Model training methods, methods for determining user consultation information, devices, and equipment

Abstract

This disclosure presents embodiments of a model training method, a user consultation information determination method, an apparatus, an electronic device, and a computer-readable medium. One specific implementation of the method includes: acquiring user selection information, first user consultation information from user feedback, and at least one second user consultation information output by a user consultation information determination model; dynamically adjusting the labels of the user data input to the user consultation information determination model to obtain training samples; dynamically adjusting the sample weights corresponding to the training samples to obtain adjusted training samples; and retraining the user consultation information determination model to obtain a trained user consultation information determination model. This implementation further trains the user consultation information determination model using the adjusted training samples, enabling the trained model to accurately and effectively determine user consultation information.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, specifically to model training methods, user consultation information determination methods, apparatus, electronic devices, and computer-readable media. Background Technology

[0002] Currently, across various business areas, human or chatbot customer service is commonly used to assist users and answer their business-related questions. Whether human or chatbot, before a user initiates a call, the system typically predicts the user's intent and presents potential questions on the front-end interface for the customer to choose from.

[0003] To predict a user's consultation intent, the common approach is to use machine learning to train a classification model by collecting training datasets offline, based on various user information before the call and the business questions the user inquires about. When a target user calls, the trained classification model is used to predict the possible consultation intent category based on the user information before the call, and then the questions under that intent category are displayed on the front-end interface for the target user to choose from.

[0004] However, using the above methods to predict and categorize potential user inquiries typically suffers from the following problems: the classification model cannot automatically optimize and update itself based on online user feedback. In the long run, this may lead to inaccurate predictions of user inquiries, thus significantly impacting user experience. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide model training methods, user consultation information determination methods, apparatuses, devices, and computer-readable media to address the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a model training method, which includes: acquiring user selection information, first user consultation information fed back by the user, and at least one second user consultation information output by a user consultation information determination model, wherein the user selection information indicates whether the user selects user consultation information recommended by the system; dynamically adjusting the labels of user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples, wherein the labels are second target user consultation information among the at least one second user consultation information; dynamically adjusting the sample weights corresponding to the training samples to obtain adjusted training samples; and retraining the user consultation information determination model based on the adjusted training samples to obtain a trained user consultation information determination model.

[0008] Optionally, the above-mentioned method of dynamically adjusting the labels of user data input to the user consultation information model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replacing the labels of the user data with the first user consultation information to obtain the training samples.

[0009] Optionally, the above-mentioned dynamic adjustment of the sample weights corresponding to the above-mentioned training samples to obtain the adjusted training samples includes: adjusting the sample weights corresponding to the above-mentioned training samples to a value α to obtain the adjusted training samples, wherein the above-mentioned value α is a value greater than 1.

[0010] Optionally, the above-mentioned method of dynamically adjusting the labels of user data input to the user consultation information model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: in response to the user selection information representing that the user selects user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replacing the labels of the user data with the first user consultation information to obtain the training samples.

[0011] Optionally, the above-mentioned method of dynamically adjusting the labels of the user data input to the model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: in response to the user selection information representing that the user selects user consultation information recommended by the system, and that the first user consultation information exists in the at least one second user consultation information, determining the first user consultation information as the label of the user data to obtain the training samples.

[0012] Optionally, the above-mentioned dynamic adjustment of the sample weights corresponding to the above-mentioned training samples to obtain the adjusted training samples includes: adjusting the sample weights corresponding to the above-mentioned training samples to a value β to obtain the adjusted training samples, wherein the above-mentioned value β is a value between 0 and 1.

[0013] Optionally, the above-mentioned dynamic adjustment of the labels of the user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information exists in the at least one second user consultation information, determining the user data and the second target user consultation information as the training samples.

[0014] Optionally, the above-mentioned dynamic adjustment of the sample weights corresponding to the above-mentioned training samples to obtain the adjusted training samples includes: keeping the sample weights of the above-mentioned training samples unchanged, and determining the above-mentioned training samples as the adjusted training samples.

[0015] Optionally, the aforementioned at least one second user consultation information is output through the following steps: in response to receiving a request for determining user consultation information, obtaining user data associated with the obtained user identifier based on the obtained user identifier; inputting the user data into a pre-trained user consultation information determination model to obtain the aforementioned at least one second user consultation information.

[0016] Optionally, the aforementioned user consultation information determination model is trained through the following steps: obtaining the target user data from the target terminal through a relevant interface; processing the target user data to obtain processed target user data; storing the processed target user data in a database; responding to a received data acquisition request, retrieving a user dataset associated with the target business scenario from the database as a training sample for the aforementioned user consultation information determination model, wherein the aforementioned user dataset includes: the aforementioned target user data; and training the aforementioned user consultation information determination model based on the aforementioned training sample.

[0017] Optionally, the database mentioned above stores user datasets associated with each business scenario in each business scenario.

[0018] Secondly, some embodiments of this disclosure provide a method for determining user consultation information. The method includes: in response to receiving a request for determining user consultation information, obtaining user data related to the user identifier of a target user obtained in advance; and inputting the user data into a pre-trained user consultation information determination model to obtain at least one piece of user consultation information.

[0019] Thirdly, some embodiments of this disclosure provide a model training apparatus, comprising: an acquisition unit configured to acquire user selection information, first user consultation information fed back by the user, and at least one second user consultation information output by a user consultation information determination model, wherein the user selection information indicates whether the user selects user consultation information recommended by the system; a first adjustment unit configured to dynamically adjust the labels of user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information, to obtain training samples, wherein the labels are second target user consultation information among the at least one second user consultation information; a second adjustment unit configured to dynamically adjust the sample weights corresponding to the training samples, to obtain adjusted training samples; and a training unit configured to retrain the user consultation information determination model based on the adjusted training samples, to obtain a trained user consultation information determination model.

[0020] Optionally, the first adjustment unit is further configured to: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replace the label of the user data with the first user consultation information to obtain the training sample.

[0021] Optionally, the second adjustment unit is further configured to: adjust the sample weights corresponding to the above training samples to a value α, thereby obtaining the adjusted training samples, wherein the above value α is a value greater than 1.

[0022] Optionally, the first adjustment unit is further configured to: in response to the user selection information indicating that the user has selected user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replace the label of the user data with the first user consultation information to obtain the training sample.

[0023] Optionally, the first adjustment unit is further configured to: in response to the user selection information characterizing the user selection of user consultation information recommended by the system, and the first user consultation information existing in at least one second user consultation information, determine the first user consultation information as the label of the user data, and obtain the training sample.

[0024] Optionally, the second adjustment unit is further configured to: adjust the sample weights corresponding to the above training samples to a value β, thereby obtaining the adjusted training samples, wherein the above value β is a value between 0 and 1.

[0025] Optionally, the first adjustment unit is further configured to: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information exists in the at least one second user consultation information, determine the user data and the second target user consultation information as the training samples.

[0026] Optionally, the second adjustment unit is further configured to: keep the sample weights of the above training samples unchanged, and determine the above training samples as the above-adjusted training samples.

[0027] Fourthly, some embodiments of this disclosure provide a user consultation information determination apparatus, the apparatus comprising: an acquisition unit configured to, in response to receiving a request for determining user consultation information, acquire user data related to the user identifier of a pre-acquired target user; and an input unit configured to input the user data into a pre-trained user consultation information determination model to obtain at least one user consultation information.

[0028] Fifthly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, which, when executed by one or more processors, cause the one or more processors to implement the methods as described in the first or third aspect.

[0029] Sixthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method as described in either the first or third aspect.

[0030] The various embodiments of this disclosure have the following beneficial effects: the model training method of some embodiments of this disclosure can further train the user consultation information determination model using adjusted training samples, so that the trained user consultation information determination model can accurately and effectively determine user consultation information. Specifically, the classification model cannot automatically optimize and update the model based on feedback information from online users. In the long run, this may lead to the classification model predicting user inquiries inaccurately, thus seriously affecting the user experience. Based on this, the model training method of some embodiments of this disclosure first obtains user selection information, first user consultation information from user feedback, and at least one second user consultation information output by the user consultation information determination model, so as to dynamically adjust the labels of user data and the sample weights corresponding to the training samples. Among them, the user selection information indicates whether the user selects the user consultation information recommended by the system. Then, based on the user selection information, the first user consultation information, and the at least one second user consultation information, the labels of the user data input to the user consultation information determination model are dynamically adjusted to obtain training samples. Among them, the labels are the second target user consultation information among the at least one second user consultation information. Here, the labels of user data are dynamically adjusted to make the labels of user data more accurate. Therefore, using the adjusted user data to train the user consultation information determination model makes the model more accurate and effective in identifying user consultation information. Furthermore, the sample weights corresponding to the training samples are dynamically adjusted to obtain adjusted training samples. It should be noted that by adjusting the sample weights, the user consultation information determination model can learn more from training samples that were misclassified by the model. This indirectly makes the model more accurate and effective in identifying user consultation information. Finally, based on the adjusted training samples, the user consultation information determination model is retrained to obtain a trained model. Thus, the above model training method can use adjusted training samples to further train the user consultation information determination model, enabling the trained model to accurately and effectively identify user consultation information. Attached Figure Description

[0031] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0032] Figure 1 This is a schematic diagram of an application scenario of the model training method of some embodiments of this disclosure;

[0033] Figure 2 This is a flowchart of some embodiments of the model training method according to this disclosure;

[0034] Figure 3 These are flowcharts of other embodiments of the model training method according to this disclosure;

[0035] Figure 4 This is a flowchart of yet another embodiment of the model training method according to the present disclosure;

[0036] Figure 5 This is a schematic diagram of an application scenario of a user consultation information determination method according to some embodiments of this disclosure;

[0037] Figure 6 This is a flowchart of some embodiments of the method for determining based on user consultation information in this disclosure;

[0038] Figure 7 This is a schematic diagram of the structure of some embodiments of the model training apparatus according to the present disclosure;

[0039] Figure 8 These are schematic diagrams of some embodiments of the device determined based on user consultation information disclosed herein;

[0040] Figure 9 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0041] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0042] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0043] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0044] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0045] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0046] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] Figure 1 This is a schematic diagram of an application scenario of the model training method of some embodiments of this disclosure.

[0048] like Figure 1As shown, the electronic device 101 can first acquire user selection information 103, first user consultation information 102 from user feedback, and at least one second user consultation information 104 output by the user consultation information determination model 105. The user selection information 103 indicates whether the user has selected the user consultation information recommended by the system. In this application scenario, the user selection information 103 can include: the user not selecting the user consultation information recommended by the system, and the user selecting the user consultation information recommended by the system. The first user consultation information 102 can be: "Mobile phone battery problem". The at least one second user consultation information 104 can include: second user consultation information 1041, second user consultation information 1042, and second user consultation information 1043. The second user consultation information 1041 can be: "Mobile phone screen problem". The second user consultation information 1042 can be: "Mobile phone virus problem". The second user consultation information 1043 can be: "Mobile phone price problem". Then, based on the user selection information 103, the first user consultation information 102, and at least one second user consultation information 104, the label 107 of the user data 106 input to the user consultation information determination model 105 is dynamically adjusted to obtain training sample 109. Here, label 107 is the second target user consultation information among the at least one second user consultation information 104. In this application scenario, electronic device 101 can adjust the label 107 of the user data 106 input to the user consultation information determination model 105 to label 108. Label 107 can be: "Mobile phone screen problem". Label 108 can be: "Mobile phone battery problem". Furthermore, the sample weight 110 corresponding to the training sample 109 is dynamically adjusted to obtain adjusted training sample 112. In this application scenario, electronic device 101 can dynamically adjust the sample weight 110 corresponding to the training sample 109 to sample weight 111. Sample weight 110 can be: "1". Sample weight 111 can be: "0.5". Finally, based on the adjusted training sample 112, the user consultation information determination model 105 is retrained to obtain the trained user consultation information determination model 113.

[0049] It should be noted that the aforementioned electronic device 101 can be either hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the electronic device is software, it can be installed in the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0050] It should be understood that Figure 1The number of electronic devices shown is merely illustrative. Any number of electronic devices can be used depending on the implementation requirements.

[0051] Continue to refer to Figure 2 The diagram illustrates a flow 200 of some embodiments of a model training method according to the present disclosure. This model training method includes the following steps:

[0052] Step 201: Obtain user selection information, first user consultation information from user feedback, and user consultation information to determine at least one second user consultation information output by the model.

[0053] In some embodiments, the entity executing the model training method (e.g. Figure 1 The electronic device 101 shown can acquire user selection information, first user consultation information from user feedback, and at least one second user consultation information output by a user consultation information determination model via wired or wireless means. The user selection information indicates whether the user has selected the user consultation information recommended by the system. The first user consultation information may or may not be present in the at least one second user consultation information.

[0054] It should be noted that the aforementioned user consultation information can be the user's consultation intent information for a specific business scenario. For example, in a shopping scenario, user consultation information could be the type of questions the user wants to ask during the shopping process. The aforementioned user selection information can indicate whether the user clicked on the user consultation information recommended by the system. The aforementioned first user consultation information can be the user's actual consultation intent information that they want to provide to the system. The aforementioned at least one second user consultation information can be the result output by inputting user data into the user consultation information determination model. Here, the aforementioned user consultation information determination model can be a network that performs classification tasks for each user consultation intent. As an example, the aforementioned user consultation information determination model can be one of the following: a distributed gradient boosting library (XGBoost, eXtremeGradient Boosting), a Support Vector Machine (SVM), or a neural network used for classification tasks.

[0055] It should be emphasized that the aforementioned execution entity can also be a prediction engine system, which can acquire user data, store user data, and provide relevant machine learning or deep learning algorithms to provide various services and reuse various services through these algorithms.

[0056] For the e-commerce sector, the overall system architecture of a predictive engine system can include: a data layer, a storage layer, an algorithm layer, a service layer, and a business layer. The data layer can continuously integrate more user information, accumulating massive amounts of user business and behavioral data, which can then be reused across different business processes. This data includes: real-time browsing and clicking information on the shopping application (APP) before a user initiates an online consultation, user history consultations, user history purchase information, user profiles, etc. The storage layer can include, but is not limited to, at least one of the following: a real-time data computing framework Flink, a distributed data storage system HBase, a remote dictionary service (Redis), or an operation support system (OSS). The aforementioned algorithm layer may include, but is not limited to, at least one of the following: DeepFM and Wide&Deep based on the DeepCTR framework (XGBoost, eXtreme Gradient Boosting), Support Vector Machine (SVM), Word2vec (word to vector) model, Principal Component Analysis (PCA), Gradient Boosting Decision Tree (GBDT), Latent Dirichlet Allocation (LDA), and logistic regression model. Some algorithms in the aforementioned algorithm layer can be used to construct models for determining user consultation information. The services provided by the aforementioned service layer may include, but are not limited to, at least one of the following: prediction of user-incoming questions, prediction of orders for user-incoming consultations, and prediction of whether a user will make repeat inquiries. The aforementioned business layer can be the business scenario used by the service layer. The aforementioned business layer may include, but is not limited to, at least one of the following: online chatbot customer service, online voice chatbot customer service, and human customer service.

[0057] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra wideband) connections, and other currently known or future wireless connection methods.

[0058] Step 202: Based on the user selection information, the first user consultation information, and at least one second user consultation information, dynamically adjust the labels of the user data input to the model based on the user consultation information to obtain training samples.

[0059] In some embodiments, the execution entity can dynamically adjust the labels of the user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and at least one second user consultation information, to obtain training samples. Here, the labels are the second target user consultation information among the at least one second user consultation information. Specifically, the second target user consultation information can be the second user consultation information with the highest output probability from the user consultation information determination model among the at least one second user consultation information. A higher output probability from the user consultation information determination model indicates that the second user consultation information corresponding to that output probability is more likely to represent the user's consultation intent information.

[0060] As an example, the aforementioned executing entity can directly adjust the labels of the user data to match the first user consultation information. Therefore, training samples can be generated based on the user data and the first user consultation information.

[0061] In some optional implementations of certain embodiments, in response to the user selection information indicating that the user has selected user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, the execution entity may replace the label of the user data with the first user consultation information to obtain the training sample.

[0062] In some optional implementations of certain embodiments, in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and the first user consultation information existing in at least one second user consultation information, the executing entity can determine the user data and the second target user consultation information as the training sample. It should be noted that the training sample includes: user data and the second target user consultation information.

[0063] Step 203: Dynamically adjust the sample weights corresponding to the above training samples to obtain the adjusted training samples.

[0064] In some embodiments, the execution entity can dynamically adjust the sample weights corresponding to the training samples to obtain adjusted training samples. The sample weights can characterize the importance of the samples. Using samples with low importance to train the user consultation information determination model allows the model to learn less feature information from the samples. Using samples with high importance to train the user consultation information determination model allows the model to learn more feature information from the samples.

[0065] As an example, the aforementioned executing entity can first determine whether the label of the user data is the first user consultation information. Then, in response to the determination that it is not, the label of the user data is adjusted to the first user consultation information to obtain training samples. Furthermore, the sample weight of the training samples is adjusted to 0.5 to obtain the adjusted training samples.

[0066] Step 204: Based on the adjusted training samples, retrain the user consultation information determination model to obtain the trained user consultation information determination model.

[0067] In some embodiments, the execution entity may retrain the user consultation information determination model based on the adjusted training samples to obtain a trained user consultation information determination model. It should be noted that the trained user consultation information determination model can be used to generate subsequent second user consultation information.

[0068] In some optional implementations of certain embodiments, the aforementioned at least one second user consultation information is output through the following steps:

[0069] The first step, in response to receiving a request to determine user inquiry information, is to retrieve user data associated with the acquired user identifier. The request to determine user inquiry information can be sent by the target terminal's system. The user data can be retrieved using the relevant user identifier.

[0070] It should be noted that the acquired user data can also be stored in a relevant database. After the user consultation information determination model outputs the corresponding results, the user data and the corresponding results can be used to further train the user consultation information determination model, making the user consultation information determination model more accurate and effective in determining user consultation information.

[0071] As an example, in response to receiving a request to determine user consultation information, the aforementioned execution entity can obtain user data associated with the user identifier from the target terminal's system through the relevant engine interface.

[0072] The second step involves inputting the aforementioned user data into a pre-trained user consultation information determination model to obtain at least one second user consultation information. The executing entity can then feed back at least one second user consultation information to the terminal and cache the output results in a database.

[0073] In some optional implementations of certain embodiments, the user consultation information determination model described above is trained through the following steps:

[0074] The first step is to obtain the target user data from the target terminal through the relevant interface. After obtaining the target user data from the target terminal through the relevant interface, the executing entity can call the configuration issue module to retrieve frequently asked questions from the user and send these frequently asked questions as a response to the target terminal.

[0075] Here, business experts analyze human customer service data offline to identify information that may be helpful in problem identification. It's important to note that the collection principle should be data that directly or indirectly determines the user's inquiry. For example, in a shopping scenario, a user submits a service order when requesting a return or exchange. The status and type of this service order are likely to determine the user's inquiry. Similar human customer service data includes work orders, dispute orders, and compensation orders. Therefore, the aforementioned target terminals can obtain target user data based on information that is helpful in problem identification.

[0076] The second step is to process the target user data to obtain processed target user data. As an example, since the target user data may be incomplete, the executing entity can perform data completion processing to obtain processed target user data.

[0077] The third step is to store the processed target user data into the database.

[0078] Fourth, in response to the received data acquisition request, the system retrieves the user dataset associated with the target business scenario from the aforementioned database as training samples for the user consultation information determination model. The aforementioned user dataset includes the target user data. The target business scenario can be the user's current business scenario.

[0079] The fifth step is to train the model based on the training samples and the user consultation information.

[0080] Optionally, the database described above stores user datasets associated with each business scenario within each business context. As an example, the database could store user datasets related to human-operated services and datasets related to chatbot customer service.

[0081] In some optional implementations of certain embodiments, the execution entity may keep the sample weights of the training samples unchanged and determine the training samples as the adjusted training samples.

[0082] The various embodiments of this disclosure have the following beneficial effects: the model training method of some embodiments of this disclosure can further train the user consultation information determination model using adjusted training samples, so that the trained user consultation information determination model can accurately and effectively determine user consultation information. Specifically, the classification model cannot automatically optimize and update the model based on feedback information from online users. In the long run, this may lead to the classification model predicting user inquiries inaccurately, thus seriously affecting the user experience. Based on this, the model training method of some embodiments of this disclosure first obtains user selection information, first user consultation information from user feedback, and at least one second user consultation information output by the user consultation information determination model, so as to dynamically adjust the labels of user data and the sample weights corresponding to the training samples. Among them, the user selection information indicates whether the user selects the user consultation information recommended by the system. Then, based on the user selection information, the first user consultation information, and the at least one second user consultation information, the labels of the user data input to the user consultation information determination model are dynamically adjusted to obtain training samples. Among them, the labels are the second target user consultation information among the at least one second user consultation information. Here, the labels of user data are dynamically adjusted to make the labels of user data more accurate. Therefore, using the adjusted user data to train the user consultation information determination model makes the model more accurate and effective in identifying user consultation information. Furthermore, the sample weights corresponding to the training samples are dynamically adjusted to obtain adjusted training samples. It should be noted that by adjusting the sample weights, the user consultation information determination model can learn more from training samples that were misclassified by the model. This indirectly makes the model more accurate and effective in identifying user consultation information. Finally, based on the adjusted training samples, the user consultation information determination model is retrained to obtain a trained model. Thus, the above model training method can use adjusted training samples to further train the user consultation information determination model, enabling the trained model to accurately and effectively identify user consultation information.

[0083] Continue to refer to Figure 3 The diagram illustrates flow 300 of some other embodiments of the model training method according to this disclosure. This model training method includes the following steps:

[0084] Step 301: Obtain user selection information, first user consultation information from user feedback, and user consultation information to determine at least one second user consultation information output by the model.

[0085] Step 302: In response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, the label of the user data is replaced with the first user consultation information to obtain the training sample.

[0086] In some embodiments, in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and the first user consultation information not existing in the at least one second user consultation information, the executing entity (e.g. Figure 1 The electronic device 101 shown can replace the labels of the above user data with the above first user consultation information to obtain the above training sample.

[0087] It should be noted that since the user did not select the user consultation information recommended by the system, and the aforementioned first user consultation information is not present in at least one of the aforementioned second user consultation information, it indicates that at least one of the second user consultation information output by the user consultation information determination model is not the consultation information the user wants. Therefore, it is necessary to replace the labels of the aforementioned user data with the aforementioned first user consultation information to adjust the accuracy of the user data labels. The resulting training samples can then be used to further train the user consultation information determination model.

[0088] Step 303: Adjust the sample weights corresponding to the above training samples to the value α to obtain the adjusted training samples.

[0089] In some embodiments, the execution entity can adjust the sample weights corresponding to the training samples to a value α to obtain the adjusted training samples. As an example, the value α can be the value "2".

[0090] Here, the adjusted training samples are used to train the user consultation information determination model, so that the user consultation information determination model pays more attention to the feature information of the adjusted training samples when learning feature information.

[0091] Step 304: Based on the adjusted training samples, retrain the user consultation information determination model to obtain the trained user consultation information determination model.

[0092] In some embodiments, the specific implementation of steps 301 and 304 and their resulting technical effects can be found in [reference]. Figure 2 Steps 201 and 204 in the corresponding embodiments will not be repeated here.

[0093] from Figure 3 It can be seen from this that, with Figure 2 Compared to the description of some corresponding embodiments, Figure 3The process flow 300 of the model training method in some corresponding embodiments illustrates the specific steps for adjusting the labels and sample weights of user data. Therefore, the schemes described in these embodiments can use the adjusted training samples to train a user consultation information determination model, making the model pay more attention to the feature information of the adjusted training samples when learning feature information, and accurately and effectively determining user consultation information using the trained model.

[0094] Continue to refer Figure 4 The diagram illustrates a flow 400 of yet another embodiment of a model training method according to the present disclosure. This model training method includes the following steps:

[0095] Step 401: Obtain user selection information, first user consultation information from user feedback, and at least one second user consultation information output by the model based on the user consultation information.

[0096] Step 402: In response to the user selection information indicating that the user selected the user consultation information recommended by the system, and that the first user consultation information exists in at least one second user consultation information, the first user consultation information is determined as the label of the user data, and the training sample is obtained.

[0097] In some embodiments, in response to the user selection information representing the user's selection of system-recommended user consultation information, and the first user consultation information existing in at least one second user consultation information, the executing entity (e.g. Figure 1 The electronic device 101 shown can identify the first user consultation information as the label of the user data to obtain the training sample.

[0098] It should be noted that since the user selects the user consultation information recommended by the system, and the aforementioned first user consultation information exists within at least one of the aforementioned second user consultation information, it indicates that the user's desired consultation information exists within at least one of the second user consultation information output by the user consultation information determination model. Therefore, it is necessary to replace the labels of the aforementioned user data with the aforementioned first user consultation information to adjust the accuracy of the user data labels. This allows the obtained training samples to be used to further train the user consultation information determination model.

[0099] Step 403: Adjust the sample weights corresponding to the above training samples to the numerical value β to obtain the adjusted training samples.

[0100] In some embodiments, the execution entity can adjust the sample weights corresponding to the training samples to a numerical value β to obtain adjusted training samples. As an example, the numerical value β can be the value "0.5".

[0101] Here, the user consultation information determination model has learned most of the feature information of the training samples, and the user consultation information determination model has a relatively accurate determination of user consultation information. Therefore, when training the user consultation information determination model again, we can appropriately reduce the learning of feature information of such samples and pay more attention to the samples that the user consultation information determination model misclassifies.

[0102] Step 404: Based on the adjusted training samples, retrain the user consultation information determination model to obtain the trained user consultation information determination model.

[0103] In some embodiments, the specific implementation of steps 401 and 404 and the resulting technical effects can be found in [reference needed]. Figure 2 Steps 201 and 204 in the corresponding embodiments will not be repeated here.

[0104] from Figure 4 It can be seen from this that, with Figure 2 Compared to the description of some corresponding embodiments, Figure 4 The process 400 of the model training method in some corresponding embodiments illustrates the specific steps of adjusting the labels and sample weights of user data. Therefore, the schemes described in these embodiments can use the adjusted training samples to train a user consultation information determination model. This allows the user consultation information determination model to focus less on the feature information of the adjusted training samples when learning feature information, and to pay more attention to misclassified samples, learning more feature information from misclassified samples. In this way, the trained user consultation information determination model can accurately and effectively determine user consultation information.

[0105] Continue to refer Figure 5 , Figure 5 This is a schematic diagram of an application scenario of the model training method of some embodiments of this disclosure.

[0106] like Figure 5As shown, in response to receiving a request 502 for determining user consultation information, the electronic device 501 can first obtain user data 504 related to the pre-acquired user identifier 503 of the target user. In this application scenario, the user identifier 503 can be "5632". Then, the user data 504 is input into a pre-trained user consultation information determination model 505 to obtain at least one user consultation information 506. In this application scenario, the at least one user consultation information 506 can include: user consultation information 5061, user consultation information 5062, and user consultation information 5063. Specifically, user consultation information 5061 can be: mobile phone screen problem; user consultation information 5062 can be: mobile phone virus problem; and user consultation information 5063 can be: mobile phone price problem.

[0107] It should be noted that the aforementioned electronic device 501 can be either hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the electronic device is software, it can be installed in the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0108] It should be understood that Figure 5 The number of electronic devices shown is merely illustrative. Any number of electronic devices can be used depending on the implementation requirements.

[0109] Continue to refer Figure 6 The flowchart 600 illustrates some embodiments of a method for determining user consultation information according to the present disclosure. This method for determining user consultation information includes the following steps:

[0110] Step 601: In response to receiving a request to determine user consultation information, obtain user data related to the user identifier of the target user obtained in advance.

[0111] In some embodiments, in response to receiving a request for determining user consultation information, the entity executing the user consultation information determination method (e.g., Figure 5 The electronic device 501 shown can obtain user data related to the user identifier of the target user based on the pre-acquired user identifier. The request for determining user inquiry information can be sent by the target terminal's system. Each user has a unique user identifier. The user data can be various information associated with the user. As an example, the user data may include: the user's name, the user's age, the user's historical order information, and the user's historical order cancellation information.

[0112] Step 602: Input the above user data into the pre-trained user consultation information determination model to obtain at least one user consultation information.

[0113] In some embodiments, the aforementioned executing entity may input the aforementioned user data into a pre-trained user consultation information determination model to obtain at least one user consultation information.

[0114] The above embodiments of this disclosure have the following beneficial effects: the user consultation information determination method of some embodiments of this disclosure can accurately and efficiently determine at least one user consultation information by using a pre-trained user consultation information determination model.

[0115] Continue to refer Figure 7 As an implementation of the above figures and methods, this disclosure provides some embodiments of a model training apparatus, which are similar to... Figure 2 Corresponding to the above-described method embodiments, the device can be specifically applied to various electronic devices.

[0116] like Figure 7 As shown, a model training apparatus 700 in some embodiments includes: an acquisition unit 701, a first adjustment unit 702, a second adjustment unit 703, and a training unit 704. The acquisition unit 701 is configured to acquire user selection information, first user consultation information from user feedback, and at least one second user consultation information output by the user consultation information determination model, wherein the user selection information indicates whether the user has selected user consultation information recommended by the system. The first adjustment unit 702 is configured to dynamically adjust the labels of the user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information, to obtain training samples, wherein the labels are the second target user consultation information among the at least one second user consultation information. The second adjustment unit 703 is configured to dynamically adjust the sample weights corresponding to the training samples, to obtain adjusted training samples. The training unit 704 is configured to retrain the user consultation information determination model based on the adjusted training samples, to obtain a trained user consultation information determination model.

[0117] In some optional implementations of some embodiments, the first adjustment unit 702 of the model training device 700 may be further configured to: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replace the label of the user data with the first user consultation information to obtain the training sample.

[0118] In some optional implementations of some embodiments, the second adjustment unit 703 of the model training device 700 may be further configured to: adjust the sample weights corresponding to the training samples to a value α to obtain the adjusted training samples, wherein the value α is a value greater than 1.

[0119] In some optional implementations of some embodiments, the first adjustment unit 702 of the model training device 700 may be further configured to: in response to the user selection information representing that the user selected user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replace the label of the user data with the first user consultation information to obtain the training sample.

[0120] In some optional implementations of some embodiments, the first adjustment unit 702 of the model training device 700 may be further configured to: in response to the user selection information characterizing the user selection of user consultation information recommended by the system, and the first user consultation information existing in at least one second user consultation information, determine the first user consultation information as the label of the user data, and obtain the training sample.

[0121] In some optional implementations of some embodiments, the second adjustment unit 703 of the model training device 700 may be further configured to: adjust the sample weights corresponding to the training samples to a value β to obtain the adjusted training samples, wherein the value β is a value between 0 and 1.

[0122] In some optional implementations of some embodiments, the first adjustment unit 702 of the model training device 700 may be further configured to: in response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and the first user consultation information being present in at least one second user consultation information, determine the user data and the second target user consultation information as the training samples.

[0123] In some optional implementations of some embodiments, the second adjustment unit 703 of the model training device 700 may be further configured to: keep the sample weights of the training samples unchanged, and determine the training samples as the adjusted training samples.

[0124] It is understandable that the units described in the device 700 are related to the reference. Figure 2 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 700 and the units contained therein, and will not be repeated here.

[0125] Continue to refer Figure 8 As an implementation of the above figures and methods, this disclosure provides some embodiments of a user consultation information determination device, which are similar to... Figure 5 Corresponding to the above-described method embodiments, the device can be specifically applied to various electronic devices.

[0126] like Figure 8 As shown, a user consultation information determination apparatus 800 in some embodiments includes an acquisition unit 801 and an input unit 802. The acquisition unit 801 is configured to, in response to receiving a request for determining user consultation information, acquire user data related to a pre-acquired user identifier of a target user. The input unit 802 is configured to input the user data into a pre-trained user consultation information determination model to obtain at least one piece of user consultation information.

[0127] It is understandable that the units described in the device 800 are related to the reference. Figure 5 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 800 and the units contained therein, and will not be repeated here.

[0128] The following is for reference. Figure 9 It shows a schematic diagram of the structure of an electronic device 900 suitable for implementing some embodiments of the present disclosure. Figure 9 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0129] like Figure 9 As shown, electronic device 900 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 901, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 902 or a program loaded from storage device 908 into random access memory (RAM) 903. RAM 903 also stores various programs and data required for the operation of electronic device 900. Processing device 901, ROM 902, and RAM 903 are interconnected via bus 904. Input / output (I / O) interface 905 is also connected to bus 904.

[0130] Typically, the following devices can be connected to I / O interface 905: input devices 906 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 907 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 908 including, for example, magnetic tapes, hard disks, etc.; and communication devices 909. Communication device 909 allows electronic device 900 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 9 An electronic device 900 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 9 Each box shown can represent a device or multiple devices as needed.

[0131] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 909, or installed from a storage device 908, or installed from a ROM 902. When the computer program is executed by the processing device 901, it performs the functions defined in the methods of some embodiments of this disclosure.

[0132] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0133] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0134] The aforementioned computer-readable medium may be included in the aforementioned device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: acquire user selection information, first user consultation information fed back by the user, and at least one second user consultation information output by a user consultation information determination model, wherein the user selection information indicates whether the user has selected user consultation information recommended by the system; dynamically adjust the labels of the user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples, wherein the labels are the second target user consultation information among the at least one second user consultation information; dynamically adjust the sample weights corresponding to the training samples to obtain adjusted training samples; and retrain the user consultation information determination model based on the adjusted training samples to obtain a trained user consultation information determination model. In response to receiving a request to determine user consultation information, the electronic device acquires user data related to the user identifier of the target user based on a pre-acquired user identifier; and inputs the user data into the pre-trained user consultation information determination model to obtain at least one user consultation information.

[0135] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0136] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0137] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including an acquisition unit, a first adjustment unit, a second adjustment unit, and a training unit. The names of these units do not necessarily limit the specific unit; for example, the acquisition unit may also be described as "a unit that acquires user selection information, first user consultation information from user feedback, and user consultation information to determine at least one second user consultation information output by the model."

[0138] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0139] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A model training method, comprising: The system obtains user selection information, first user consultation information based on user feedback, and at least one second user consultation information output by the model based on user consultation information, wherein the user selection information indicates whether the user selects user consultation information recommended by the system. Based on the user selection information, the first user consultation information, and the at least one second user consultation information, the labels of the user data input to the user consultation information determination model are dynamically adjusted to obtain training samples. The labels are the second target user consultation information in the at least one second user consultation information. This includes: in response to the user selection information indicating that the user has not selected the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, the labels of the user data are replaced with the first user consultation information to obtain the training samples. The sample weights corresponding to the training samples are dynamically adjusted to obtain the adjusted training samples, including: adjusting the sample weights corresponding to the training samples to a value α to obtain the adjusted training samples, wherein the value α is a value greater than 1. Based on the adjusted training samples, the user consultation information determination model is retrained to obtain the trained user consultation information determination model.

2. The method according to claim 1, wherein, The step of dynamically adjusting the labels of the user data input to the model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: In response to the user selection information indicating that the user has selected user consultation information recommended by the system, and the first user consultation information not existing in the at least one second user consultation information, the label of the user data is replaced with the first user consultation information to obtain the training sample.

3. The method according to claim 1, wherein, The step of dynamically adjusting the labels of the user data input to the model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: In response to the user selection information representing the user's selection of system-recommended user consultation information, and the first user consultation information existing in the at least one second user consultation information, the first user consultation information is determined as the label of the user data, and the training sample is obtained.

4. The method according to claim 3, wherein, The step of dynamically adjusting the sample weights corresponding to the training samples to obtain the adjusted training samples includes: The sample weights corresponding to the training samples are adjusted to the value β to obtain the adjusted training samples, wherein the value β is a value between 0 and 1.

5. The method according to claim 1, wherein, The step of dynamically adjusting the labels of the user data input to the model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples includes: In response to the user selection information indicating that the user did not select the user consultation information recommended by the system, and the first user consultation information being present in the at least one second user consultation information, the user data and the second target user consultation information are determined as the training samples.

6. The method according to claim 2 or 5, wherein, The step of dynamically adjusting the sample weights corresponding to the training samples to obtain the adjusted training samples includes: Keeping the sample weights of the training samples unchanged, the training samples are determined as the adjusted training samples.

7. The method according to claim 1, wherein, The at least one second user consultation information is output through the following steps: In response to receiving a request to determine user consultation information, user data associated with the obtained user identifier is obtained based on the acquired user identifier; The user data is input into a pre-trained user consultation information determination model to obtain at least one second user consultation information.

8. The method according to claim 1, wherein, The user consultation information determination model is trained through the following steps: Obtain target user data from the target terminal through relevant interfaces; The target user data is processed to obtain processed target user data; The processed target user data is stored in the database; In response to receiving a data acquisition request, the system retrieves a user dataset associated with the target business scenario from the database as a training sample for the user consultation information determination model, wherein the user dataset includes the target user data; The model is trained based on the training samples and the user consultation information.

9. The method according to claim 8, wherein, The database stores user datasets associated with each business scenario within each business scenario.

10. A method for determining user inquiry information, comprising: In response to receiving a request to determine user consultation information, user data related to the user identifier is obtained based on the user identifier of the target user obtained in advance; The user data is input into a pre-trained user consultation information determination model to obtain at least one user consultation information, wherein the user consultation information determination model is trained using the method described in any one of claims 1-9.

11. A model training device, comprising: The acquisition unit is configured to acquire user selection information, first user consultation information from user feedback, and at least one second user consultation information output by the user consultation information determination model, wherein the user selection information indicates whether the user selects the user consultation information recommended by the system. The first adjustment unit is configured to dynamically adjust the labels of user data input to the user consultation information determination model based on the user selection information, the first user consultation information, and the at least one second user consultation information to obtain training samples. The labels are second target user consultation information in the at least one second user consultation information. The adjustment unit includes: in response to the user selection information indicating that the user has not selected the user consultation information recommended by the system, and that the first user consultation information does not exist in the at least one second user consultation information, replacing the labels of the user data with the first user consultation information to obtain the training samples. The second adjustment unit is configured to dynamically adjust the sample weights corresponding to the training samples to obtain the adjusted training samples, including: adjusting the sample weights corresponding to the training samples to a value α to obtain the adjusted training samples, wherein the value α is a value greater than 1. The training unit is configured to retrain the user consultation information determination model based on the adjusted training samples to obtain a trained user consultation information determination model.

12. A user inquiry information determination device, comprising: The acquisition unit is configured to, in response to receiving a request for determining user consultation information, acquire user data related to the user identifier of the target user based on the user identifier of the target user acquired in advance; An input unit is configured to input the user data into a pre-trained user consultation information determination model to obtain at least one user consultation information, wherein the user consultation information determination model is trained using the method described in any one of claims 1-9.

13. An electronic device, comprising: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any one of claims 1-9.

14. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-9.

Images