CN110222161B - Intelligent response method and device for conversation robot - Google Patents
Intelligent response method and device for conversation robot Download PDFInfo
- Publication number
- CN110222161B CN110222161B CN201910375687.XA CN201910375687A CN110222161B CN 110222161 B CN110222161 B CN 110222161B CN 201910375687 A CN201910375687 A CN 201910375687A CN 110222161 B CN110222161 B CN 110222161B
- Authority
- CN
- China
- Prior art keywords
- user
- processing result
- conversation
- information
- feedback
- 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.)
- Active
Links
Images
Classifications
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
Abstract
The application discloses an intelligent response method and device for a conversation robot. The method comprises the steps of obtaining a user request; returning at least one processing result similar to the user request; and receiving user feedback of the processing result, and returning the next processing result according to the user feedback. The method and the device solve the technical problem that the conversation robot cannot accurately understand the intention of the user. Through the application, the conversation robot can more accurately understand various expressions of the user, and the conversation robot more fits the expression habits of the user along with the time. In addition, the method and the device are suitable for intelligent response scenes.
Description
Technical Field
The application relates to the field of intelligent conversation, in particular to an intelligent response method and device for a conversation robot.
Background
With the development of AI technology, the application of conversation robots in daily life is more and more common.
The inventor finds that the dialog robot cannot accurately understand the user's dialog intention due to the variety of the language itself and the input method, and further, the purpose of human-computer dialog cannot be achieved smoothly and the expression habit of the user cannot be fitted.
Aiming at the problem that the dialog robot in the related art cannot accurately understand the intention of the user, an effective solution is not provided at present.
Disclosure of Invention
The application mainly aims to provide an intelligent response method and device for a conversation robot, so as to solve the problem that the conversation robot cannot accurately understand the intention of a user.
In order to achieve the above object, according to a first aspect of the present application, there is provided a dialogue robot intelligent response method.
The intelligent response method of the conversation robot comprises the following steps: acquiring a user request; returning at least one processing result similar to the user request; and receiving user feedback of the processing result, and returning the next processing result according to the user feedback.
Further, returning at least one processing result similar to the user request comprises: generating a candidate set expressed by the current standard according to the information of the user dialogue; calculating candidate results in the candidate set according to a preset similarity model; and fusing the candidate results, and outputting the candidate results as processing results according to the sequence.
Further, the user dialog context information includes: domain information for judging which scene the dialog is currently processing; purpose intention information for a user to have a conversation; dialog action information for questioning information and selecting contents.
Further, after receiving user feedback on the processing result and returning the next processing result according to the user feedback, the method further includes: and judging whether to continue to execute the current conversation or to carry out clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
Further, after receiving user feedback on the processing result and returning the next processing result according to the user feedback, the method further includes: and if the user feedback indicates that the user request is correctly understood, storing the processing result into an examination library to be examined so that the conversation robot intelligently responds to the next request of the user through the understanding result in the examination library to be examined.
In order to achieve the above object, according to a second aspect of the present application, there is provided a dialogue robot intelligent response device.
The intelligent answering device of the conversation robot according to the application comprises: the acquisition module is used for acquiring a user request; a return module for returning at least one processing result similar to the user request; and the receiving module is used for receiving the user feedback of the processing result and returning the next processing result according to the user feedback.
Further, the return module is also used for generating a candidate set expressed by the current standard according to the user dialogue context information; calculating candidate results in the candidate set according to a preset similarity model; and fusing the candidate results, and outputting the candidate results as processing results according to the sequence.
Further, the user dialog context information in the return module includes: domain information for judging which scene the dialog is currently processing; purpose intention information for a user to have a conversation; dialog action information for questioning information and selecting contents.
Further, the apparatus further comprises: and the clarification interaction module is used for judging whether to continue to execute the current conversation or to carry out clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
Further, the apparatus further comprises: and the enhancement module is used for storing the processing result into the to-be-examined checking library so that the conversation robot intelligently responds to the next request of the user through the understanding result in the to-be-examined checking library if the user feeds back that the user request is correctly understood.
In order to achieve the above object, according to a third aspect of the present application, the present application provides a device for intelligent response of a conversational robot, the device comprising a processor, a display, a memory, a network interface and a bus system, wherein the processor, the display, the memory and the network interface are connected with each other through the bus system. The memory is configured to store instructions and the processor is configured to execute the instructions stored by the memory, and when executed, the processor performs the method of the first aspect or any possible implementation manner of the first aspect through the network interface.
In order to achieve the above object, according to a fourth aspect of the present application, there is provided a computer-readable medium storing a computer program including instructions for executing the intelligent answer method of the dialogue robot.
In the embodiment of the application, the intelligent response method and the intelligent response device for the dialogue robot adopt a mode of acquiring the user request, and achieve the purposes of receiving the user feedback of the processing result and returning the next processing result according to the user feedback by returning at least one processing result similar to the user request, thereby realizing the technical effects of intelligently understanding the user request and intelligently enhancing learning, and further solving the technical problem that the dialogue robot cannot accurately understand the intention of the user.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, are included to provide a further understanding of the application and to enable other features, objects, and advantages of the application to be more apparent. The drawings and the description of the exemplary embodiments of the present application are provided for explaining the present application and do not constitute an undue limitation on the present application. In the drawings:
FIG. 1 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 2 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 3 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 4 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 5 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 6 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
FIG. 7 is a flowchart illustrating a conversational robot intelligent response method according to an embodiment of the application;
FIG. 8 is a flow chart of a conversation robot intelligent response method according to an embodiment of the application;
fig. 9 is a flowchart illustrating a conversational robot intelligent response method according to an embodiment of the application.
Detailed Description
In order to make the technical solutions of the present application better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that embodiments of the application described herein may be used. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The intelligent response method of the conversation robot in the application comprises the following steps: acquiring a user request; returning at least one processing result similar to the user request; and receiving user feedback of the processing result, and returning a next processing result according to the user feedback. The expression of the user is extracted by using various models, and the understanding range is automatically expanded in a continuous machine learning mode, so that the conversation robot can more accurately understand various expressions of the user and is more fit with the expression habits of the user along with the lapse of time.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
As shown in fig. 1, the method includes steps S to S as follows:
step S102, obtaining a user request;
and acquiring a user request of the user in a conversation process with the conversation robot.
The robot is used for asking questions of the user and obtaining responses of the user to the questions.
Namely, the following steps are for further operations after the conversation robot acquires the user request.
It should be noted that, in the embodiment of the present application, the manner or the type of the user request is not limited as long as the condition for obtaining the user request can be satisfied.
Step S104, returning at least one processing result similar to the user request;
and returning a processing result similar to the user request to the user through the conversation robot.
The processing result returned by the conversation robot similar to the user request may include a plurality of processing results, and is not limited to returning only one processing result. If the dialog robot returns the processing result belonging to the result with higher confidence, only one processing result can be returned, namely the processing result can satisfy the current use of the user. If the dialog robot has uncertainty about the returned processing result and needs to be further clarified by the user to determine, a plurality of possible results may be returned and the user may be given a choice.
Similar to the user request refers to a situation similar to or consistent with the user's purpose, intent.
Specifically, when at least one processing result similar to the user request is returned, a processing result prompt is given to the user, and one or more processing results for the user request may be included in the processing result.
And step S106, receiving user feedback of the processing result, and returning the next processing result according to the user feedback.
And returning the next processing result by the conversation robot according to the user feedback by receiving the user feedback of the processing result. The user feedback includes at least: feedback of the accuracy or satisfaction of the user to the answer to the question. And when the next processing result is returned according to the user feedback, the next conversation state can be carried out or the current conversation state can be continued.
Specifically, the processing result is returned to the user for confirmation and clarification, and after the feedback of the user is obtained, the successfully understood dialog segment can be added into the corpus and the machine learning is continued.
From the above description, it can be seen that the following technical effects are achieved by the present application:
according to the intelligent response method and device for the conversation robot in the embodiment of the application, the purpose of receiving the user feedback of the processing result and returning the next processing result according to the user feedback is achieved by returning at least one processing result similar to the user request in a mode of acquiring the user request, so that the technical effects of intelligently understanding the user request and intelligently enhancing learning are achieved, and the technical problem that the conversation robot cannot accurately understand the intention of the user is solved.
According to the embodiment of the present application, as a preference in the embodiment, as shown in fig. 2, returning at least one processing result similar to the user request includes:
step S202, generating a candidate set expressed by the current standard according to the information on the user dialog;
and after receiving a request of a user, adding user dialogue information, and generating a candidate set expressed by the current standard according to the user dialogue information.
For example, is the question which stays well? And generating a current standard expression A1 to be used as a candidate set of the current standard expression for American good study, B2 to be used as a candidate set of English high study fee and C3 to be used as a candidate set of the current standard expression for Netherlands to watch windmills according to the information on the user dialog.
Step S204, calculating a candidate result in the candidate set according to a preset similarity model;
and judging and calculating a candidate result in the candidate set according to the preset similarity model.
The preset similarity model can be subject similarity calculation, pinyin similarity calculation, semantic similarity calculation or pronunciation similarity calculation and the like.
And step S206, fusing the candidate results, and outputting the candidate results as a processing result according to the sequence.
And fusing the candidate results in the candidate set according to the calculation results of different similarity calculation models, and outputting the fusion as a processing result according to the sequence.
Specifically, after query of a user is obtained, text information is added to a current user session, a candidate set expressed by all current standards is obtained, and then evaluation and sorting are performed on the candidates by using various methods based on word vectors, pinyin similarity and the like, so that one or more most similar results are obtained.
Various complex language scenes can be solved by the mixed use of a plurality of models. For example, the pinyin-based model can solve the problems of homophones and wrong characters and wrong voice recognition, and the vector-representation-based model can solve the problems of synonymous expression, new words and the like.
According to the embodiment of the present application, as a preferred feature in the embodiment, the user dialog context information includes: the domain information is used for judging which scene is processed by the dialog at present; purpose intention information for a user to have a conversation; dialog action information for questioning information and selecting contents.
Specifically, the above information in the dialog mainly includes, as domain information, what scene the dialog is currently in. Intention information for the purpose of conducting a conversation as a user. And the dialogue action is used for asking what information, selecting what content and the like.
In the embodiment of the application, by using the above of the dialog, the scope to be understood can be limited to a relatively small range, and at this time, compared with the general semantic understanding, more aggressive understanding strategies can be used, so that the success rate of the understanding is greatly enhanced.
According to the embodiment of the present application, as a preferred embodiment in the present application, as shown in fig. 3, after receiving user feedback on the processing result and returning the next processing result according to the user feedback, the method further includes:
and step S302, judging whether to continue executing the current conversation or to perform clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
Specifically, after the above information is supplemented, all possible candidates are obtained first, then the candidates are scored by using multiple text judgment models, finally the scores of the multiple models are fused and then sorted according to the fusion score from high to low, and the part meeting the threshold is returned to the current conversation. And judging whether the next action is to continue to execute the conversation or to carry out clarification interaction with the user according to the number of returned results in the current conversation.
According to the embodiment of the present application, as a preferred embodiment in the present embodiment, as shown in fig. 4, after receiving user feedback on the processing result and returning the processing result of the next time according to the user feedback, the method further includes:
step S402, if the user feedback is that the user request is correctly understood, the processing result is stored in a to-be-examined checking library so that the conversation robot can intelligently respond to the next request of the user through the understanding result in the to-be-examined checking library.
Specifically, after the feedback of the user is obtained, the matching segments (correct understanding) in the query are added into the to-be-examined library, and the options in the to-be-examined library also participate in the understanding of the next query of the user on the premise of punishment, so that the automatic continuous learning of the conversation robot is realized.
In addition, a dialog manager is allowed to manually check all candidates to be checked, and the penalty weight coefficient is adjusted to guide the understanding direction of the dialog robot.
It should be noted that, after the query of the user is successfully understood each time, similar segments in the query are supplemented into the to-be-examined kernel library, and as a next understanding, the content to be examined can also be one of the candidates for understanding. Meanwhile, in order to prevent the false recall, the option to be audited is added with a penalty weight. And if the conversation robot generates a false recall, a punishment mechanism is provided, so that the automatic continuous learning of the conversation robot is realized.
It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer-executable instructions and that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different than here.
According to an embodiment of the present application, there is also provided a dialogue robot intelligent response device for implementing the above method, as shown in fig. 5, the device includes: an obtaining module 10, configured to obtain a user request; a returning module 20, configured to return at least one processing result similar to the user request; and a receiving module 30, configured to receive user feedback on the processing result, and return a next processing result according to the user feedback.
The obtaining module 10 of the embodiment of the present application obtains the user request of the user in the conversation process with the conversation robot. The user request includes: the user queries the request. Namely, the following steps are for further operations after the dialogue robot acquires the user request.
It should be noted that, in the embodiment of the present application, the manner or the type of the user request is not limited as long as the condition for obtaining the user request can be satisfied.
The return module 20 of the embodiment of the present application returns a processing result similar to the user request to the user. And if the user request is a query request, the corresponding processing result is a query result.
The returned processing results similar to the user request may include multiple processing results, and are not limited to returning only one processing result. If the dialog robot returns the processing result belonging to the result with higher confidence, only one processing result can be returned, namely the processing result can satisfy the current use of the user. If the dialog robot has uncertainty about the returned processing result and needs to be further clarified by the user to determine, a plurality of possible results may be returned and the user may be given a choice.
Similar to the user request refers to a situation similar to or consistent with the user's purpose, intent.
Specifically, when at least one processing result similar to the user request is returned, a processing result prompt is given to the user, and one or more processing results for the user request may be included in the processing result.
In the receiving module 30 of the embodiment of the present application, by receiving the user feedback for the processing result, the conversation robot returns the next processing result according to the user feedback. The user feedback includes at least: feedback of the accuracy or satisfaction of the user to the answer to the question. And when a next processing result is returned according to the user feedback, the next conversation state can be carried out or the current conversation state can be continued.
Specifically, the processing result is returned to the user for confirmation and clarification, and after the feedback of the user is obtained, the successfully understood dialog segment can be added into the corpus and the machine learning is continued.
According to the embodiment of the present application, as a preferred option in the embodiment, as shown in fig. 6, the returning module is further configured to generate a candidate set expressed by the current standard according to the user dialog context information; calculating candidate results in the candidate set according to a preset similarity model; and fusing the candidate results, and outputting the candidate results as a processing result according to the sequence.
In the returning module 20 of the embodiment of the present application, after receiving a request from a user, information on a user dialog text is added, and a candidate set expressed by a current standard is generated according to the information on the user dialog text.
For example, is the question which stays well? And generating a current standard expression A1 to go to American to study well, B2 to go to English to study at high fee and C3 to go to the Netherlands to look at windmills as a candidate set of the current standard expression according to the user dialogue information.
In the returning module 20 of the embodiment of the present application, the candidate result in the candidate set is determined and calculated according to the preset similarity model.
The preset similarity model can be subject similarity calculation, pinyin similarity calculation, semantic similarity calculation or pronunciation similarity calculation and the like.
In the returning module 20 of the embodiment of the present application, the candidate results in the candidate set are fused according to the calculation results of the different similarity calculation models, and are output as the processing result according to the ranking.
Specifically, after query of a user is obtained, text information is added to a current user session, a candidate set expressed by all current standards is obtained, and then evaluation and sorting are performed on the candidates by using various methods based on word vectors, pinyin similarity and the like, so that one or more most similar results are obtained.
Various complex language scenes can be solved by mixed use of a plurality of models. For example, the pinyin-based model can solve the problems of homophones and wrong characters and wrong voice recognition, and the vector-representation-based model can solve the problems of synonymous expression, new words and the like.
According to the embodiment of the present application, as a preferred embodiment in the present application, as shown in fig. 7, the user dialog context information in the returning module 20 includes: the domain information is used for judging which scene is processed by the dialog at present; purpose intention information for a user to have a conversation; dialog action information for questioning information and selecting contents.
Specifically, the above information in the dialog mainly includes, as domain information, what scene the dialog is currently in. Intention information for the purpose of conducting a conversation as a user. And the dialogue action is used for asking what information, selecting what content and the like.
In the embodiment of the application, by using the above of the dialog, the scope to be understood can be limited to a relatively small range, and at this time, compared with the general semantic understanding, more aggressive understanding strategies can be used, so that the success rate of the understanding is greatly enhanced.
According to the embodiment of the present application, as a preference in the embodiment, as shown in fig. 6, the apparatus further includes: and the clarification interaction module 40 is configured to determine whether to continue to execute the current conversation or perform clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
Specifically, in the clarification interaction module 40 according to the embodiment of the present application, after the above information is supplemented, all possible candidates are obtained first, then the candidates are scored using a plurality of text determination models, and finally the scores of the plurality of models are fused and then sorted from high to low according to the fusion score, and the part satisfying the threshold is returned to the current conversation. And judging whether the next action is to continue to execute the conversation or to carry out clarification interaction with the user according to the number of returned results in the current conversation.
According to the embodiment of the present application, as a preference in the embodiment, as shown in fig. 7, the apparatus further includes: and the enhancing module 50 is configured to, if the user feeds back that the user request is correctly understood, store the processing result in the to-be-examined checking library so that the conversation robot intelligently responds to the next request of the user through the understanding result in the to-be-examined checking library.
Specifically, in the enhancement module 50 of the embodiment of the present application, after the feedback of the user is obtained, the matching segment (correct understanding) in the query is added to the to-be-examined kernel library, and the options in the to-be-examined kernel library also participate in the understanding of the next query of the user on the premise of punishment, so as to implement the automatic continuous learning of the conversation robot.
In addition, a dialog manager is allowed to manually check all candidates to be checked, and the penalty weight coefficient is adjusted to guide the understanding direction of the dialog robot.
It should be noted that, after the query of the user is successfully understood each time, similar segments in the query are supplemented into the to-be-examined kernel library, and as a next understanding, the content to be examined can also be one of the candidates for understanding. Meanwhile, in order to prevent the false recall, the option to be audited is added with a penalty weight. And if the conversation robot generates a false recall, a punishment mechanism is provided, so that the automatic continuous learning of the conversation robot is realized.
The principle of implementation of the present application is shown in fig. 8, and in particular, it is increasingly difficult for a conversation robot to want to accurately understand the meaning of a user as described in the background. The endless new words, strange grammar, etc. are caused by the diversity of the language itself. Homophonic and different characters, background noise during speech recognition, and the like are caused by the diversity of input methods. In order to enable the conversation robot to more accurately understand various expressions of the user, the implementation manner of the application mainly includes: a dialogue management module and an intelligent response module.
Specifically, the functions in the intelligent answering module include: candidate acquisition and candidate sorting. The dialogue management module comprises: the management module comprises an upper management module and a state management module.
After query of a user is obtained in the dialogue management module, all candidate sets expressed by the current standard are obtained in the intelligent response module according to the field, intention, dialogue action and other above information in the dialogue, then evaluation and sequencing are carried out on the candidates by using various methods based on word vectors, pinyin similarity and the like, one or more most similar results are obtained, and the results are returned to the user for confirmation and clarification. Furthermore, after the feedback of the user is obtained, the matching segment in the query is added into a to-be-examined check library of the state management module, and options in the to-be-examined check library also participate in the understanding of the next user query on the premise of punishment, so that the robot can automatically and continuously learn. And meanwhile, a dialog manager is allowed to manually check all candidates to be checked, and the punished weight coefficient is adjusted to guide the understanding direction of the robot.
The above information in the dialog in the above management module mainly includes: fields, intents, dialog actions, etc. By using the above of the dialog, the scope to be understood can be limited to a relatively small range, and at the moment, compared with the general semantic understanding, more aggressive understanding strategies can be used, so that the success rate of understanding is greatly enhanced.
As shown in fig. 9, after the dialog management module supplements the above information, the dialog management module is handed to the intelligent response module for processing. All possible candidates are obtained at the intelligent response module, the candidates are scored by using multiple preset text judgment models, finally, the scores of the multiple models are fused and then are sorted according to the fusion score from high to low, and the part meeting the threshold value is returned to the dialogue management module. And the conversation management module judges whether the next action is to continue to execute the conversation or to carry out clarification interaction with the user according to the number of the returned results.
Preferably, various complex language scenarios can be addressed through the hybrid use of multiple models. For example, the pinyin-based model can solve the problems of homophones and wrong characters and wrong voice recognition, and the vector-representation-based model can solve the problems of synonymous expression, new words and the like.
As shown in fig. 9, which is an implementation principle of intelligent reinforcement learning, in the present application, after a query of a user is successfully understood each time, similar segments in the query are supplemented into a to-be-reviewed kernel library, and the content to be reviewed can be used as one of candidates for understanding in the next understanding. Meanwhile, in order to prevent the false recall, the option to be audited is added with the weight of punishment.
As shown in fig. 9, during the learning process, some candidates cannot be recalled temporarily according to the initial candidate set, but if a query segment that has been successfully understood is added to the set to be audited, the intelligent comprehension range of the dialog robot at the next time can be continuously expanded by the candidates to be audited. In addition, the manager of the conversation manually audits the candidate to be audited, if the audit is successful, the candidate is added into the initial candidate set, if the audit is failed, the candidate is used as a negative sample to prevent the re-understanding error, and if the audit is successful, the quick learning of the conversation robot can be realized.
In another embodiment of the present application, there is provided a device for intelligent answering of a dialogue robot, the device including a processor, a display, a memory, a network interface, and a bus system, wherein the processor, the display, the memory, and the network interface are connected to each other through the bus system. The memory is used for storing instructions, and the processor is used for executing the instructions stored by the memory, and when the instructions are executed, the processor executes the intelligent response method of the dialogue robot through the network interface.
In yet another embodiment of the present application, a computer-readable medium is provided for storing a computer program comprising instructions for performing a conversational robot intelligent response method.
It will be apparent to those skilled in the art that the modules or steps of the present application described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and they may alternatively be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, or fabricated separately as individual integrated circuit modules, or fabricated as a single integrated circuit module from multiple modules or steps. Thus, the present application is not limited to any specific combination of hardware and software.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (6)
1. A conversation robot intelligent response method is characterized by comprising the following steps:
acquiring a user request;
returning at least one processing result similar to the user request; and
receiving user feedback of the processing result, and returning a next processing result according to the user feedback;
if the user feedback is that the user request is correctly understood, after the feedback of the user is obtained, the matched segment is added into a to-be-examined check library;
adding a punished weight coefficient to the option in the to-be-examined kernel library, and then executing the next processing of the user request;
manually checking all the options to be checked in the check library to be checked, and adjusting the punished weight coefficient;
returning at least one processing result similar to the user request comprises:
generating a candidate set expressed by the current standard according to the information of the user dialogue;
calculating candidate results in the candidate set according to a plurality of different preset similarity models;
and fusing the candidate results, and outputting the candidate results as processing results according to the sequence.
2. The intelligent answering method for dialogue robot according to claim 1, wherein the user dialogue above information comprises:
the domain information is used for judging which scene is processed by the dialog at present;
purpose intention information for a user to have a conversation;
dialog action information for questioning information and selecting contents.
3. The intelligent answering method for dialogue robots, as recited in claim 1, wherein the method receives user feedback on the processing result, and further comprises the following steps after returning the next processing result according to the user feedback:
and judging whether to continue to execute the current conversation or to carry out clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
4. A dialogue robot intelligent answering device is characterized by comprising:
the acquisition module is used for acquiring a user request;
a return module for returning at least one processing result similar to the user request; and
the receiving module is used for receiving user feedback of the processing result and returning a next processing result according to the user feedback;
the enhancement module is used for adding the anastomotic segment into a to-be-examined check library after the feedback of the user is obtained if the feedback of the user is correct to understand the user request;
adding a punished weight coefficient to the option in the to-be-examined kernel library, and then executing the next processing of the user request;
manually checking all the options to be checked in the check library to be checked, and adjusting the punished weight coefficient;
the return module is also configured to,
generating a candidate set expressed by the current standard according to the user dialogue context information;
calculating candidate results in the candidate set according to a plurality of different preset similarity models; and fusing the candidate results, and outputting the candidate results as processing results according to the sequence.
5. The intelligent answering device of dialogue robot of claim 4, wherein the user dialogue context information in the return module comprises:
domain information for judging which scene the dialog is currently processing;
purpose intention information for a user to have a conversation;
dialog action information for questioning information and selecting contents.
6. The intelligent answering device of dialogue robot of claim 4, further comprising: and the clarification interaction module is used for judging whether to continue to execute the current conversation or to carry out clarification interaction with the user according to the number of the processing results fed back by the user in the current conversation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910375687.XA CN110222161B (en) | 2019-05-07 | 2019-05-07 | Intelligent response method and device for conversation robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910375687.XA CN110222161B (en) | 2019-05-07 | 2019-05-07 | Intelligent response method and device for conversation robot |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110222161A CN110222161A (en) | 2019-09-10 |
| CN110222161B true CN110222161B (en) | 2022-10-14 |
Family
ID=67820626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910375687.XA Active CN110222161B (en) | 2019-05-07 | 2019-05-07 | Intelligent response method and device for conversation robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110222161B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111611364B (en) * | 2020-05-15 | 2023-08-15 | 北京百度网讯科技有限公司 | Intelligent response method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2557510A1 (en) * | 2011-08-12 | 2013-02-13 | Accenture Global Services Limited | Context and process based search ranking |
| CN107423440A (en) * | 2017-08-04 | 2017-12-01 | 逸途（北京）科技有限公司 | A kind of question and answer context switching based on sentiment analysis is with strengthening system of selection |
| CN108763355A (en) * | 2018-05-16 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of intelligent robot interaction data processing system and method based on user |
| CN109002477A (en) * | 2018-06-19 | 2018-12-14 | 深圳市元征科技股份有限公司 | Information processing method, device, terminal and medium |
| CN109543006A (en) * | 2018-10-12 | 2019-03-29 | 北京来也网络科技有限公司 | Method of quality control and device for corpus processing |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9824083B2 (en) * | 2015-07-07 | 2017-11-21 | Rima Ghannam | System for natural language understanding |
| CN105159996B (en) * | 2015-09-07 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Depth question and answer service providing method based on artificial intelligence and device |
| CN106503156B (en) * | 2016-10-24 | 2019-09-03 | 北京百度网讯科技有限公司 | Man-machine interaction method and device based on artificial intelligence |
| CN107291867B (en) * | 2017-06-13 | 2021-07-20 | 北京百度网讯科技有限公司 | Dialog processing method, device and equipment based on artificial intelligence and computer readable storage medium |
| JP6857581B2 (en) * | 2017-09-13 | 2021-04-14 | 株式会社日立製作所 | Growth interactive device |
| CN108170792B (en) * | 2017-12-27 | 2021-12-28 | 北京百度网讯科技有限公司 | Question and answer guiding method and device based on artificial intelligence and computer equipment |
| CN108256009B (en) * | 2018-01-03 | 2022-02-15 | 国网江苏省电力有限公司电力科学研究院 | Method for improving answer accuracy of electric intelligent answer robot |
-
2019
- 2019-05-07 CN CN201910375687.XA patent/CN110222161B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2557510A1 (en) * | 2011-08-12 | 2013-02-13 | Accenture Global Services Limited | Context and process based search ranking |
| CN107423440A (en) * | 2017-08-04 | 2017-12-01 | 逸途（北京）科技有限公司 | A kind of question and answer context switching based on sentiment analysis is with strengthening system of selection |
| CN108763355A (en) * | 2018-05-16 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of intelligent robot interaction data processing system and method based on user |
| CN109002477A (en) * | 2018-06-19 | 2018-12-14 | 深圳市元征科技股份有限公司 | Information processing method, device, terminal and medium |
| CN109543006A (en) * | 2018-10-12 | 2019-03-29 | 北京来也网络科技有限公司 | Method of quality control and device for corpus processing |
Non-Patent Citations (2)
| Title |
|---|
| A Dialogue-Based Approach Enhanced with Situation Awareness and Reinforcement Learning for Ubiquitous Access to Linked Data;Giuseppe D"Aniello，等;《2014 International Conference on Intelligent Networking and Collaborative Systems》;20150312;第249-256页 * |
| 基于文本相似度的智能查号引擎研究;乔世权等;《河北科技大学学报》;20180629(第03期);第96-102页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110222161A (en) | 2019-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11322153B2 (en) | Conversation interaction method, apparatus and computer readable storage medium | |
| US10453454B2 (en) | Dialog system with self-learning natural language understanding | |
| CN108509619B (en) | Voice interaction method and device | |
| CN111177359A (en) | Multi-turn dialogue method and device | |
| CN111402894B (en) | Speech recognition method and electronic equipment | |
| CN111191450A (en) | Corpus cleaning method, corpus entry device and computer-readable storage medium | |
| CN111858854A (en) | Question-answer matching method based on historical dialogue information and related device | |
| CN111259124A (en) | Dialogue management method, device, system and storage medium | |
| CN112084317A (en) | Method and apparatus for pre-training a language model | |
| CN116821290A (en) | Multitasking dialogue-oriented large language model training method and interaction method | |
| CN112417107A (en) | Information processing method and device | |
| CN112199486A (en) | Task type multi-turn conversation method and system for office scene | |
| CN109165286A (en) | Automatic question-answering method, device and computer readable storage medium | |
| CN111402864A (en) | Voice processing method and electronic equipment | |
| CN110222161B (en) | Intelligent response method and device for conversation robot | |
| CN112182189B (en) | Dialogue processing method and device, electronic equipment and storage medium | |
| CN113591463A (en) | Intention recognition method and device, electronic equipment and storage medium | |
| CN107886940A (en) | Voiced translation processing method and processing device | |
| CN109726002B (en) | Processing flow adjusting method and device | |
| CN109002498B (en) | Man-machine conversation method, device, equipment and storage medium | |
| CN114372476B (en) | Semantic truncation detection method, device, equipment and computer readable storage medium | |
| CN114860910A (en) | Intelligent dialogue method and system | |
| CN109273004A (en) | Predictive audio recognition method and device based on big data | |
| CN114202363A (en) | Artificial intelligence based call method, device, computer equipment and medium | |
| CN113470625A (en) | Voice conversation processing method, device, 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |