CN116681060A - Dialogue data processing method, man-machine interaction method, equipment and storage medium - Google Patents
Dialogue data processing method, man-machine interaction method, equipment and storage medium Download PDFInfo
- Publication number
- CN116681060A CN116681060A CN202310559957.9A CN202310559957A CN116681060A CN 116681060 A CN116681060 A CN 116681060A CN 202310559957 A CN202310559957 A CN 202310559957A CN 116681060 A CN116681060 A CN 116681060A
- Authority
- CN
- China
- Prior art keywords
- dialogue
- data
- dialogue data
- processed
- training
- 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
Abstract
The embodiment of the invention provides a dialogue data processing method, a man-machine interaction method, equipment and a storage medium, wherein the method comprises the following steps: the processing device obtains statistical features of the dialogue data to be processed containing at least one round of questions and answers, the statistical features are irrelevant to dialogue contents of the dialogue data to be processed, and further, whether the dialogue data to be processed has question and answer mismatch or not can be determined according to the statistical features. In the above manner, since the statistical feature of the dialogue data to be processed is a feature which is irrelevant to the dialogue content and has correlation with whether the dialogue data to be processed has question-answer mismatch, the dialogue data of different contents generated in different scenes can be detected by using the method to detect whether the question-answer mismatch exists, thereby improving the generalization of the question-answer mismatch detection.
Description
Technical Field
The present invention relates to the field of artificial intelligence technologies, and in particular, to a dialogue data processing method, a man-machine interaction method, a device, and a storage medium.
Background
Man-machine conversations have been enabled using artificial intelligence techniques and may also have application in various scenarios. For example, a user may be engaged in a conversation using a hardware interaction device integrated with a conversation system. The interaction device may be a service robot, an intelligent sound box, a mobile terminal device, etc. For example, when the service robot is in particular a guidance robot deployed in a mall, restaurant, the service robot may guide the user by having multiple rounds of conversations with the user. For example, a dialogue system deployed in the cloud can be used for conducting dialogue with a user, such as online intelligent customer service, and the like, so as to meet the query requirement of the user on related information.
In different scenarios, the content of man-machine conversations is quite rich. And in any scene, a frequently occurring condition in the human-computer interaction process is question-answer mismatch, namely, question-answer questions, which can seriously affect the user experience. Therefore, how to accurately detect the mismatching of questions and answers in the human-computer conversation process under different scenes becomes a problem to be solved urgently.
Disclosure of Invention
In view of this, the embodiments of the present invention provide a dialogue data processing method, a man-machine interaction method, a device and a storage medium, which are used for accurately detecting whether question-answer mismatch exists in dialogue data of different scenes.
In a first aspect, an embodiment of the present invention provides a method for processing dialogue data, including:
acquiring statistical features of dialogue data to be processed containing at least one round of questions and answers, wherein the statistical features are irrelevant to dialogue content of the dialogue data to be processed;
and determining whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics.
In a second aspect, an embodiment of the present invention provides a human-computer interaction method, including:
displaying a prompt message, wherein the prompt message is generated when the number of abnormal dialogue data is larger than the preset number, the abnormal dialogue data comprises dialogue data with question-answer mismatch, the dialogue data is generated by a user and the dialogue system in a man-machine dialogue process, and whether the dialogue data has question-answer mismatch is determined by the statistical characteristics of the dialogue data;
Responding to the processing operation of the prompt message.
In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory, where the memory is configured to store one or more computer instructions, where the one or more computer instructions, when executed by the processor, implement the method for processing dialogue data in the first aspect, or the method for man-machine interaction in the second aspect. The electronic device may also include a communication interface for communicating with other devices or communication systems.
In a fourth aspect, embodiments of the present invention provide a non-transitory machine-readable storage medium having executable code stored thereon, which when executed by a processor of an electronic device, causes the processor to implement at least the method for processing dialogue data as in the first aspect or the method for human-computer interaction as in the second aspect.
According to the dialogue data processing method provided by the embodiment of the invention, the processing equipment acquires the statistical characteristics of the dialogue data to be processed comprising at least one round of questions and answers, the statistical characteristics are irrelevant to dialogue contents of the dialogue data to be processed, and further, whether the questions and answers of the dialogue data to be processed are not matched can be determined according to the statistical characteristics. In the above manner, since the statistical characteristics of the dialogue data to be processed are irrelevant to the dialogue content, the dialogue data of different contents generated in different scenes can be detected by using the method to detect whether the question and answer mismatch exists, so that the generalization of the question and answer mismatch detection is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for processing dialogue data according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an operation interface according to an embodiment of the present invention;
FIG. 3 is a flowchart of a test model training method according to an embodiment of the present invention;
FIG. 4 is a flowchart of another method for processing dialogue data according to an embodiment of the present invention;
FIG. 5 is a flowchart of another test model training method according to an embodiment of the present invention;
FIG. 6 is a flowchart of another test model training method according to an embodiment of the present invention;
fig. 7 is an application schematic diagram of a dialogue data processing method according to an embodiment of the present invention;
FIG. 8 is a flowchart of a man-machine interaction method according to an embodiment of the present invention;
FIG. 9 is a flowchart of another man-machine interaction method according to an embodiment of the present invention;
FIG. 10 is a schematic diagram of a dialogue data processing device according to an embodiment of the present invention;
fig. 11 is a schematic structural diagram of a man-machine interaction device according to an embodiment of the present invention;
fig. 12 is a schematic structural diagram of an electronic device according to an embodiment of the present invention;
fig. 13 is a schematic structural diagram of another electronic device according to an embodiment of the present invention;
fig. 14 is a schematic structural diagram of another man-machine interaction device according to an embodiment of the present invention;
fig. 15 is a schematic structural diagram of still another electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The terminology used in the embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise, the "plurality" generally includes at least two, but does not exclude the case of at least one.
It should be understood that the term "and/or" as used herein is merely one relationship describing the association of the associated objects, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In addition, the character "/" herein generally indicates that the front and rear associated objects are an "or" relationship.
The words "if", as used herein, may be interpreted as "at … …" or "at … …" or "in response to a determination" or "in response to an identification", depending on the context. Similarly, the phrase "if determined" or "if identified (stated condition or event)" may be interpreted as "when determined" or "in response to a determination" or "when identified (stated condition or event)" or "in response to an identification (stated condition or event), depending on the context.
It should be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a product or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such product or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a commodity or system comprising such elements.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present invention are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide corresponding operation entries for the user to select authorization or rejection.
It should be noted that, similarly to the description in the background art, an intelligent robot such as a service robot, a greeting robot, a self-moving vending robot, or the like may be integrated with a dialogue system. In addition, smart terminals such as mobile terminals, smart home appliances, smart wearable devices, etc. may also be integrated with dialog systems. Each of the above hardware devices may be considered as an interactive device. In addition, the dialogue system can be deployed at the cloud to provide services such as online shopping, online consultation and the like for users. In general, any dialog system deployed on a hardware device or cloud that supports human-machine dialog with a user, either in voice or text, may generate dialog data with the user. And the following embodiments of the present invention can determine task-type conversations generated by users and conversational systems during a man-machine conversation as to-be-processed conversational data.
Before describing the dialogue data processing method and the man-machine interaction method provided in the following embodiments of the present invention in detail, the related concepts related to the following embodiments may be explained:
task-type conversations: at least one round of dialogue generated by the dialogue system aiming at a certain requirement of a user is at least one round of question and answer. The dialog system may determine the user's intent by way of understanding, clarification, etc., and then complete the user's needs by way of a reply.
Question-answer mismatch: the dialogue system does not directly give out an answer when answering the questions given by the user, or does not answer the actual meaning of the questions, and the questions and answers are not matched, so that the questions and answers are actually questions and questions.
For example, in an online service scenario such as an online banking consultation scenario, the following dialogue data may be generated between a user and a service robot set in a banking lobby:
service robot: please ask what you need to transact?
The user: i want to change some dollars.
Service robot: please ask you if you carry an identification card and a bank card?
The user: both the identity card and the bank card are carried.
Service robot: please place the identification card in the induction area, select the exchange of the foreign exchange and take the number.
In the conversation, the service robot finally and successfully guides the user to finish the number taking of the foreign exchange conversation, so that no question-answer mismatch exists in the conversation data.
For example, in an online service scenario, such as an online government information consulting scenario, the following dialogue data may be generated between a user and an online intelligent customer service:
on-line intelligent customer service: you can provide information consultation in human social security for you, you can say pension adjustment, payroll, medical insurance, childbirth benefits, unemployment insurance, etc.
The user: and (5) manual service.
On-line intelligent customer service: you can say that you want the problem of consulting, if you can't answer, will transfer the manual customer service for you.
The user: the first few days have the problem of producing a body composition for calling.
On-line intelligent customer service: please ask you if you are childbirth or abortion.
The user: no branch name was written the last time a fertility post was applied.
On-line intelligent customer service: bad meaning, i don't hear clearly, you say.
The user: and (5) manual service.
In the dialogue, the online intelligent customer service does not understand the user intention finally, so that the dialogue data have question-answer mismatch.
In practice, the user experience is seriously affected by the question-answer mismatch, and the method mentioned in the following embodiments of the present invention can be used to detect whether there is a question-answer mismatch in the dialogue data.
Some embodiments of the invention will now be described in detail with reference to the accompanying drawings, based on the foregoing description. In the case where there is no conflict between the embodiments, the following embodiments and features in the embodiments may be combined with each other. In addition, the sequence of steps in the method embodiments described below is only an example and is not strictly limited.
Fig. 1 is a flowchart of a method for processing dialogue data according to an embodiment of the present invention. The dialogue data processing method provided by the embodiment of the invention can be executed by processing equipment. The processing device may be either a physical device or a virtual device. As shown in fig. 1, the method may include the steps of:
s101, acquiring statistical characteristics of the dialogue data to be processed containing at least one round of questions and answers, wherein the statistical characteristics are irrelevant to dialogue contents of the dialogue data to be processed.
In the man-machine conversation process of the user and the conversation system, the conversation data generated by the user and the conversation system are the conversation data to be processed. And in response to the initiation of the human-machine conversation, the conversation system may record the user generated pending conversation data itself. Wherein, the section of dialogue data to be processed can comprise at least one round of questions and answers generated by a user and the dialogue system in a dialogue process. Such as the online-offline banking consultation scenario described above or the dialogue data generated in the online government information consultation scenario.
Alternatively, both the user and the dialog system may actively initiate a human-machine dialog. And under the condition that the user actively triggers the dialogue opening operation, the online banking consultation scene is received, and optionally, the user can actively trigger the dialogue opening operation on an operation interface provided by the service robot so as to actively open the man-machine dialogue. The dialog data to be processed generated at this time may generally be presented in speech form or in text form. And after receiving the on-line government information consultation scene, the user can actively dial a government information consultation telephone to actively start man-machine conversation. At this time, the generated dialog data to be processed may generally be presented in the form of speech. The user can also actively trigger a dialogue opening operation on an operation interface provided for the government website so as to actively open the man-machine dialogue. At this time, the dialog data to be processed may generally be presented in the form of text data.
In the case that the dialogue system actively triggers a dialogue opening operation, optionally, in an online customer service return visit scene, a return visit telephone can be actively dialed to the user by the dialogue system to open a man-machine dialogue, and the dialogue data to be processed generated in such a scene can be expressed in a voice form.
Optionally, for the acquisition of statistical features, the dialog system may record the generation time of each sentence in the dialog data to be processed, while recording the sentence. Based on this, statistics may be performed on the session data to be processed by the session system or other device or system based on this generation time to obtain statistical features of the session data to be processed. The processing device may obtain this statistical feature from the dialog system.
Alternatively, the statistical features of the dialog data to be processed may comprise features of the dialog data that are not related to the dialog content. Optionally, specifically, at least one of the following may be included: the number of rounds of conversations, the number of times the system has not given an answer, whether the user generated conversations are too long, the number of times the user has not generated a conversation within a preset time period, the number of times the user has turned to a manual conversation, whether the user generated conversations contain negative emotions, the number of times the conversation has been referred to, whether the last round of conversations has a situation that the system has not given an answer and the user hangs up, whether the user has not generated a conversation within a preset time period when the last round of conversations, and the like. From the statistical features given above, the statistical features are actually numerical type features.
The above statistics can also be explained as follows:
the absence of an answer from the dialog system may be manifested as a response from the dialog system that "bad meaning, i am not listening, you please say" or "bad meaning, please speak again" etc. indicate that the user's intent is not understood.
The inclusion of a negative emotion in a user-generated dialog may be manifested as the inclusion of a preset emotional word in the user-generated dialog, and the emotional word may represent a negative emotion.
If the user does not generate a dialogue within the preset time length, namely the user generates a dialogue timeout, the number of times that the user does not generate the dialogue within the preset time length is the number of times that the user generates the dialogue timeout. Similarly, whether the user does not generate the dialogue within the preset duration in the last round of dialogue is the time-out of the last round of dialogue of the user.
The statistics of whether the dialog generated by the user is too long, whether the dialog generated by the user contains negative emotion, and whether the user does not generate the dialog in the last dialog time can also be represented by different identifiers, for example, the situation can be represented by "1", and the situation can not be represented by "0".
The statistical characteristics of any of the dialog data to be processed may be schematically shown in the following table:
the contents in the table above represent: the dialogue turn of the dialogue data to be processed is 3 times, 1 time of the manual dialogue is changed, and 1 time of the manual dialogue is indicated in the dialogue.
And it should be noted that, whether each feature in the statistical features is not matched with question and answer in the dialogue data is of specific relevance. Alternatively, the degree of correlation is specifically measured using any one of pearson correlation coefficient, spearman (Spearman Rank Correlation Coefficient, src for short) rank correlation coefficient, kendel (Kendall) correlation coefficient, and the like.
Specifically, the statistics of "number of rounds of dialogue", "number of times the dialogue system does not give an answer", "whether the dialogue generated by the user is too long" and "number of times the user does not generate a dialogue within a preset duration" are highly positively correlated with the presence of question-answer mismatch with the dialogue data. I.e., the more rounds of dialog, the longer the dialog generated by the user, the more times the user has not generated a dialog within a preset duration, indicating a higher likelihood that the dialog data has a question-answer mismatch.
The statistical characteristics of whether the dialogue system does not give an answer and the user hangs up in the last round of dialogue, whether the user does not generate dialogue in the preset duration in the last round of dialogue, whether negative emotion is contained in the dialogue generated by the user and the number of times of referring in the dialogue are positively correlated with question-answer mismatch of dialogue data.
The statistical feature of 'the number of times of turning to the manual dialogue' is inversely related to the question-answer mismatch of the dialogue data, namely, the lower the number of times of turning to the manual dialogue, the lower the possibility that the question-answer mismatch of the dialogue data exists.
S102, determining whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics.
The processing device may then determine whether a question-answer mismatch exists for the pending dialog directly from the statistical characteristics of the pending dialog data.
Optionally, for the statistical features of the numerical value types, a weight coefficient may be set for each statistical feature according to the correlation of the question-answer mismatch between different statistical features and the dialogue data, and the statistical features and the corresponding weight systems thereof may be weighted and summed, and the summed result may be compared with a preset numerical value to determine whether the question-answer mismatch exists in the dialogue data to be processed.
In this embodiment, the processing device obtains a statistical feature of the to-be-processed dialogue data including at least one round of questions and answers, where the statistical feature is irrelevant to dialogue content of the to-be-processed dialogue data, and further determines whether there is a question and answer mismatch in the to-be-processed dialogue data according to the statistical feature. In the above manner, since the statistical feature of the dialogue data to be processed is a feature unrelated to the dialogue content, the dialogue data of different contents generated in different scenes can be detected by using the method to detect whether the question and answer mismatch exists, so that the generalization of the question and answer mismatch detection is improved. In addition, the statistical features of the numerical type contain smaller amounts of data than the features of the feature vector type, and therefore, the use of the above method can also reduce the amount of computation in detecting whether there is a question-answer mismatch.
It is easy to understand that the more the number of dialogue data having question-answer mismatch indicates that the performance of the dialogue system is worse, so the processing device may detect all dialogue data generated in the history period as dialogue data to be processed in the manner of the embodiment shown in fig. 1, and may know the number of abnormal dialogue data having question-answer mismatch in the history period according to the detection result. Alternatively, the number may also be displayed on a statistical interface provided by the processing device. If the number of the abnormal dialogue data is larger than the preset number, a prompt message can be displayed on the statistical interface. At this time, the operator of the dialogue system can choose whether to optimize the dialogue system according to the prompt message displayed on the statistical interface.
Optionally, a change curve of the number of session data in the historical time period may be displayed on the statistical interface, and other information related to the abnormal session data may be displayed on the statistical interface, such as the total number of session data in the current historical time period, the duty ratio of the abnormal session data in the current historical time period, the change rate of the number of abnormal session data in the current historical time period compared with the number of abnormal session data in the previous historical time period, and so on. Alternatively, the historical time end may be one week, one month, or several months, etc.
The statistical interface provided by the processing device may be as shown in fig. 2.
For determining whether there is a question-answer mismatch in the dialog data to be processed based on statistical features, this can be achieved by means of a detection model, in addition to the weighted summation approach in the embodiment shown in fig. 1. Specifically, the statistical characteristics of the dialogue data to be processed can be input into the detection model, so that whether the dialogue data to be processed has question-answer mismatch or not can be directly output by the detection model. The detection model is essentially a classification model, and optionally, the detection model may be any model that can implement a classification function, such as various Neural Networks (NN), logistic regression (Logistic regression) models, support vector machines (Support Vector Machine, SVM), and so on.
Alternatively, an initial model may be trained using supervised training to obtain the detection model used in the various embodiments described above. Alternatively, multiple initial models may be trained to obtain the detection model described above. Fig. 3 is a flowchart of a test model training method according to an embodiment of the present invention. The dialogue data processing method provided by the embodiment of the invention can be executed by training equipment. Alternatively, the training device may be the processing device mentioned in the above embodiment, or may be another device other than the processing device. As shown in fig. 3, the following steps may be included:
S201, acquiring statistical characteristics of training dialogue data containing at least one round of questions and answers, and reflecting whether the training dialogue data has a question and answer unmatched labeling result.
The training device may first obtain training session data comprising at least one round of questions and answers. Alternatively, the training session data may be collected over a network, or may be session data generated by the user over a historical period of time. Meanwhile, the training device can also obtain a labeling result of the training dialogue data, wherein the labeling result is used for reflecting whether the training dialogue data has question-answer mismatch or not, and optionally, the labeling result can be labeled manually.
S202, training the initial models with the statistical features of training dialogue data as training data and the labeling results as supervision information to obtain alternative models with different model parameters.
The training device may then also acquire a preset number of initial models. Then, in an alternative manner, the preset number of initial models may have the same model structure and different initial model parameters, and the statistical features of the training session data in step S201 may be used as training data, and the labeling results of the training session data may be used as supervision information to respectively train the preset number of initial models. Since the initial model parameters of the initial model are different, alternative models with different model parameters can be obtained even using the same training data.
Alternatively, the training session data in step S201 may be divided by a preset number. And then, respectively taking the statistical characteristics of each training dialogue data as training data of a preset number of initial models, and respectively training the preset number of initial models by taking the labeling results of the training dialogue data as supervision information. In this manner, the predetermined number of initial models may have the same model structure and the same initial model parameters. Because the training dialogue data used for training the initial model is different, model parameters of the obtained alternative model are different.
Alternatively, the loss calculation may be performed according to the following formula, and training of the initial model may be achieved by adopting a gradient descent manner:
wherein p is the confidence of the detection result which is output by the detection model and indicates that the training dialogue sample has question-answer mismatch, sigma is a loss function used for calculating a loss value between the detection result and the standard result, x=w1x1+w2x2+ + wkxk+b, X1, X2..
S203, determining model parameters of the detection model according to the model parameters of the alternative model to obtain the detection model.
Finally, the training device may integrate model parameters of each of the preset number of candidate models to determine model parameters of the detection model. Alternatively, the model parameters of each of the preset number of candidate models may be averaged or weighted and summed, and the calculation result may be used as the model parameter of the detection model, thereby obtaining the detection model. The detection model has the same model structure as the initial model and the alternative model.
In this embodiment, the detection accuracy of the detection model is improved by comprehensively considering model parameters of a plurality of candidate models, that is, a parameter integration mode. Specifically, the training device trains a plurality of alternative models, and then obtains model parameters of the detection model by comprehensively considering respective model parameters of the alternative models, so that the accuracy of the model parameters of the detection model is improved, namely the detection accuracy of the detection model is improved.
In order to improve the accuracy of detection of question-answer mismatch, alternatively, after a preset number of alternative models are obtained according to the embodiment shown in fig. 3, an integrated learning (Ensemble Learning) mechanism may be combined, that is, multiple detection results may be comprehensively considered to obtain more accurate detection results.
Specifically, the detection results of the dialogue data to be processed and the confidence coefficient of the detection results may be output by a preset number of candidate models, and then the detection results are synthesized, for example, the confidence coefficient is averaged or weighted and summed, so as to finally obtain the detection result of the dialogue data to be processed.
Based on the embodiment shown in fig. 1, fig. 4 is a flowchart of another processing method of dialogue data according to the embodiment of the present invention in order to further improve accuracy of question-answer mismatch detection. As shown in fig. 4, the following steps may be included:
s301, acquiring statistical characteristics of the dialogue data to be processed, which contain at least one round of dialogue, wherein the statistical characteristics are irrelevant to dialogue contents of the dialogue data to be processed.
The specific implementation process of step S301 may refer to the specific description of the relevant steps in the embodiment shown in fig. 1, which is not repeated here.
S302, feature combination is carried out on the statistical features to obtain combined features of the dialogue data to be processed.
S303, determining whether the dialogue data to be processed has question-answer mismatch according to the combination characteristics of the dialogue data to be processed.
The processing device may perform feature combination, i.e. feature interleaving, on the statistical features of the dialogue data to be processed to obtain combined features of the dialogue data to be processed. Wherein the following can also be understood for a combination of features: any of the features of at least two of the various statistical features may be combined as set forth in the above table. It should be noted that the two features may be the same or different. And as with the statistical features, the combined features are also independent of the dialog content of the dialog data to be processed.
Alternatively, feature combinations may be performed in the form of binomials or higher order polynomials. For example, for statistical features [ a, b, c]The combined features obtained by performing binomial feature crossing on the features can comprise [ a, b, c, ab, bc, ac, a 2 ,b 2 ,c 2 ]The method comprises the steps of carrying out a first treatment on the surface of the The combined features after the three-term feature crossing can comprise [ a, b, c, ab, bc, ac, a 3 ,b 3 ,c 3 ，a 2 b,ab 2 ,a 2 c,ac 2 ,b 2 c,bc 2 ,abc]。
Wherein a, b, c may represent any three of the plurality of statistical features mentioned in the embodiment shown in fig. 1. In the numerical sense, a, b, c are expressed as numerical values, ab, bc, a 2 ,c 2 And the like is the product of the numerical values. In a physical sense, for example, a represents whether the dialog generated by the user is too long, b represents whether the dialog generated by the user contains negative emotion, c represents the number of times of converting the manual dialog, ab represents whether the dialog generated by the user is too long, meanwhile, whether the dialog generated by the user contains negative emotion, and ac represents whether the dialog generated by the user is too long, and meanwhile, whether the dialog generated by the user generates the manual dialog and the number of times of converting the manual dialog.
Compared with the original statistical features, the number of the statistical features can be increased through feature combination, and the expanded combined features can describe the features of the dialogue data to be processed more comprehensively, so that whether question-answer mismatch exists in the dialogue data to be processed can be determined more accurately.
In this embodiment, by combining features of statistical features of the dialogue data to be processed, a combination feature with more abundant quantity and meaning can be obtained, and the combination feature can describe the features of the dialogue data to be processed more comprehensively, so as to determine whether the dialogue data to be processed has question-answer mismatch or not more accurately. In addition, the details of the embodiment, which are not described in detail, and the technical effects that can be achieved can be referred to the related description in the embodiment shown in fig. 1, which are not described herein.
Optionally, for the manner provided by the embodiment shown in fig. 4, optionally, feature combinations of statistical features may also be performed by the detection model, and the combined features are further input into the detection model, so that a detection result of whether there is a question-answer mismatch in the dialogue data to be processed is output by the detection model.
That is, compared with the detection models mentioned in the embodiments shown in fig. 1 and 3, the detection model corresponding to the embodiment shown in fig. 4 also has a feature combination function, and fig. 5 is a flowchart of another model training provided in the embodiment of the present invention for the detection model for feature combination. This embodiment may equally be performed by the training device described above. As shown in fig. 5, the following steps may be included in the steps:
S401, acquiring statistical characteristics of training dialogue data comprising at least one round of dialogue, and reflecting whether the training dialogue data has a question-answer mismatch or not.
The specific implementation process of the above step S401 may refer to the specific description of the related steps in the embodiment shown in fig. 3, which is not repeated herein.
S402, feature combination is carried out on the statistical features of the training dialogue data to obtain combined features of the training dialogue data.
The feature combination process of the statistical features of the training session data is similar to the feature combination process of the statistical features of the session data to be processed, and the specific implementation process of the step 402 may refer to the specific description of the related steps in the embodiment shown in fig. 4, which is not repeated here.
S403, taking the combined features of the training dialogue data as training data, and taking the labeling result as supervision information to perform model training so as to obtain a detection model.
Finally, the combined characteristics of the training dialogue data can be used as training data, and the labeling result of the training dialogue data is used as supervision information to carry out supervised training. Alternatively, the loss calculation may be performed according to the following formula, and the detection model training may be implemented in a gradient descent manner:
Wherein p ' is the confidence of the detection result which is output by the detection model and indicates that the training dialogue sample has question-answer mismatch, sigma is a loss function used for calculating a loss value between the detection result and the standard result, x=w1 ' X1' +w2' X2' + Wk ' Xk ' +b ', X1', X2'. Xk ' is a combined feature of the training dialogue data, W1', W2'. Wk ' is a weight value of the detection model, b ' is a bias value of the detection model, and the weight value and the bias value jointly form a model parameter of the detection model.
In this embodiment, the statistical features of the training session data may be combined first to obtain the combined features. And then model training is carried out based on more combination features with more meanings so as to improve the detection accuracy of the detection model.
In order to further improve accuracy of the detection model, fig. 6 is a flowchart of another model training provided in an embodiment of the present invention. This embodiment may equally be performed by the training device described above. As shown in fig. 6, the following steps may be included in the steps:
s501, acquiring statistical characteristics of training dialogue data containing at least one round of questions and answers, and reflecting whether the training dialogue data has a question and answer unmatched labeling result.
S502, feature combination is carried out on the statistical features of the training dialogue data so as to obtain combined features of the training dialogue data.
The specific implementation process of the above steps S501 to S502 may refer to the specific description of the related steps in the embodiment shown in fig. 5, which is not repeated herein.
S503, taking the combined characteristics of the training dialogue data as training data, and taking the labeling results as supervision information to respectively train a preset number of initial models so as to obtain alternative models with different model parameters.
Then, the training device may further obtain a preset number of initial models, and in an alternative manner, the preset number of initial models may have the same model structure and different initial model parameters, and the combined features of the training session data in step S502 may be used as training data, and the labeling results of the training session data may be used as supervision information to respectively train the preset number of initial models. Since the initial model parameters of the initial model are different, alternative models with different model parameters can be obtained even using the same training data.
Alternatively, the training session data in step S501 may be divided by a preset number. And then, respectively taking the combined characteristics of each training dialogue data as training data of a preset number of initial models, and respectively training the preset number of initial models by taking the labeling results of the training dialogue data as supervision information. The preset number of initial models may have the same model structure and the same initial model parameters. Because the training dialogue data for training the initial model is different, the combination characteristics of the training dialogue data are different, and the model parameters of the finally obtained alternative model are different.
S504, determining parameters of the detection model according to the parameters of the alternative model to obtain the detection model.
Finally, the training device may determine model parameters of the detection model according to respective model parameters of a preset number of candidate models. Alternatively, the model parameters of each of the preset number of candidate models may be averaged or weighted and summed, and the calculation result of the model parameters may be used as the model parameters of the detection model, thereby obtaining the detection model. The detection model has the same model structure as the initial model and the alternative model.
In this embodiment, the detection accuracy of the detection model is improved by combining features and comprehensively considering model parameters, i.e., parameter integration. Specifically, the training device trains a plurality of alternative models by using the combined features, and then obtains model parameters of the detection model by comprehensively considering respective model parameters of the alternative models, thereby improving the accuracy of the model parameters of the detection model, namely improving the detection accuracy of the detection model.
In order to further improve the accuracy of detecting the question-answer mismatch, in addition to the embodiment shown in fig. 6, alternatively, a plurality of detection results may be comprehensively considered to obtain more accurate detection results. Specifically, after training to obtain a preset number of candidate models according to the embodiment shown in fig. 6, the preset number of candidate models may also output the detection results of the to-be-processed dialogue data and the confidence of the detection results, and then synthesize a plurality of detection results, for example, average the confidence or weight and sum the confidence, so as to finally obtain the detection results of the to-be-processed dialogue data.
For ease of understanding, the following may exemplarily describe a specific implementation procedure of the session data processing method provided in the above embodiments. The following process may also be understood in conjunction with fig. 7.
The dialogue system can be deployed in a cloud or hardware device to provide consultation services for users in different scenes. The user and dialog system may generate the multi-segment dialog presented in the above-described embodiments during a human-machine dialog. In the man-machine conversation process, the conversation system can record conversations generated by two parties of the conversation in real time, and can also count conversation data after the conversation is finished so as to obtain the statistical characteristics of the conversation data.
The processing device may obtain the statistical feature or, alternatively, the processing device may directly use this statistical feature, which is independent of the dialog content and appears as a numerical value, to determine whether there is a question-answer mismatch in the dialog data. Optionally, the processing device may also perform feature combination on the statistical features to obtain a combined feature. The processing device may further utilize this combined feature to determine if there is a question-answer mismatch in the dialog data. The detection result may also be displayed on the operation interface shown in fig. 2.
Alternatively, the above-described processing procedure performed by the processing apparatus may be specifically performed by a detection model deployed in the processing apparatus. The specific training process of the detection model may be referred to the related description in the above embodiments, which is not repeated herein.
The embodiments shown in fig. 1 to 7 describe in detail the processing of dialogue data by the processing device. On the basis, a man-machine interaction method can be provided from the interaction flow of the user. Fig. 8 is a flowchart of a man-machine interaction method according to an embodiment of the present invention. The execution subject of the method may specifically be a processing device providing an operation interface. As shown in fig. 8, the method may include the steps of:
s601, displaying a prompt message, wherein the prompt message is generated when the number of abnormal dialogue data is larger than the preset number, the abnormal dialogue data comprises dialogue data with question-answer mismatch, the dialogue data is generated by a user and a dialogue system in a man-machine dialogue process, whether the question-answer mismatch exists in the dialogue data is determined by statistical characteristics of the dialogue data, and the statistical characteristics are irrelevant to dialogue contents of the dialogue data.
The processing device may detect whether there is a question-answer mismatch for each of the dialog data to be processed generated during the history period in the manner provided in the embodiments described above. Wherein the pending dialogue data for which there is a question-answer mismatch may be referred to as abnormal dialogue data. If the number of abnormal dialogue data is greater than the preset threshold, a prompt message can be displayed on an operation interface provided by the processing equipment.
The alert message may include a text alert that optimizes the dialog system, such as the operator interface shown in fig. 2. Optionally, other information related to the abnormal dialogue data may be displayed on the statistical interface, such as the total number of dialogue data in the current history period, the duty ratio of the abnormal dialogue data in the current history period, the change rate of the abnormal dialogue data in the current history period compared with the abnormal dialogue data in the previous history period, and so on.
S602, responding to the processing operation of the prompt message.
The processing device may then determine whether to optimize the dialog system in response to the operation of the prompt message by the operation staff member.
In this embodiment, with the aid of an operation interface provided by the processing device, an operator of the dialog system can interact with the processing device to determine whether to process the dialog system. Meanwhile, the operation and maintenance personnel can accurately know the performance of the dialogue system by means of various information displayed on the operation interface. In addition, the details not described in detail in this embodiment and the technical effects that can be achieved may be related to the descriptions in the above embodiments, which are not described herein.
Alternatively, the dialog system may be deployed in an online customer service platform, and the prompting message may be displayed on a statistical interface provided by the online customer service platform.
The solution of mismatching questions and answers provided by the embodiments above may also be provided as a software operation service (Software as aService, saaS for short) to an on-line customer service platform or other on-line platform operators deployed with a dialogue system. The service may also be deployed in a service platform in the cloud. Fig. 9 is a flowchart of another man-machine interaction method according to an embodiment of the present invention. The execution subject of the method may be a service platform. As shown in fig. 9, the method may include the steps of:
s701, in response to the triggering of the detection operation, acquiring the statistical characteristics of the to-be-processed dialogue data comprising at least one round of questions and answers, wherein the statistical characteristics are irrelevant to the dialogue content of the to-be-processed dialogue data.
S702, detecting whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics.
S703, displaying the detection result of the dialogue data to be processed.
Optionally, the operation and maintenance personnel may trigger a detection operation on the cloud operation platform, and then the service platform may obtain the to-be-processed dialogue data including at least one round of question and answer, and detect whether the to-be-processed dialogue data has a question and answer mismatch according to the manner provided by the above embodiments. The final detection result can also be displayed on a statistical interface mentioned by the service platform.
The details and the technical effects that can be achieved in this embodiment may be related to the descriptions in the above embodiments, which are not described in detail herein.
A dialogue data processing apparatus of one or more embodiments of the present invention will be described in detail below. Those skilled in the art will appreciate that these dialog data processing devices may be configured using commercially available hardware components through the steps taught by the present solution.
Fig. 10 is a schematic structural diagram of a session data processing apparatus according to an embodiment of the present invention, as shown in fig. 10, where the apparatus includes:
the feature acquisition module 11 is configured to acquire a statistical feature of the to-be-processed dialogue data including at least one question and answer, where the statistical feature is irrelevant to dialogue content of the to-be-processed dialogue data.
And the determining module 12 is used for determining whether the question-answer mismatch exists in the dialogue data to be processed according to the statistical characteristics.
Optionally, the determining module 12 is configured to input the statistical feature into a detection model, so as to output, by the detection model, whether the dialog data to be processed has a question-answer mismatch.
Optionally, the apparatus further comprises: the training module 13 is used for acquiring statistical characteristics of training dialogue data comprising at least one round of questions and answers and reflecting whether the training dialogue data has a labeling result of mismatching of the questions and the answers; respectively training a preset number of initial models by taking the statistical characteristics of the training dialogue data as training data and the labeling results as supervision information to obtain alternative models with different model parameters; and determining the model parameters of the detection model according to the model parameters of the alternative model to obtain the detection model.
Optionally, the determining module 12 is configured to perform feature combination on any at least two of the plurality of statistical features to obtain a combined feature of the dialog data to be processed; and determining whether the dialogue data to be processed has question-answer mismatch according to the combined characteristics of the dialogue data to be processed.
Optionally, the determining module 12 is configured to input the statistical feature into a detection model, so as to perform feature combination on the statistical feature by the detection model; and inputting the combined characteristics into the detection model so as to output a detection result reflecting whether the dialogue data to be processed has question-answer mismatch or not by the detection model.
Optionally, the training module 13 is configured to obtain statistical features of training dialogue data including at least one round of questions and answers, and a labeling result reflecting whether the training dialogue data has a question and answer mismatch; feature combination is carried out on the statistical features of the training dialogue data so as to obtain combined features of the training dialogue data; and taking the combined characteristics of the training dialogue data as training data, and taking the labeling result as supervision information to perform model training so as to obtain the detection model.
Optionally, the training module 13 is configured to train a preset number of initial models respectively by using the combined feature of the training dialogue data as training data and the labeling result as supervision information, so as to obtain alternative models with different model parameters; and determining parameters of the detection model according to the parameters of the alternative model to obtain the detection model.
The obtaining module 11 is configured to obtain the statistical feature output by the dialogue system, and the statistical feature is output by the dialogue system after the man-machine dialogue is completed by counting the dialogue data to be processed. Wherein the dialog data to be processed are generated by the user and the dialog system during a man-machine dialog.
The apparatus further comprises: a statistics module 14 and an information display module 15.
The statistics module 14 is configured to count the number of abnormal dialogue data that have mismatching questions and answers in the historical time period.
The information display module 15 is configured to display the number of abnormal dialogue data on a statistics interface; and if the number of the abnormal dialogue data is larger than the preset number, displaying a prompt message for optimizing the dialogue system on the statistical interface.
The dialogue data to be processed comprises texts or voices generated by a user and a dialogue system in a man-machine dialogue process; the statistical features of the dialog data to be processed include numerical type features.
The apparatus shown in fig. 10 may perform the method of the embodiment shown in fig. 1 to 6, and reference is made to the relevant description of the embodiment shown in fig. 1 to 6 for parts of this embodiment not described in detail. The implementation process and technical effects of this technical solution are described in the embodiments shown in fig. 1 to 6, and are not described herein.
Fig. 11 is a schematic structural diagram of a man-machine interaction device according to an embodiment of the present invention, as shown in fig. 11, where the device includes:
the display module 21 is configured to display a prompt message, where the prompt message is generated when the number of abnormal dialogue data is greater than a preset number, where the abnormal dialogue data includes dialogue data with question-answer mismatch, where the dialogue data is generated by a user and a dialogue system during a man-machine dialogue, and whether the dialogue data has question-answer mismatch is determined by a statistical feature of the dialogue data, where the statistical feature is unrelated to dialogue content of the dialogue data.
A response module 22, configured to respond to the processing operation for the hint message.
The apparatus shown in fig. 11 may perform the method of the embodiment shown in fig. 8, and reference is made to the relevant description of the embodiment shown in fig. 8 for parts of this embodiment not described in detail. The implementation process and the technical effect of this technical solution refer to the description in the embodiment shown in fig. 8, and are not repeated here.
In one possible design, the method for processing dialogue data provided in the foregoing embodiments may be applied to an electronic device, as shown in fig. 12, where the electronic device may include: a first processor 31 and a first memory 32. The first memory 32 is used for storing a program for supporting the electronic device to execute the session data processing method provided in the embodiment shown in fig. 1 to 6, and the first processor 31 is configured to execute the program stored in the first memory 32.
The program comprises one or more computer instructions, wherein the one or more computer instructions, when executed by the first processor 31, are capable of performing the steps of:
acquiring statistical features of dialogue data to be processed containing at least one round of questions and answers, wherein the statistical features are irrelevant to dialogue content of the dialogue data to be processed;
and determining whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics.
Optionally, the first processor 31 is further configured to perform all or part of the steps in the embodiments shown in fig. 1 to 6.
The electronic device may further include a first communication interface 33 in a structure for the electronic device to communicate with other devices or communication systems.
In addition, an embodiment of the present invention provides a computer storage medium storing computer software instructions for the electronic device, which includes a program for executing the session data processing method shown in fig. 1 to 6.
In one possible design, the human-computer interaction method provided in the above embodiments may be applied to another electronic device, as shown in fig. 13, where the electronic device may include: a second processor 41 and a second memory 42. The second memory 42 is used for storing a program for supporting the electronic device to execute the man-machine interaction method provided in the embodiment shown in fig. 8, and the second processor 41 is configured to execute the program stored in the second memory 42.
The program comprises one or more computer instructions, wherein the one or more computer instructions, when executed by the second processor 41, are capable of performing the steps of:
displaying a prompt message, wherein the prompt message is generated when the number of abnormal dialogue data is larger than a preset number, the abnormal dialogue data comprises dialogue data with question-answer mismatch, the dialogue data is generated by a user and the dialogue system in a man-machine dialogue process, whether the dialogue data has question-answer mismatch is determined by the statistical characteristics of the dialogue data, and the statistical characteristics are irrelevant to dialogue contents of the dialogue data;
Responding to the processing operation of the prompt message.
Optionally, the second processor 41 is further configured to perform all or part of the steps in the embodiment shown in fig. 8.
The electronic device may further include a second communication interface 43 in its structure for communicating with other devices or communication systems.
In addition, an embodiment of the present invention provides a computer storage medium, configured to store computer software instructions for the electronic device, where the computer storage medium includes a program for executing the man-machine interaction method shown in fig. 8.
Fig. 14 is a schematic structural diagram of another man-machine interaction device according to an embodiment of the present invention, as shown in fig. 14, the device includes:
an obtaining module 51, configured to obtain, in response to a trigger of the detection operation, a statistical feature of the to-be-processed dialogue data including at least one round of questions and answers, where the statistical feature is irrelevant to dialogue content of the to-be-processed dialogue data.
And the detection module 52 is used for detecting whether the question-answer mismatch exists in the dialogue data to be processed according to the statistical characteristics.
And a result display module 53, configured to display a detection result of the dialog data to be processed.
The apparatus shown in fig. 14 may perform the method of the embodiment shown in fig. 9, and reference is made to the relevant description of the embodiment shown in fig. 9 for parts of this embodiment not described in detail. The implementation process and the technical effect of this technical solution are described in the embodiment shown in fig. 9, and are not described herein.
In one possible design, the human-computer interaction method provided in the above embodiments may be applied to another electronic device, as shown in fig. 15, where the electronic device may include: a third processor 61 and a third memory 62. The third memory 62 is used for storing a program for supporting the electronic device to execute the man-machine interaction method provided in the embodiment shown in fig. 9, and the third processor 61 is configured to execute the program stored in the third memory 62.
The program comprises one or more computer instructions, wherein the one or more computer instructions, when executed by the third processor 61, are capable of performing the steps of:
in response to triggering of a detection operation, obtaining statistical features of the dialogue data to be processed comprising at least one round of questions and answers, wherein the statistical features are irrelevant to dialogue contents of the dialogue data to be processed;
detecting whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics;
and displaying the detection result of the dialogue data to be processed.
Optionally, the third processor 61 is further configured to perform all or part of the steps in the embodiment shown in fig. 9.
A third communication interface 63 may also be included in the structure of the electronic device for the electronic device to communicate with other devices or communication systems.
In addition, an embodiment of the present invention provides a computer storage medium, configured to store computer software instructions for the electronic device, where the computer storage medium includes a program for executing the man-machine interaction method shown in fig. 9.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (14)
1. A method of processing dialogue data, comprising:
acquiring statistical features of dialogue data to be processed comprising at least one round of questions and answers, wherein the statistical features are irrelevant to dialogue content of the dialogue data to be processed;
and determining whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics.
2. The method of claim 1, wherein said determining whether there is a question-answer mismatch in the dialog data to be processed based on the statistical features comprises:
And inputting the statistical characteristics into a detection model to output whether the dialogue data to be processed have question-answer mismatch or not by the detection model.
3. The method according to claim 2, wherein the method further comprises:
acquiring statistical characteristics of training dialogue data containing at least one round of questions and answers, and reflecting whether the training dialogue data has a labeling result of mismatching of the questions and the answers;
respectively training a preset number of initial models by taking the statistical characteristics of the training dialogue data as training data and the labeling results as supervision information to obtain alternative models with different model parameters;
and determining the model parameters of the detection model according to the model parameters of the alternative model to obtain the detection model.
4. The method of claim 1, wherein the statistical features comprise a plurality of; the step of determining whether the dialogue data to be processed has question-answer mismatch according to the statistical characteristics comprises the following steps:
performing feature combination on any at least two features in a plurality of statistical features to obtain combined features of the dialogue data to be processed;
and determining whether the dialogue data to be processed has question-answer mismatch according to the combined characteristics of the dialogue data to be processed.
5. The method of claim 4, wherein the feature combining the statistical features to obtain the combined features of the dialog data to be processed comprises:
inputting the statistical features into a detection model to perform feature combination on the statistical features by the detection model;
the determining whether the dialogue data to be processed has question-answer mismatch according to the combination features comprises the following steps:
and inputting the combined characteristics into the detection model so as to output a detection result reflecting whether the dialogue data to be processed has question-answer mismatch or not by the detection model.
6. The method of claim 5, wherein the method further comprises:
acquiring statistical characteristics of training dialogue data containing at least one round of questions and answers, and reflecting whether the training dialogue data has a labeling result of mismatching of the questions and the answers;
feature combination is carried out on the statistical features of the training dialogue data so as to obtain combined features of the training dialogue data;
and taking the combined characteristics of the training dialogue data as training data, and taking the labeling result as supervision information to perform model training so as to obtain the detection model.
7. The method of claim 6, wherein model training the combined features of the training session data as training data and the labeling results as supervision information to obtain the detection model comprises:
respectively training a preset number of initial models by taking the combined features of the training dialogue data as training data and the labeling results as supervision information to obtain alternative models with different model parameters;
and determining parameters of the detection model according to the parameters of the alternative model to obtain the detection model.
8. The method according to claim 1, wherein the dialog data to be processed is generated by a user and a dialog system during a man-machine dialog;
the step of obtaining the statistical characteristics of the dialogue data to be processed comprises the following steps:
and acquiring the statistical characteristics output by the dialogue system, and outputting the statistical characteristics after the dialogue system counts the dialogue data to be processed after the man-machine dialogue is finished.
9. The method of claim 8, wherein the method further comprises:
counting the number of abnormal dialogue data with mismatching questions and answers in a historical time period;
Displaying the number of abnormal dialogue data on a statistical interface;
and if the number of the abnormal dialogue data is larger than the preset number, displaying a prompt message for optimizing the dialogue system on the statistical interface.
10. The method of claim 1, wherein the dialog data to be processed comprises text or speech generated by a user and a dialog system during a human-machine dialog; the statistical features of the dialog data to be processed include numerical type features.
11. A human-computer interaction method, comprising:
displaying a prompt message, wherein the prompt message is generated when the number of abnormal dialogue data is larger than the preset number, the abnormal dialogue data comprises dialogue data with question-answer mismatch, the dialogue data is generated by a user and a dialogue system in a man-machine dialogue process, and whether the dialogue data has question-answer mismatch is determined by the statistical characteristics of the dialogue data;
responding to the processing operation of the prompt message.
12. The method of claim 11, wherein the dialog system is deployed in an online customer service platform; the displaying of the prompt message includes:
and displaying the prompt message on a statistical interface provided by the online customer service platform.
13. An electronic device, comprising: a memory, a processor; wherein the memory has stored thereon executable code which, when executed by the processor, causes the processor to perform the dialog data processing method of any of claims 1 to 10 or to perform the human interaction method of claim 11 or 12.
14. A non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform the dialog data processing method of any of claims 1 to 10 or to perform the human-machine interaction method of claim 11 or 12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310559957.9A CN116681060A (en) | 2023-05-15 | 2023-05-15 | Dialogue data processing method, man-machine interaction method, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310559957.9A CN116681060A (en) | 2023-05-15 | 2023-05-15 | Dialogue data processing method, man-machine interaction method, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116681060A true CN116681060A (en) | 2023-09-01 |
Family
ID=87788203
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310559957.9A Pending CN116681060A (en) | 2023-05-15 | 2023-05-15 | Dialogue data processing method, man-machine interaction method, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116681060A (en) |
-
2023
- 2023-05-15 CN CN202310559957.9A patent/CN116681060A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110096191B (en) | Man-machine conversation method and device and electronic equipment | |
| US10623573B2 (en) | Personalized support routing based on paralinguistic information | |
| US10262195B2 (en) | Predictive and responsive video analytics system and methods | |
| CN112988991B (en) | Method and system for performing anti-fraud intervention through man-machine conversation | |
| US20210191965A1 (en) | Dialogue management system with hierarchical classification and progression | |
| US20200274970A1 (en) | Automated systems and methods for natural language processing with speaker intention inference | |
| CN107818798A (en) | Customer service quality evaluating method, device, equipment and storage medium | |
| US9818405B2 (en) | Dialog management system | |
| US11336770B2 (en) | Systems and methods for analyzing coaching comments | |
| CN109981910A (en) | Business recommended method and apparatus | |
| KR102225603B1 (en) | Device and Method of Scoring Emotion for Psychological Consultation | |
| CN110610705A (en) | Voice interaction prompter based on artificial intelligence | |
| CN116681060A (en) | Dialogue data processing method, man-machine interaction method, equipment and storage medium | |
| CN115271932A (en) | Outbound risk identification method and device | |
| EP4020468A1 (en) | System, method and apparatus for conversational guidance | |
| CN110765242A (en) | Method, device and system for providing customer service information | |
| CN115379054A (en) | Method and device for processing call-out operation | |
| CN114157763A (en) | Information processing method and device in interactive process, terminal and storage medium | |
| CN114138943A (en) | Dialog message generation method and device, electronic equipment and storage medium | |
| US11755652B2 (en) | Information-processing device and information-processing method | |
| Bang et al. | UX Design and Evaluation on Conversational Bot Supporting Multi-Turn and Multi-Domain Dialogues | |
| WO2018012412A1 (en) | Information output system, information output method, and recording medium | |
| EP4093005A1 (en) | System method and apparatus for combining words and behaviors | |
| CN111161706A (en) | Interaction method, device, equipment and system | |
| CN115905642B (en) | Method, system, terminal and storage medium for enhancing speaking emotion |
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 |