved procedure includes providing to guarantee the minimum
Dialogue scheme and the individual operation interface of offer.The dialogue scheme guaranteed the minimum can not inherently repair dialogue misinterpretation,
It is unable to the defect of actual repair chat system.Individual operation interface needs to be repaired by modes of operation such as mouse, touch-controls
Multiple, this can interrupt normal dialog procedure, influence the fluency of dialogue, while also add the operation difficulty of conversational system.Cause
This, the repair mode validity of traditional dialogue misinterpretation is poor and efficiency is lower, and both modes are all to mend afterwards
The scheme rescued does not account for eliminating the problem of may cause understanding failure in time during talking with and understanding.
Embodiment of the disclosure proposes a kind of scheme for improving dialogue understanding effect by natural language interaction.According to this
Disclosed embodiment is not only allowed users to be corrected the mistake in dialogue in a manner of natural language interaction, but also to chat
Its robot can understand that the quality of result acts to provide corresponding dialogue according to dialogue, which thereby enhance chatting service quality
And the user experience is improved.Therefore, embodiment of the disclosure can repair dialogue misinterpretation, and user is helped to reach real
Talk with purpose, further, since just can solve dialogue using the interactive mode based on natural language in dialog procedure and understand quality
Problem avoids the process for jumping out normal dialog.Below with reference to some example embodiments of the attached drawing 1-8 detailed description disclosure.
Fig. 1, which shows embodiment of the disclosure, can be realized schematic diagram in example context 100 wherein.In example context
In 100, user 110 carries out chat conversations with chat robots 120 (also referred to as " chat engine " or " chat system ").It can
Selection of land, user 110 can be in the locals of chat robots 120, i.e. user 110 can directly carry out with chat robots 120 pair
Words.Alternatively, its local device (such as laptop computer, desktop computer, smart phone, flat also can be used in user 110
Plate computer etc.) by network and the progress chat conversations of chat robots 120, network can be arbitrary wiredly and/or wirelessly net
Network.Therefore, chat robots 120 can be both deployed in local electronic equipment, can also be deployed to remote server
Or in cloud, or by distributed deployment.
With reference to Fig. 1, user 110 sends message 121 (referred to as " first message ") to chat robots 120, chat robots
120 handle message 121 and provide corresponding reply 122 (referred to as " first replys ") to user 110.So far, user 110 with chat
The first round dialogue of its robot 120 has been completed.Optionally, message 121 and reply 122 can be word message.Alternatively
Ground, message 121 and reply 122 or speech message can be by the corresponding texts of message 121 in the scene of voice-enabled chat
This recognition result is simultaneously displayed in the display of user equipment, allows users to apparent know current conversation content.
In embodiment of the disclosure, due to replying 122 demand (such as the chat robots 120 for not being able to satisfy user 110
To the understanding of message 121, there are mistakes, cannot accurately identify the intention of user 110), user can be to chat robots 120
Further message 123 (referred to as " second message ") is sent so that for correcting or clarifying, chat robots 120 handle message
It 123 and is provided to user 110 and corresponding replys 124 (referred to as " second replys ").In accordance with an embodiment of the present disclosure, user 110
The message 123 that natural language form can be used is corrected or is clarified to message 121 and/or reply 122, thus chatting machine
Device people 120 can more accurately identify the intention of user 110 by combined message 121 and 123.In addition, according to the disclosure
Embodiment, chat robots 120 can also assess the quality that dialogue understands result automatically, and be directed to the understanding knot of different quality
Fruit provides different types of reply.For example, if chat robots 120 determine that dialogue understands that result confidence level is lower, it can
Further confirmation or clarification are obtained to inquire user, user may not met by thus avoiding or reducing directly to present to user
The reply of intention.
Fig. 2 shows the example GUI of the example dialogue between user according to an embodiment of the present disclosure and chat robots
200 diagram, for example, chat conversations shown in GUI 200 can be the user 110 above with reference to described in Fig. 1 and chat
Chat conversations between robot 120.
Optionally, after user 110 opens the chat window with chat robots 120, chat robots 120 can be first
First issue greeting message 201 (such as " you are good, I is chat robots, now you can talk with me, have a try
?.").As shown in Fig. 2, GUI 200 further includes the window that user 110 inputs message other than chat conversations 201-216
250 and user send message button 260.It should be appreciated that user 110 can also by other means (such as voice) with chat
Its robot 120 engages in the dialogue.
With reference to the 211-212 in Fig. 2, user 110 can initiate the first round with chat robots 120 in chat window and chat
Its dialogue.For example, user 110 issues message 211 (such as " I will remove San Litun "), and the reply that chat robots 120 provide
212 (such as " where joyful? ") it is intention about amusement, from the expection (such as navigation purposes) of user 110 and different.
It can be seen that there is no the intentions of correct identification user message 211 for chat robots 120 in first round dialogue 211-212.
In order to correct mistake present in first round dialogue, user 110 can issue the message 213 of natural language form
(such as " I will navigate ") corrects the misinterpretations of chat robots 120.Since message 213 is the correction to replying 212, because
And chat robots 120 can accurately identify the real of user 110 and be intended that navigation.In addition, according to an embodiment of the present disclosure
Chat robots 120 can also assess the quality that dialogue understands result.Although for example, chat robots 120 have been identified and have been led
Boat is intended to, but not high for the confidence level of the word slot information (such as " destination is San Litun ") in navigation purposes, such as
Lower than predefined thresholds, thus chat robots 120 trigger the dialog logic of a confirmation form, provide the reply including inquiry
214 (such as " it is good, it is navigating, does is San Litun destination? "), so that user 110 confirms whether word slot information is accurate.
With continued reference to Fig. 2, when user is believed by the word slot of message 215 (such as " yes ") confirmation " San Litun is destination "
After breath, chat robots can prepare to understand the intention and word slot of user at this time, then provide and reply 216 and (such as " advising
The route in three li of villages is scratched, please later ").In some embodiments, chat robots 120 can execute corresponding movement, such as
Invocation map application navigates to three li of collecting for tasks to execute from local.In accordance with an embodiment of the present disclosure, chat robots 120 are gone back
The quality that dialogue understands result can be assessed automatically, and when finding that low quality understands result, it can be with natural language interaction
Mode actively to user initiate inquire, confirmation dialogue understands correcting errors for result, thus avoiding or reducing unnecessary mistake is in
It is existing.
Fig. 3 shows according to an embodiment of the present disclosure for handling the flow chart of the method 300 of natural language dialogue.
It should be appreciated that method 300 can be executed by the chat robots 120 above with reference to described in Fig. 1 or Fig. 2.
It provides and is replied for the first of first message in response to receiving first message from the user in frame 302.Example
Such as, with reference to figure 2 above, after receiving message 211 from user 110, chat robots 120 provide corresponding reply 212.
In some embodiments, chat robots 120 can parse message 211 to obtain corresponding intention type and word slot information.?
In the example, intention type is erroneously identified as example entertaining by chat robots 120, and leading to reply 212, there are misinterpretations.
Setting for the parsing result for being directed to second message is determined in response to receiving second message from the user in frame 304
Reliability, wherein second message is the feedback replied first message or first.For example, being received with reference to figure 2 above from user 110
Message 213, wherein message 213 is clarification to message 211 and/or the correction to replying 212, then chat robots 120
Determine parsing result and its confidence level.In this example, it is intended that type has been corrected as navigation purposes, thus intention type
Confidence level is very high (such as 90%), and the confidence level of the word slot value " San Litun " in word slot type " destination " is confirmed as not
High (such as 75%).
In frame 306, confidence level based on analysis result is provided and is replied for the second of second message.For example, being directed to message
213, the confidence level of identified intention type (such as " navigation purposes ") is greater than predefined thresholds (such as 80%), and institute is really
The confidence level of fixed word slot information (such as " San Litun is destination ") is less than predefined thresholds (such as 80%, it should be understood that meaning
Graph type and word slot information can have different predefined thresholds).In this case it is necessary to provide a confirmation word slot letter
The second of breath is replied, as shown in the reply 214 in Fig. 2.In some embodiments, if intention type and word in parsing result
The confidence level of slot information is respectively greater than predefined thresholds, then illustrates chat robots with respect to correctly understanding disappearing for user
Breath, because without initiating to inquire to user.
Therefore, method 300 according to an embodiment of the present disclosure, not only user can be corrected in a manner of natural language interaction
Mistake in dialogue, and chat robots can understand that the quality of result acts to provide corresponding dialogue according to dialogue, by
This improves chatting service quality and the user experience is improved.
Fig. 4 shows according to an embodiment of the present disclosure for improving the side of dialogue understanding effect by natural language interaction
The flow chart of method 400.It should be appreciated that method 400 can be held by the chat robots 120 above with reference to described in Fig. 1 or Fig. 2
Row, in addition, method 400 (can provide corresponding return based on some user message for a wheel session described in figure 3 above
Example implementation again).
In frame 402, chat robots receive user message from user.In frame 404, the user message received is carried out certainly
Dynamic error correction.It is, for example, possible to use language model, co-occurrence statistics, word alignments etc. to attempt to be automatically repaired present in user message
Speech recognition errors, wrong word etc..
In frame 406, judge whether user message is feedback for last round of dialogue.For example, after based on automatic error-correcting
User message judges that the user message is that a wheel is normally talked with, or for the feedback of upper wheel dialogue.In some embodiments
In, template or model can be used to identify whether user message is feedback to last round of dialogue.For example, if user message
With " [word slot type] is [word slot value] " template matching, then illustrate that the user message is the correction or confirmation to word slot information, because
And this user information can be determined that it is feedback to last round of dialogue.
In some embodiments, if active user's message be used to correct speech recognition errors in previous user message or
Typing error, or if active user's message can determine that active user's message is pair for clarifying previous user message
The feedback of previous user message.In some embodiments, if active user's message is used to correct the understanding mistake in previous reply
Accidentally, it or is used to answer the inquiry in previous reply in response to current message, determines that active user's message is to previous reply
Feedback.
If determining that user message is not to illustrate that this is the dialogue of a new round, in frame to last round of feedback in frame 406
408, execute normal dialog understanding process.For example, dialogue understanding process can be called to obtain the parsing result being intended to word slot.It is right
Template, machine learning model, or both can be used to combine to realize for words understanding process.In addition, can also be wrapped in parsing result
Include the confidence level of parsing result.It should be appreciated that confidence level can be the numerical value for having fixed value range, value is higher, table
Bright chat robots are stronger to the correct confidence of parsing result.After executing normal dialog understanding process 408, held in frame 414
Row dialogue understands quality analysis.
If determining that user message is that last round of feedback in frame 410, is executed feedback dialogue and understood in frame 406
Journey.For example, chat robots it will be appreciated that for upper wheel talk with understand that the situation of result can include but is not limited to: to intention
The parsing result of word slot correction or negative, speech recognition errors existing for previous user message or wrong word are entangled
Just, express previous user message again in a manner of repetition, carry out when previous user message is imperfect supplementing, etc..One
In a little embodiments, sentiment analysis can also be carried out for the feedback of user, if expressed (such as in user message there are changeable in mood
Abuse or complaint of user etc.), chat robots are triggered specific dialog logic and are pacified to user by setting dialogue state
It comforts, and inform how user more efficiently expresses self-demand in guided manner.Talk with understanding process by feedback
410, chat robots can further identify intention type and word slot information.
In frame 412, judge whether chat robots successfully understand feedback.If successfully understood, in frame
414, it executes dialogue and understands quality analysis.Such as chat robots are based on the parsing result and corresponding confidence being intended to word slot
Information is spent, the quality of parsing result is analyzed, for low-quality parsing result, triggering clarification dialog logic initiates to be based on to user
The inquiry of natural language to confirm that parsing result it is whether correct.It is described below with reference to Fig. 6 and understands matter for analyzing dialogue
The example implementation of amount.If determining that chat robots do not understand feedback successfully, then it is corresponding right to trigger in frame 418 in frame 412
Words movement, such as trigger the movement of failure.For example, feedback dialogue understanding process can be responsible for checking automatic error-correcting and active error correction
Completion quality, for there are serious error, the user information that can not restore through automatic error-correcting or the active that can not be understood are entangled
Mistake, unsuccessfully dialogue acts for triggering, and informs that user's epicycle message can not understand, and user is guided to express in a manner of more reasonable
Self-demand.
Next, updating dialogue state in frame 416, information source needed for updating dialogue state may include: normal right
Talk about intention type and word slot information obtained in understanding process 408;The user that feedback dialogue understanding process 410 obtains is for upper wheel
The active feedback information of dialogue or user are directed to the answer of the inquiry of chat robots.Chat robots can be according to these letters
The information in breath source, which executes, updates operation, wherein the content updated may include intention type, word slot information and confidence level, and
And it also can indicate that whether still to have and fail clear intention type and word slot information etc..
In frame 418, dialogue movement is triggered.For example, chat robots can be triggered and be exported corresponding according to dialogue state
Reply movement.For completed understand and the sufficiently high dialogue of result confidence level, can export understand after intention type and
The structured messages such as word slot information;Low-quality dialogue is understood as a result, corresponding clarification dialogue movement can be exported;And it is right
In the still unsuccessful dialogue of multiple clarification and automatic error-correcting and active error correction can not successful user message, output indicates pair
Words understand the dialogue movement of failure, and user is guided preferably to express self-demand.Therefore, side according to an embodiment of the present disclosure
Method 400 can be directed to the parsing result of different confidence levels, trigger different types of movement, enable chat robots not
Determine the feedback that user is inquired in the case that user is intended to, to reduce the situation of low-quality parsing result or misinterpretation,
It effectively improves chatting service quality and the user experience is improved.
Fig. 5 shows according to an embodiment of the present disclosure for parsing the schematic diagram 500 of user message.Such as Fig. 5 institute
Show, for user message 510 (such as " 6 points of tonight helps me to reserve a box, ten people in Quanjude "), parsing result
520 can be resolved to it is as follows: intention type is predetermined dining room, includes three word slots under such intention, i.e., dining room name,
Time and number, word slot value are respectively " Quanjude ", " 2018.06.21 18:00 " and " 10 ".According to the implementation of the disclosure
Example, can calculate the confidence level of the parsing result of intention type and each word slot.
In some embodiments, the word slot in every kind of intention type can be preset.For example, for the intention of navigation,
Word slot may include starting point and destination.Alternatively, the word slot of navigation purposes can also include by way of ground, as optional word
Slot.In some embodiments, if user message has included word slot value, but chat robots only can determine the meaning of user
Graph type, without can determine that all word slot values, then chat robots can ask in reply that user clarifies or user can be with needle
The reply that chat robots provide actively is clarified.
Fig. 6 shows the flow chart according to an embodiment of the present disclosure for understanding the method 600 of quality for analyzing dialogue.
It should be appreciated that method 600 can be executed by the chat robots 120 above with reference to described in Fig. 1 or Fig. 2, in addition, method
600 can be the example implementation of the described movement 414-418 of figure 4 above.
In frame 602, the parsing result and its confidence level of user message are obtained, wherein parsing result includes word slot type and word
Slot information, as shown in figure 5 above.In frame 604, judge that parsed intention type whether there is ambiguity.If be intended to
There are ambiguities for type, then in frame 606, provide multiple intention types and select for user.For example, if being parsed from user message
Intention type similar in two confidence levels, then chat robots can trigger the dialog logic of a selection form, it is desirable that user
Select required intention type.
If ambiguity is not present in intention type, such as only parses single intention, then in frame 608, the meaning parsed is judged
Whether the confidence level of graph type is less than first threshold.If confidence level, which is less than first threshold, provides single intention in frame 610
Type confirms for user.For example, chat robots trigger the dialog logic of a confirmation form, it is desirable that user confirms intention type
Whether be user true intention.
If whether the confidence level of intention type is greater than first threshold, small in the confidence level of frame 612, grammatical term for the character slot information
In second threshold.If the confidence level of word slot information is less than second threshold, in frame 614, word slot information is provided and is confirmed for user.
Although word slot parsing result confidence level is lower than for example, confirmation, which parses the confidence level being individually intended to, is higher than predefined thresholds
Predefined threshold value, chat robots can trigger the dialog logic of a confirmation form, it is desirable that user confirms that word slot information is
It is no correct.
In frame 616, grammatical term for the character slot type whether there is ambiguity.If there are ambiguities to provide in frame 618 for word slot type
Multiple word slot types are selected for user.For example, in the case where same word slot value belongs to multiple and different word slot types, such as
Under " navigation " is intended to, word slot value " Beijing " is resolvable to " departure place " and " destination " simultaneously, at this point, chat robots can touch
Send out the dialog logic of a selection form, it is desirable that user selects the word slot type of the word slot value, that is, allowing user to select Beijing is
Hair ground or destination.
In frame 620, grammatical term for the character slot value whether there is ambiguity.If there are ambiguities to provide multiple in frame 622 for word slot value
Word slot value is selected for user.For example, multiple word slot values occur belongs to same word slot type, and the type word slot does not support multivalue
The case where, such as under " navigation " intention, " Beijing " and " Shanghai " is resolvable to word slot type " departure place " simultaneously, at this point, chatting
Its robot can trigger the dialog logic of a selection form, it is desirable that user selects the value that equivalent slot should take, that is, allows use
It is Beijing or Shanghai that family, which selects departure place,.It should be appreciated that although showing frame 612, frame 616, frame 620 in Fig. 6 is successively to hold
Row, however, they can also be performed simultaneously or to be performed with order in a different order shown in Fig. 6.
With continued reference to Fig. 6, if ambiguity is all not present for intention type and word slot information and respective confidence level all meets
Predetermined condition then illustrates that current intention type and word slot information confidence level are higher, in frame 624, based on the intention parsed
Type and storage information are replied to generate.That is, chat robots are not necessarily in the very high situation of parsing result confidence level
It inquires user, and reply can be directly generated, improve execution efficiency.
In some embodiments, from the user for answering the third message of the second reply in response to receiving, it is based on
Third message comes update intent type and word slot information, is then based on updated intention type and word slot information, provides and be directed to
The third of third message replys (such as 216 are replied shown in Fig. 2).For example, frame 606 in Fig. 6,610,614,
618, after 622, after the inquiry that user answers chat robots, can answer based on user further update
Parsing result, and further reply is provided using updated parsing result.
Fig. 7 shows according to an embodiment of the present disclosure for handling the block diagram of the device 700 of natural language dialogue.Such as figure
Shown in 7, device 700 includes: that the first reply provides module 710, is configured to respond to receive first message from the user,
It provides and is replied for the first of first message;Parsing result determining module 720 is configured to respond to receive from the user
Second message determines the confidence level of the parsing result for second message, and wherein second message is to first message or first time
Multiple feedback;And second reply provide module 730, be configured as confidence level based on analysis result, provide and disappear for second
The second of breath is replied.
In some embodiments, device 700 can also include: the first feedback determining module, be configured to respond to second
Message is used to correct the speech recognition errors or typing error in first message, or in response to second message for clarifying first
Message determines that second message is the feedback to first message.
In some embodiments, device 700 can also include: the second feedback determining module, be configured to respond to second
Message is used to correct the misinterpretation in the first reply, or is used to answer the inquiry in the first reply in response to second message,
Determine that second message is the feedback replied first.
In some embodiments, wherein parsing result determining module 720 may include: that parsing result obtains module, be matched
Being set to through parsing second message acquisition intention type and word slot information, word slot information includes word slot type and word slot value;And
Confidence determination module is configured to determine that the confidence level of intention type and the confidence level of word slot information.
In some embodiments, it wherein it may include: the first inquiry module that the second reply, which provides module 730, is configured as
It is less than predetermined threshold in response to the confidence level of at least one in intention type and word slot information, provides including intention type and word
At least one second in slot information replys for user's confirmation.
In some embodiments, it wherein it may include: the second inquiry module that the second reply, which provides module 730, is configured as
Include first intention type and second intention type in response to intention type, provides including first intention type and second intention class
The second of type is replied for selection by the user.
In some embodiments, it wherein it may include: third inquiry module that the second reply, which provides module 730, is configured as
Include multiple candidate items in response at least one in word slot type and word slot value, provides second including multiple candidate items and reply
For selection by the user.
In some embodiments, wherein it may include: the second reply generation module that the second reply, which provides module 730, matched
The confidence level being set in response to intention type and word slot information is all satisfied predetermined condition, based on the intention type and word slot parsed
Information generates the second reply.
In some embodiments, device 700 can also include: intention update module, be configured to respond to receive and
From user for answering the third message of the second reply, based on third message come update intent type and word slot information;And
Third, which is replied, provides module, is configured as based on the intention type and word slot information updated, provides for third message
Three reply.
It should be appreciated that the first reply shown in Fig. 7 provides module 710, parsing result determining module 720 and second
Replying offer module 730 can be included in the chat robots 120 with reference to described in Fig. 1 or Fig. 2.Moreover, it should manage
Solution, module shown in Fig. 7 can execute with reference to embodiment of the disclosure method or in the process the step of or movement.
Fig. 8 shows the schematic block diagram that can be used to implement the example apparatus 800 of embodiment of the disclosure.It should manage
Solution, equipment 800 can be used to implement the described device 700 or chat machine for handling natural language dialogue of the disclosure
People 120.As shown, equipment 800 includes central processing unit (CPU) 801, it can be according to being stored in read-only memory
(ROM) it the computer program instructions in 802 or is loaded into random access storage device (RAM) 803 from storage unit 808
Computer program instructions, to execute various movements appropriate and processing.In RAM 803, can also it store needed for equipment 800 operates
Various programs and data.CPU 801, ROM 802 and RAM 803 are connected with each other by bus 804.Input/output (I/O)
Interface 805 is also connected to bus 804.
Multiple components in equipment 800 are connected to I/O interface 805, comprising: input unit 806, such as keyboard, mouse etc.;
Output unit 807, such as various types of displays, loudspeaker etc.;Storage unit 808, such as disk, CD etc.;And it is logical
Believe unit 809, such as network interface card, modem, wireless communication transceiver etc..Communication unit 809 allows equipment 800 by such as
The computer network of internet and/or various telecommunication networks exchange information/data with other equipment.
Processing unit 801 executes each method and process as described above, such as method 300, method 400 and method
600.For example, in some embodiments, method can be implemented as computer software programs, machine readable by being tangibly embodied in
Medium, such as storage unit 808.In some embodiments, some or all of of computer program can be via ROM 802
And/or communication unit 809 and be loaded into and/or be installed in equipment 800.When computer program loads to RAM 803 and by
When CPU 801 is executed, the one or more movements or step of method as described above can be executed.Alternatively, in other implementations
In example, CPU 801 can be configured as execution method by other any modes (for example, by means of firmware) appropriate.
Function described herein can be executed at least partly by one or more hardware logic components.Example
Such as, without limitation, the hardware logic component for the exemplary type that can be used include: field programmable gate array (FPGA), specially
With integrated circuit (ASIC), Application Specific Standard Product (ASSP), the system (SOC) of system on chip, load programmable logic device
(CPLD), etc..
For implement disclosed method program code can using any combination of one or more programming languages come
It writes.These program codes can be supplied to the place of general purpose computer, special purpose computer or other programmable data processing units
Device or controller are managed, so that program code makes defined in flowchart and or block diagram when by processor or controller execution
Function/operation is carried out.Program code can be executed completely on machine, partly be executed on machine, as stand alone software
Is executed on machine and partly execute or executed on remote machine or server completely on the remote machine to packet portion.
In the context of the disclosure, machine readable media can be tangible medium, may include or is stored for
The program that instruction execution system, device or equipment are used or is used in combination with instruction execution system, device or equipment.Machine can
Reading medium can be machine-readable signal medium or machine-readable storage medium.Machine readable media can include but is not limited to electricity
Son, magnetic, optical, electromagnetism, infrared or semiconductor system, device or equipment or above content any conjunction
Suitable combination.The more specific example of machine readable storage medium will include the electrical connection of line based on one or more, portable meter
Calculation machine disk, hard disk, random access memory (RAM), read-only memory (ROM), Erasable Programmable Read Only Memory EPROM (EPROM
Or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage facilities or
Any appropriate combination of above content.
Although this should be understood as requiring acting in this way in addition, depicting each movement or step using certain order
Or step is executed with shown certain order or in sequential order, or requires the movement of all diagrams or step that should be performed
To obtain desired result.Under certain environment, multitask and parallel processing be may be advantageous.Similarly, although above
Several specific implementation details are contained in discussion, but these are not construed as the limitation to the scope of the present disclosure.In list
Certain features described in the context of only embodiment can also be realized in combination in single realize.On the contrary, single
Various features described in the context of realization can also be realized individually or in any suitable subcombination multiple
In realization.
Although having used the implementation specific to the language description of the structure feature and/or method logical action disclosure
Example it should be appreciated that theme defined in the appended claims is not necessarily limited to special characteristic described above or dynamic
Make.On the contrary, special characteristic described above and movement are only to realize the exemplary forms of claims.
Claims (20)
1. a kind of method for handling natural language dialogue, comprising:
In response to receiving first message from the user, provides and replied for the first of the first message;
In response to receiving the second message from the user, the confidence of the parsing result for the second message is determined
Degree, the second message are the feedbacks replied the first message or described first;And
The confidence level based on the parsing result is provided and is replied for the second of the second message.
2. according to the method described in claim 1, further include:
It is used to correct the speech recognition errors or typing error in the first message, or response in response to the second message
In the second message for clarifying the first message, determine that the second message is the feedback to the first message.
3. according to the method described in claim 1, further include:
It is used to correct the misinterpretation in first reply in response to the second message, or in response to the second message
For answering the inquiry in first reply, determine that the second message is the feedback replied described first.
4. according to the method described in claim 1, wherein determining that the confidence level of the parsing result for the second message includes:
Intention type and word slot information are obtained by parsing the second message, institute's predicate slot information includes word slot type and word slot
Value;And determine the confidence level of the intention type and the confidence level of institute's predicate slot information.
5. according to the method described in claim 4, wherein provide for the second message second reply include:
In response in the intention type and institute's predicate slot information at least one of confidence level be less than predetermined threshold, provide including
At least one described second in the intention type and institute's predicate slot information replys for user confirmation.
6. according to the method described in claim 4, wherein provide for the second message second reply include:
Include first intention type and second intention type in response to the intention type, provides including the first intention type
It replys with described the second of second intention type for user selection.
7. according to the method described in claim 4, wherein provide for the second message second reply include:
Include multiple candidate items in response at least one in institute's predicate slot type and institute's predicate slot value, provides including the multiple
Described the second of candidate item is replied for user selection.
8. according to the method described in claim 4, wherein provide for the second message second reply include:
It is all satisfied predetermined condition in response to the confidence level of the intention type and institute's predicate slot information, based on the meaning parsed
Graph type and institute's predicate slot information are replied to generate described second.
9. the method according to any one of claim 5-7, further includes:
It is used to answer the described second third message replied from the user in response to receiving, is based on the third message
To update the intention type and institute's predicate slot information;And it is based on the updated intention type and institute's predicate slot information,
The third provided for the third message is replied.
10. a kind of for handling the device of natural language dialogue, comprising:
First replys offer module, is configured to respond to receive first message from the user, provide for described first
The first of message is replied;
Parsing result determining module is configured to respond to receive the second message from the user, determine for described
The confidence level of the parsing result of second message, the second message are the feedbacks replied the first message or described first;
And second reply provide module, be configured as the confidence level based on the parsing result, provide and disappear for described second
The second of breath is replied.
11. device according to claim 10, further includes:
First feedback determining module, is configured to respond to the second message and knows for correcting the voice in the first message
Not mistake or typing error, or the second message is determined for clarifying the first message in response to the second message
It is the feedback to the first message.
12. device according to claim 10, further includes:
It is wrong for correcting the understanding in first reply to be configured to respond to the second message for second feedback determining module
Accidentally, or in response to the second message it is used to answer the inquiry in first reply, determines that the second message is to institute
State the feedback of the first reply.
13. device according to claim 10, wherein the parsing result determining module includes:
Parsing result obtains module, is configured as obtaining intention type and word slot information by parsing the second message, described
Word slot information includes word slot type and word slot value;And confidence determination module, it is configured to determine that setting for the intention type
The confidence level of reliability and institute's predicate slot information.
14. device according to claim 13, wherein the second reply offer module includes:
First inquiry module, the confidence level of at least one being configured to respond in the intention type and institute's predicate slot information
Less than predetermined threshold, provide including in the intention type and institute's predicate slot information at least one of it is described second reply for
User's confirmation.
15. device according to claim 13, wherein the second reply offer module includes:
Second inquiry module, being configured to respond to the intention type includes first intention type and second intention type, is mentioned
For including second reply of the first intention type and second intention type for user selection.
16. device according to claim 13, wherein the second reply offer module includes:
Third inquires module, at least one being configured to respond in institute's predicate slot type and institute's predicate slot value includes multiple times
Option provides described second including the multiple candidate item and replys for user selection.
17. device according to claim 13, wherein the second reply offer module includes:
Second replys generation module, be configured to respond to the intention type and institute's predicate slot information confidence level be all satisfied it is pre-
Fixed condition is generated described second based on the intention type and institute's predicate slot information that parse and replied.
18. device described in any one of 4-16 according to claim 1, further includes:
It is intended to update module, is configured to respond to receive the third for being used to answer second reply from the user
Message updates the intention type and institute's predicate slot information based on the third message;And third replys and provides module, quilt
It is configured to the updated intention type and institute's predicate slot information, the third reply for the third message is provided.
19. a kind of electronic equipment, the electronic equipment include:
One or more processors;And storage device, the phase is for storing one or more programs, one or more of programs
It is executed when by one or more of processors, so that the electronic equipment is realized according to claim 1 described in any one of -9
Method.
20. a kind of computer readable storage medium is stored thereon with computer program, realization when described program is executed by processor
Method according to claim 1 to 9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810696649.XA CN109002501A (en) | 2018-06-29 | 2018-06-29 | For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810696649.XA CN109002501A (en) | 2018-06-29 | 2018-06-29 | For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109002501A true CN109002501A (en) | 2018-12-14 |
Family
ID=64601000
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810696649.XA Pending CN109002501A (en) | 2018-06-29 | 2018-06-29 | For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109002501A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109783704A (en) * | 2019-01-03 | 2019-05-21 | 中国科学院自动化研究所 | Man-machine mixed answer method, system, device |
CN109857487A (en) * | 2019-02-02 | 2019-06-07 | 上海奔影网络科技有限公司 | Data processing method and device for task execution |
CN110232190A (en) * | 2019-06-14 | 2019-09-13 | 上海流利说信息技术有限公司 | A kind of method, apparatus, storage medium and electronic equipment that dialogue generates |
CN110289000A (en) * | 2019-05-27 | 2019-09-27 | 北京蓦然认知科技有限公司 | A kind of audio recognition method, device |
CN110472030A (en) * | 2019-08-08 | 2019-11-19 | 网易(杭州)网络有限公司 | Man-machine interaction method, device and electronic equipment |
CN111160002A (en) * | 2019-12-27 | 2020-05-15 | 北京百度网讯科技有限公司 | Method and device for analyzing abnormal information in output spoken language understanding |
CN111222322A (en) * | 2019-12-31 | 2020-06-02 | 联想(北京)有限公司 | Information processing method and electronic device |
CN111427992A (en) * | 2019-01-10 | 2020-07-17 | 华为技术有限公司 | Conversation processing method and device |
CN111737408A (en) * | 2019-03-25 | 2020-10-02 | 阿里巴巴集团控股有限公司 | Dialogue method and equipment based on script and electronic equipment |
CN111813900A (en) * | 2019-04-10 | 2020-10-23 | 北京猎户星空科技有限公司 | Multi-turn conversation processing method and device, electronic equipment and storage medium |
CN112397060A (en) * | 2019-07-31 | 2021-02-23 | 北京声智科技有限公司 | Voice instruction processing method, system, device and medium |
CN112527979A (en) * | 2020-11-10 | 2021-03-19 | 联想(北京)有限公司 | Information interaction method and device and readable storage medium |
WO2022199784A1 (en) * | 2021-03-22 | 2022-09-29 | Telefonaktiebolaget Lm Ericsson (Publ) | Electronic communications device, and methods of the electronic communications device for updating a previously sent text message |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101566998A (en) * | 2009-05-26 | 2009-10-28 | 华中师范大学 | Chinese question-answering system based on neural network |
CN101834809A (en) * | 2010-05-18 | 2010-09-15 | 华中科技大学 | Internet instant message communication system |
CN103440287A (en) * | 2013-08-14 | 2013-12-11 | 广东工业大学 | Web question-answering retrieval system based on product information structuring |
CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
CN107526723A (en) * | 2017-08-21 | 2017-12-29 | 百度在线网络技术(北京)有限公司 | A kind of method, apparatus, server and the storage medium of semantic disambiguation |
-
2018
- 2018-06-29 CN CN201810696649.XA patent/CN109002501A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101566998A (en) * | 2009-05-26 | 2009-10-28 | 华中师范大学 | Chinese question-answering system based on neural network |
CN101834809A (en) * | 2010-05-18 | 2010-09-15 | 华中科技大学 | Internet instant message communication system |
CN103440287A (en) * | 2013-08-14 | 2013-12-11 | 广东工业大学 | Web question-answering retrieval system based on product information structuring |
CN104050256A (en) * | 2014-06-13 | 2014-09-17 | 西安蒜泥电子科技有限责任公司 | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method |
CN107526723A (en) * | 2017-08-21 | 2017-12-29 | 百度在线网络技术(北京)有限公司 | A kind of method, apparatus, server and the storage medium of semantic disambiguation |
Cited By (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109783704A (en) * | 2019-01-03 | 2019-05-21 | 中国科学院自动化研究所 | Man-machine mixed answer method, system, device |
US11977815B2 (en) | 2019-01-10 | 2024-05-07 | Huawei Technologies Co., Ltd. | Dialogue processing method and device |
CN111427992A (en) * | 2019-01-10 | 2020-07-17 | 华为技术有限公司 | Conversation processing method and device |
CN109857487A (en) * | 2019-02-02 | 2019-06-07 | 上海奔影网络科技有限公司 | Data processing method and device for task execution |
CN111737408B (en) * | 2019-03-25 | 2024-05-03 | 阿里巴巴集团控股有限公司 | Dialogue method and equipment based on script, and electronic equipment |
CN111737408A (en) * | 2019-03-25 | 2020-10-02 | 阿里巴巴集团控股有限公司 | Dialogue method and equipment based on script and electronic equipment |
CN111813900A (en) * | 2019-04-10 | 2020-10-23 | 北京猎户星空科技有限公司 | Multi-turn conversation processing method and device, electronic equipment and storage medium |
CN111813900B (en) * | 2019-04-10 | 2023-12-08 | 北京猎户星空科技有限公司 | Multi-round dialogue processing method and device, electronic equipment and storage medium |
CN110289000A (en) * | 2019-05-27 | 2019-09-27 | 北京蓦然认知科技有限公司 | A kind of audio recognition method, device |
CN110232190A (en) * | 2019-06-14 | 2019-09-13 | 上海流利说信息技术有限公司 | A kind of method, apparatus, storage medium and electronic equipment that dialogue generates |
CN112397060A (en) * | 2019-07-31 | 2021-02-23 | 北京声智科技有限公司 | Voice instruction processing method, system, device and medium |
CN112397060B (en) * | 2019-07-31 | 2024-02-23 | 北京声智科技有限公司 | Voice instruction processing method, system, equipment and medium |
CN110472030A (en) * | 2019-08-08 | 2019-11-19 | 网易(杭州)网络有限公司 | Man-machine interaction method, device and electronic equipment |
CN111160002A (en) * | 2019-12-27 | 2020-05-15 | 北京百度网讯科技有限公司 | Method and device for analyzing abnormal information in output spoken language understanding |
JP2021108095A (en) * | 2019-12-27 | 2021-07-29 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Method for outputting information on analysis abnormality in speech comprehension |
KR102382421B1 (en) * | 2019-12-27 | 2022-04-05 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Method and apparatus for outputting analysis abnormality information in spoken language understanding |
US11482211B2 (en) | 2019-12-27 | 2022-10-25 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and apparatus for outputting analysis abnormality information in spoken language understanding |
JP7182584B2 (en) | 2019-12-27 | 2022-12-02 | ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド | A method for outputting information of parsing anomalies in speech comprehension |
KR20210084207A (en) * | 2019-12-27 | 2021-07-07 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Method and apparatus for outputting analysis abnormality information in spoken language understanding |
EP3843090A1 (en) * | 2019-12-27 | 2021-06-30 | Beijing Baidu Netcom Science And Technology Co. Ltd. | Method and apparatus for outputting analysis abnormality information in spoken language understanding |
CN111222322A (en) * | 2019-12-31 | 2020-06-02 | 联想(北京)有限公司 | Information processing method and electronic device |
CN112527979A (en) * | 2020-11-10 | 2021-03-19 | 联想(北京)有限公司 | Information interaction method and device and readable storage medium |
WO2022199784A1 (en) * | 2021-03-22 | 2022-09-29 | Telefonaktiebolaget Lm Ericsson (Publ) | Electronic communications device, and methods of the electronic communications device for updating a previously sent text message |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109002501A (en) | For handling method, apparatus, electronic equipment and the computer readable storage medium of natural language dialogue | |
US10460029B2 (en) | Reply information recommendation method and apparatus | |
CN108763548A (en) | Collect method, apparatus, equipment and the computer readable storage medium of training data | |
CN108877792A (en) | For handling method, apparatus, electronic equipment and the computer readable storage medium of voice dialogue | |
CN109033223A (en) | For method, apparatus, equipment and computer readable storage medium across type session | |
US20210280190A1 (en) | Human-machine interaction | |
CN112597754B (en) | Text error correction method, apparatus, electronic device and readable storage medium | |
CN106383872A (en) | Artificial intelligence-based information processing method and apparatus | |
CN107430616A (en) | The interactive mode of speech polling re-forms | |
CN109543010A (en) | The interactive method and system of fused data library inquiry | |
CN107239450B (en) | Method for processing natural language based on interactive context | |
CN112487173A (en) | Man-machine conversation method, device and storage medium | |
CN110288995A (en) | Exchange method, device, storage medium and electronic equipment based on speech recognition | |
CN116737908A (en) | Knowledge question-answering method, device, equipment and storage medium | |
CN112307188B (en) | Dialog generation method, system, electronic device and readable storage medium | |
CN115309877A (en) | Dialog generation method, dialog model training method and device | |
CN109165286A (en) | Automatic question-answering method, device and computer readable storage medium | |
CN114722171B (en) | Multi-round dialogue processing method and device, electronic equipment and storage medium | |
CN114490994B (en) | Conversation management method and device | |
CN112002325B (en) | Multi-language voice interaction method and device | |
CN113743127A (en) | Task type dialogue method and device, electronic equipment and storage medium | |
CN114202363A (en) | Artificial intelligence based call method, device, computer equipment and medium | |
CN110222161B (en) | Intelligent response method and device for conversation robot | |
CN113554062A (en) | Training method, device and storage medium of multi-classification model | |
CN112687271A (en) | Voice translation method and device, electronic equipment and storage medium |
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 |