A user service model training method, device and equipment
By receiving service anomaly information from clients and generating correction guidelines, users can correct the anomalies and train the model. This solves the problem of time-consuming and labor-intensive acquisition of training sample data, and achieves practical application fit of the model and service accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA CONSTRUCTION BANK
- Filing Date
- 2023-07-06
- Publication Date
- 2026-07-03
Smart Images
Figure CN116842387B_ABST
Abstract
Description
Technical Field
[0001] The embodiments in this specification relate to the field of artificial intelligence technology, and in particular to a user service model training method, apparatus, and device. Background Technology
[0002] With the development of artificial intelligence technology, models built using AI can provide personalized services based on user needs. For example, AI services can perform semantic analysis based on user needs, and then provide relevant information or services directly; they can also directly analyze image data and process or provide relevant image data based on user needs.
[0003] Currently, before applying AI services, it's often necessary to obtain training sample data to train the relevant models. This training sample data is typically collected from various sources or generated by specialists. However, the currently available training sample data requires significant time and effort to compile or collect, and it may deviate from real-world application scenarios, rendering models trained on this data ineffective in meeting practical needs. Therefore, a method for effectively acquiring training sample data for AI services is urgently needed. Summary of the Invention
[0004] The purpose of the embodiments in this specification is to provide a user service model training method, apparatus, and device to solve the problem of how to effectively obtain training sample data for artificial intelligence services.
[0005] To address the aforementioned technical problems, embodiments of this specification provide a user service model training method applied to a server. The method includes: receiving a service call request sent by a client; invoking a user service model to generate user service content corresponding to the service call request; feeding back the user service content to the client; upon receiving service exception information sent by the client, generating corresponding correction guidance information based on the service exception information; the service exception information indicating an anomaly in the user service content; the correction guidance information guiding the user to correct the anomaly in the user service content; sending the correction guidance information to the client, enabling the user to input correction information corresponding to the anomaly based on the client; acquiring the correction information and using the correction information to train the user service model.
[0006] In some implementations, the user service content includes image data; generating corresponding correction guidance information based on the service anomaly information includes: locating abnormal element information in the image data based on the service anomaly information; the abnormal element information includes at least one of the following: the position of the abnormal element in the image data, the content of the abnormal element, and the meaning information of the abnormal element.
[0007] Based on the above implementation method, the correction information includes at least one of the following: the correct position information of the abnormal element in the image data, the correction content of the abnormal element, and the correct meaning information of the abnormal element.
[0008] In some implementations, the correction information is also used to send the user service content and the correction information back to the client together after receiving the service call request sent by the client.
[0009] In some implementations, the correction guidance information includes a prompt; the prompt is used to prompt the user to correct the abnormal content.
[0010] In some implementations, obtaining the correction information and using the correction information to train the user service model includes: extracting labeled data corresponding to the user service content from the correction information; constructing training sample data based on the labeled data and the user service content; and retraining the user service model using the training sample data.
[0011] This specification also proposes a user service model training method applied to a client. The method includes: sending a service call request to a server; receiving and displaying user service content fed back by the server; the user service content being generated by the server using a user service model in response to the service call request; upon receiving a service exception instruction input by the user, sending service exception information to the server; the service exception information indicating an exception in the user service content; receiving and displaying correction guidance information fed back by the server; the correction guidance information guiding the user to correct the exception content in the user service content; obtaining correction information input by the user corresponding to the exception content; and sending the correction information to the server so that the server can use the correction information to train the user service model.
[0012] This specification also proposes a user service model training device applied to a server. The device includes: a service call request receiving module for receiving service call requests sent by a client; a user service content generation module for calling a user service model to generate user service content corresponding to the service call request; a user service content feedback module for feeding back the user service content to the client; a correction guidance information generation module for generating corresponding correction guidance information based on service exception information sent by the client upon receiving such information; the service exception information indicates that the user service content is abnormal; the correction guidance information guides the user to correct the abnormal content in the user service content; a correction guidance information sending module for sending the correction guidance information to the client so that the user can input correction information corresponding to the abnormal content based on the client; and a user service model training module for acquiring the correction information and training the user service model using the correction information.
[0013] This specification also proposes a user service model training device applied to a client. The device includes: a service call request receiving module for receiving service call requests sent by the client; a user service content generation module for calling a user service model to generate user service content corresponding to the service call request; a user service content feedback module for feeding back the user service content to the client; a correction guidance information generation module for generating corresponding correction guidance information based on service exception information sent by the client upon receiving such information; the service exception information indicates that the user service content is abnormal; the correction guidance information guides the user to correct the abnormal content in the user service content; a correction guidance information sending module for sending the correction guidance information to the client so that the user can input correction information corresponding to the abnormal content based on the client; and a user service model training module for acquiring the correction information and training the user service model using the correction information.
[0014] This specification also proposes an electronic device, including a memory and a processor; the memory is used to store computer programs / instructions; the processor is used to execute the computer programs / instructions to implement the steps of the user service model training method described above.
[0015] This specification also proposes a computer-readable storage medium storing a computer program / instruction thereon, which, when executed by a processor, implements the steps of the user service model training method described above.
[0016] This specification also proposes a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the user service model training method described above.
[0017] As can be seen from the technical solutions provided in the embodiments of this specification above, after the user service model is invoked to provide user service content to the client, if the user reports an anomaly in the service content based on the client, corresponding correction guidance information can be generated to enable the user to correct the anomaly based on the client. Correspondingly, after the user completes the correction of the anomaly, the corresponding correction information can be obtained. The obtained correction information can then be used to train the user service model. Through the above method, when providing services to users, the correction of anomalies can be completed by guiding the user, ensuring not only the effective correction of errors but also the effective utilization of correction information, ensuring the accuracy and effectiveness of subsequent user services. By using correction information as training data, not only is the training data used closely aligned with actual applications, enabling the optimized user service to better meet the needs of actual applications, but the acquisition of training data is also simplified, eliminating the need to collect or generate corresponding training data externally, saving time and manpower, and ensuring the practical application value. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a flowchart illustrating a user service model training method as described in this specification.
[0020] Figure 2 This is a flowchart illustrating a user service model training method as described in this specification.
[0021] Figure 3 This is a block diagram of a user service model training device according to an embodiment of this specification;
[0022] Figure 4 This is a block diagram of a user service model training device according to an embodiment of this specification. Detailed Implementation
[0023] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.
[0024] To address the aforementioned technical problems, this specification proposes a user service model training method. The execution entity of the user service model training method is a server. For example... Figure 1 As shown, the user service model training method may include the following specific implementation steps.
[0025] S110: Receive service call requests sent by clients.
[0026] A service call request is a request to invoke a corresponding service. When a user needs a service, they can generate and submit such a service call request through the relevant software or application on their client. For example, if a user wants to identify objects in an image, they can upload the image and submit a request to identify it.
[0027] In the embodiments of this specification, the corresponding services are those provided based on artificial intelligence technology. For example, the artificial intelligence service has a pre-set image recognition model. After receiving a user's image recognition request, the image can be input into the image recognition model, which then identifies the content in the image and obtains the corresponding recognition result. In practical applications, other types of user services can also be set up, such as performing semantic analysis and parsing on user input and providing corresponding parsed or replying content. The specific type of user service can be set according to the actual application situation and is not limited thereto.
[0028] The construction methods for user services, such as the training process of relevant models for user services, can also be set based on the needs of actual applications, and will not be elaborated here.
[0029] S120: Invoke the user service model to generate user service content corresponding to the service invocation request.
[0030] Upon receiving the service call request, the system can generate corresponding user service content based on pre-defined logical relationships using the associated user service model, and then feed it back to the client for the user to view or use. For example, when the service call request is used to determine an object in an image, the feedback user service content could be the object's location marker in the image, its name, and related information.
[0031] Specifically, the service call request can be pre-marked with the corresponding service type, and then the corresponding user service model can be obtained based on the service type. The specific process of calling the user service model can be configured according to the actual application situation, and will not be elaborated here.
[0032] Furthermore, user service content can also be generated based on relevant requirements in the service call request; for example, additional content can be provided according to the user's personalized needs. The specific information included in the user service content can be set according to the actual application situation, and there are no restrictions on this.
[0033] S130: Feedback the user service content to the client.
[0034] After generating the user service content, the user service content can be sent to the client so that the client can display the user service content to the user, thereby enabling the user to obtain query information corresponding to the service call request.
[0035] S140: Upon receiving service exception information sent by the client, generate corresponding correction guidance information based on the service exception information; the service exception information is used to indicate that there is an exception in the user service content; the correction guidance information is used to guide the user to correct the exception content in the user service content.
[0036] When the user service content is functioning normally, users can directly utilize the user service content, such as performing the next step based on the recognition results of objects in the image.
[0037] In some cases, defects in the user service itself may lead to certain anomalies in the user service content. For example, when recognizing images, there may be problems such as incorrect object location, incorrect object recognition, or incorrect retrieval of object-related information.
[0038] To address this issue, when a user determines that the received user service content is abnormal, they can report a service exception message through the client. This service exception message indicates that the user service content is abnormal.
[0039] Preferably, the service error information can also indicate the specific type of error. The specific type can be set according to the specific content of the user service. For example, when the user service is image recognition, the specific type of error could be location error, recognition error, or interpretation error. In practical applications, the specific type of error is determined based on the specific circumstances, and will not be elaborated further here.
[0040] For service anomaly information, corresponding correction guidance information can be generated. This correction guidance information can be used to instruct users to correct the abnormal content in their service content. Correction can involve marking the abnormal content or correcting it to obtain the correct user service content.
[0041] In cases where the specific content of the anomaly is uncertain, the correction guidance information may be used only to prompt the user to assist in correcting the anomaly.
[0042] When the type of anomaly is determined or the direction of correction for data in the format of user service content is clarified, the correction guidance information can be associated with the user service content. Specifically, it can be combined with service anomaly information to identify the abnormal content in the user service content and to mark the abnormal content accordingly. The added markings are used to guide users to correct the abnormal content, thereby ensuring the correctness and reliability of the correction results.
[0043] For example, when the user service content includes image data, generating correction guidance information can be based on locating abnormal element information in the image data according to the service anomaly information. The abnormal element information includes at least one of the following: the location of the abnormal element in the image data, the content of the abnormal element, and the meaning information of the abnormal element. The location of the abnormal element in the image data can be used to determine the location of the incorrect location in the case of a positioning error; the content of the abnormal element can be the incorrectly identified image content; and the meaning information of the abnormal element can be the incorrect interpretation of an object in the image. In practical applications, other types of abnormal element information can also be set, and there are no restrictions on this.
[0044] After identifying the abnormal element information, corresponding correction guidance information can be generated to instruct the correction of the abnormal element information, thereby ensuring that users can complete the correction of abnormal parts in user service content through effective means.
[0045] S150: Send the correction guidance information to the client so that the user can input correction information corresponding to the abnormal content based on the client.
[0046] After generating the correction guidance information, it can be sent to the client, allowing the user to input correction information corresponding to the abnormal content based on the client's input. Since the correction guidance information instructs the user to correct the service content, the user, upon receiving the guidance, can correct the abnormal content based on the relevant prompts and the identification of the corresponding error in the feedback service exception information.
[0047] The specific correction process could involve setting up a corresponding input interface on the client side, allowing users to mark and annotate abnormal content. Preferably, with technical support, users can also be provided with relevant technical means to directly correct abnormal content into modified content.
[0048] Correction information refers to the information entered by the user regarding abnormal content. As described in the example above, it can be a comment on the abnormal content or the corrected user service content. In some examples, when the user service content is image data, the correction information may be at least one of the following: the correct location information of the abnormal element in the image data, the corrected content of the abnormal element, or the correct meaning information of the abnormal element. The specific type of correction information can be determined according to the actual situation and is not limited thereto.
[0049] S160: Obtain the correction information and use the correction information to train the user service model.
[0050] After the user inputs correction information, the correction information can be retrieved. Retrieving the correction information can be done either by the user actively submitting the correction information after input, or by directly retrieving the content of the correction information input by the user from the client through a corresponding interface.
[0051] Accordingly, after obtaining the correction information, it can be saved as corresponding to the user service content. Then, upon receiving the same service call request, the user service content and the correction information can be sent back to the client together, ensuring that the user receives the correct service content.
[0052] In some cases, the user service content can be processed by correcting the information to generate the correct user service content. Then, upon receiving a service call request, the processed user service content is directly fed back.
[0053] Once the correction information is obtained, it can be used to train the user service model. Because the correction information not only indicates anomalies in the user service content but also includes corrections to those anomalies, it can be used not only to identify defects in the user service model but also to determine the optimization direction for its practical application. The specific optimization process can be configured according to the type of user service.
[0054] Specifically, the correction information can be used as training sample data corresponding to the user service model to retrain the model. Since the correction information includes not only the errors that the model will actually produce, but also the corresponding annotations and corrected information, the model trained with this training sample data has more practical application value.
[0055] In some implementations, retraining the model can begin by extracting labeled data corresponding to the user service content from the correction information, then constructing training sample data based on the labeled data and the user service content, and finally retraining the user service model using the training sample data. The labeled data can be data on the annotation effectiveness of abnormal content in the user service content, thereby enabling effective annotation of the user service content.
[0056] By extracting labeled data based on correction information, it is possible to effectively indicate anomalies in the model during application, thereby further optimizing the training effect of the model.
[0057] The specific process for training the model can be set according to the actual application, and will not be elaborated here.
[0058] Based on the above embodiments, after the method calls the user service model to provide user service content to the client, if the user reports an anomaly in the service content based on the client, corresponding correction guidance information can be generated to enable the user to correct the anomaly based on the client. Correspondingly, after the user completes the correction of the anomaly, the corresponding correction information can be obtained. This correction information can then be used to train the user service model. Through this method, when providing services to users, the correction of anomalies can be completed by guiding the user, ensuring not only the effective correction of errors but also the effective utilization of correction information, ensuring the accuracy and effectiveness of subsequent user services. By using the correction information as training data, not only is the training data used closely aligned with actual applications, allowing the optimized user service to better meet the needs of practical applications, but the acquisition of training data is also simplified, eliminating the need to collect or generate corresponding training data externally, saving time and manpower, and ensuring the practical application value.
[0059] based on Figure 1 Regarding the corresponding user service model training method, this specification also proposes a user service model training method in its embodiments. The executing entity of the user service model training method is the corresponding client. For example... Figure 2 As shown, the user service model training method may include the following specific implementation steps.
[0060] S210: Send a service call request to the server.
[0061] A service call request is a request to invoke a corresponding service. When a user needs a service, they can generate and submit such a service call request through the relevant software or application on their client. For example, if a user wants to identify objects in an image, they can upload the image and submit a request to identify it.
[0062] In the embodiments of this specification, the corresponding services are those provided based on artificial intelligence technology. For example, the artificial intelligence service has a pre-set image recognition model. After receiving a user's image recognition request, the image can be input into the image recognition model, which then identifies the content in the image and obtains the corresponding recognition result. In practical applications, other types of user services can also be set up, such as performing semantic analysis and parsing on user input and providing corresponding parsed or replying content. The specific type of user service can be set according to the actual application situation and is not limited thereto.
[0063] The construction methods for user services, such as the training process of relevant models for user services, can also be set based on the needs of actual applications, and will not be elaborated here.
[0064] S220: Receive and display the user service content fed back by the server; the user service content is generated by the server using the user service model in response to the service call request.
[0065] Upon receiving the service call request, the server can invoke the corresponding process to generate user service content corresponding to the service call request using the associated user service, and then feed it back to the client for the user to view or use. For example, when the service call request is used to determine an object in an image, the feedback user service content could be the object's location marker in the image, as well as the object's name and related information.
[0066] Furthermore, user service content can also be generated based on relevant requirements in the service call request; for example, additional content can be provided according to the user's personalized needs. The specific information included in the user service content can be set according to the actual application situation, and there are no restrictions on this.
[0067] User service content typically includes displayable information, such as text, images, and annotations. This user service content can be directly displayed to the user through the client.
[0068] When the user service content is functioning normally, users can directly utilize the user service content, such as performing the next step based on the recognition results of objects in the image.
[0069] S230: Upon receiving a service exception instruction from the user, send service exception information to the server; the service exception information is used to indicate that there is an exception in the user service content.
[0070] In some cases, defects in the user service itself may lead to certain anomalies in the user service content. For example, when recognizing images, there may be problems such as incorrect object location, incorrect object recognition, or incorrect retrieval of object-related information.
[0071] To address this issue, when a user determines that the received user service content is abnormal, they can input a service exception command through the client. This service exception command indicates that the user service content is abnormal. Specifically, for example, a corresponding exception feedback button can be set on the client's display interface. When the user notices an anomaly in the user service content, they can directly click the exception feedback button. Correspondingly, based on the feedback from the exception feedback button, the relevant program generates a corresponding service exception command.
[0072] Upon detecting a service exception command, the client can generate corresponding service exception information. This information is sent to the server to indicate that an exception exists in the user service content.
[0073] Preferably, the service error information can also indicate the specific type of error. The specific type can be set according to the specific content of the user service. For example, when the user service is image recognition, the specific type of error could be location error, recognition error, or interpretation error. In practical applications, the specific type of error is determined based on the specific circumstances, and will not be elaborated further here.
[0074] S240: Receive and display correction guidance information fed back by the server; the correction guidance information is used to guide the user to correct the abnormal content in the user service content.
[0075] In response to service anomaly information, the server can generate corresponding correction guidance information. This correction guidance information can be used to instruct users to correct the abnormal content in their service offerings. Correction can involve marking the abnormal content or revising it to obtain the correct service offering.
[0076] In cases where the specific content of the anomaly is uncertain, the correction guidance information may be used only to prompt the user to assist in correcting the anomaly.
[0077] When the type of anomaly is determined or the direction of correction for data in the format of user service content is clarified, the correction guidance information can be associated with the user service content. Specifically, it can be combined with service anomaly information to identify the abnormal content in the user service content and to mark the abnormal content accordingly. The added markings are used to guide users to correct the abnormal content, thereby ensuring the correctness and reliability of the correction results.
[0078] For example, when the user service content includes image data, generating correction guidance information can be based on locating abnormal element information in the image data according to the service anomaly information. The abnormal element information includes at least one of the following: the location of the abnormal element in the image data, the content of the abnormal element, and the meaning information of the abnormal element. The location of the abnormal element in the image data can be used to determine the location of the incorrect location in the case of a positioning error; the content of the abnormal element can be the incorrectly identified image content; and the meaning information of the abnormal element can be the incorrect interpretation of an object in the image. In practical applications, other types of abnormal element information can also be set, and there are no restrictions on this.
[0079] After identifying the abnormal element information, corresponding correction guidance information can be generated to instruct the correction of the abnormal element information, thereby ensuring that users can complete the correction of abnormal parts in user service content through effective means.
[0080] After providing correction guidance, this guidance can be sent to the client, allowing the user to input correction information corresponding to the anomaly. Since the correction guidance instructs the user to correct the service content, the user, upon receiving the guidance, can correct the anomaly based on the relevant prompts and the identification of the corresponding error in the returned service anomaly information.
[0081] The specific correction process could involve setting up a corresponding input interface on the client side, allowing users to mark and annotate abnormal content. Preferably, with technical support, users can also be provided with relevant technical means to directly correct abnormal content into modified content.
[0082] S250: Obtain correction information corresponding to the abnormal content input by the user.
[0083] Correction information refers to the information entered by the user regarding abnormal content. As described in the example above, it can be a comment on the abnormal content or the corrected user service content. In some examples, when the user service content is image data, the correction information may be at least one of the following: the correct location information of the abnormal element in the image data, the corrected content of the abnormal element, or the correct meaning information of the abnormal element. The specific type of correction information can be determined according to the actual situation and is not limited thereto.
[0084] After the user inputs correction information, the correction information can be retrieved. Retrieving the correction information can be done either by the user actively submitting the correction information after input, or by directly retrieving the content of the correction information input by the user from the client through a corresponding interface.
[0085] Accordingly, after obtaining the correction information, it can be saved as corresponding to the user service content. Then, upon receiving the same service call request, the user service content and the correction information can be sent back to the client together, ensuring that the user receives the correct service content.
[0086] In some cases, the user service content can be processed by correcting the information to generate the correct user service content. Then, upon receiving a service call request, the processed user service content is directly fed back.
[0087] In some implementations, the user service is the corresponding service content generated based on a user service model. The user service model can be a corresponding machine learning model. After pre-building the model architecture, the model is trained using sample data to ensure that the model parameters meet application requirements, and then the model is applied to the corresponding user service. For example, the user service model can be an image recognition model capable of outputting corresponding image recognition results based on the image input by the user.
[0088] S260: The correction information is sent to the server so that the server can use the correction information to train the user service model.
[0089] Once the correction information is obtained, it can be used to train the user service model. Because the correction information not only indicates anomalies in the user service content but also includes corrections to those anomalies, it can be used not only to identify defects in the user service model but also to determine the optimization direction for its practical application. The specific optimization process can be configured according to the type of user service.
[0090] Specifically, the correction information can be used as training sample data corresponding to the user service model to retrain the model. Since the correction information includes not only the errors that the model will actually produce, but also the corresponding annotations and corrected information, the model trained with this training sample data has more practical application value.
[0091] In some implementations, retraining the model can begin by extracting labeled data corresponding to the user service content from the correction information, then constructing training sample data based on the labeled data and the user service content, and finally retraining the user service model using the training sample data. The labeled data can be data on the annotation effectiveness of abnormal content in the user service content, thereby enabling effective annotation of the user service content.
[0092] By extracting labeled data based on correction information, it is possible to effectively indicate anomalies in the model during application, thereby further optimizing the training effect of the model.
[0093] The specific process for training the model can be set according to the actual application, and will not be elaborated here.
[0094] Based on the above embodiments, after the method calls the user service model to provide user service content to the client, if the user reports an anomaly in the service content based on the client, corresponding correction guidance information can be generated to enable the user to correct the anomaly based on the client. Correspondingly, after the user completes the correction of the anomaly, the corresponding correction information can be obtained. This correction information can then be used to train the user service model. Through this method, when providing services to users, the correction of anomalies can be completed by guiding the user, ensuring not only the effective correction of errors but also the effective utilization of correction information, ensuring the accuracy and effectiveness of subsequent user services. By using the correction information as training data, not only is the training data used closely aligned with actual applications, allowing the optimized user service to better meet the needs of practical applications, but the acquisition of training data is also simplified, eliminating the need to collect or generate corresponding training data externally, saving time and manpower, and ensuring the practical application value.
[0095] based on Figure 1 This specification describes a user service model training device, based on an embodiment of the user service model training method. The user service model training device can be installed on a server. For example... Figure 3 As shown, the user service model training device includes the following modules.
[0096] The service call request receiving module 310 is used to receive service call requests sent by the client.
[0097] The user service content generation module 320 is used to call the user service model to generate user service content corresponding to the service call request.
[0098] The user service content feedback module 330 is used to feed back the user service content to the client.
[0099] The correction guidance information generation module 340 is used to generate corresponding correction guidance information based on the service exception information sent by the client when receiving service exception information; the service exception information is used to indicate that there is an exception in the user service content; the correction guidance information is used to guide the user to correct the exception content in the user service content.
[0100] The correction guidance information sending module 350 is used to send the correction guidance information to the client so that the user can input correction information corresponding to the abnormal content based on the client.
[0101] User service model training module 360 is used to acquire the correction information and train the user service model using the correction information.
[0102] based on Figure 2 This specification describes a user service model training device according to an embodiment of the corresponding user service model training method. The user service model training device can be installed on the client side. For example... Figure 4 As shown, the user service model training device includes the following modules.
[0103] The service call request sending module 410 is used to send service call requests to the server.
[0104] The user service content display module 420 is used to receive and display the user service content fed back by the server; the user service content is generated by the server using the user service model in response to the service call request.
[0105] The service exception information sending module 430 is used to send service exception information to the server when a user inputs a service exception command; the service exception information is used to indicate that the user service content is abnormal.
[0106] The correction guidance information display module 440 is used to receive and display correction guidance information fed back by the server; the correction guidance information is used to guide users to correct abnormal content in the user service content.
[0107] The correction information acquisition module 450 is used to acquire correction information corresponding to the abnormal content input by the user.
[0108] The correction information sending module 460 is used to send the correction information to the server so that the server can use the correction information to train the user service model.
[0109] based on Figure 1 The corresponding user service optimization model training method is described in this specification, and an electronic device is provided in the embodiments. The electronic device can be equivalent to the aforementioned server or client. The electronic device may include a memory and a processor.
[0110] In this embodiment, the memory can be implemented in any suitable manner. For example, the memory can be a read-only memory, a hard disk drive, a solid-state drive, or a USB flash drive, etc. The memory can be used to store computer programs / instructions.
[0111] In this embodiment, the processor can be implemented in any suitable manner. For example, the processor can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc. When applied to a server, the processor can execute the computer program instructions to implement, for example, Figure 1 The corresponding user service model training method; when applied to the client, the computer program instructions can be executed to achieve the following: Figure 2 The corresponding user service model training method.
[0112] This specification provides a computer-readable storage medium storing a computer program / instructions thereon. The computer-readable storage medium can be read by a processor via the internal bus of a device, and the processor can then implement the program instructions in the computer-readable storage medium.
[0113] In this embodiment, the computer-readable storage medium can be implemented in any suitable manner. The computer-readable storage medium includes, but is not limited to, random access memory (RAM), read-only memory (ROM), cache, hard disk drive (HDD), memory card, etc. The computer storage medium stores computer program instructions. When the computer-readable storage medium is disposed on a server, the computer program instructions, when executed, implement this specification. Figure 1 The corresponding user service model training method's program instructions or modules; when the computer-readable storage medium is set on the client, the computer program instructions are executed to implement this specification. Figure 2 The corresponding user service model training method's program instructions or modules.
[0114] This specification also provides a computer program product, including a computer program / instructions. The computer program product may be a program written in a corresponding computer programming language, stored in a corresponding storage device in a program manner, and can be transmitted via a computer network. The computer program product can be executed by a processor. In this specification embodiment, when the computer program product is executed on a server, it implements the following... Figure 1 The corresponding user service model training method's program instructions or modules, or those implemented when set up and executed on the client. Figure 2The corresponding user service model training method's program instructions or modules.
[0115] It should be noted that the above-mentioned user service optimization model training methods, devices and equipment can be applied to the field of artificial intelligence technology, or to other technical fields other than artificial intelligence technology, without any restrictions.
[0116] Furthermore, all operations related to the acquisition, processing, storage, and forwarding of data in the aforementioned user service model training methods, devices, and equipment comply with relevant national laws and regulations.
[0117] Although the process described above includes multiple operations that occur in a specific order, it should be clearly understood that these processes may include more or fewer operations, which may be executed sequentially or in parallel (e.g., using parallel processors or a multithreaded environment).
[0118] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0121] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0122] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0123] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape storage, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0124] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of computer program products implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0125] The embodiments described in this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0126] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, system embodiments are basically similar to method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. In the description of this specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the embodiments in this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples.
[0127] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A user service model training method, applied to a server; characterized in that, The method includes: Receive service call requests sent by clients; The user service model is invoked to generate user service content corresponding to the service invocation request; The user service content is then fed back to the client. Upon receiving service exception information from the client, corresponding correction guidance information is generated based on the service exception information; the service exception information is used to indicate that there is an exception in the user service content; the correction guidance information is used to guide the user to correct the exception content in the user service content. The correction guidance information is sent to the client so that the user can input correction information corresponding to the abnormal content based on the client. Obtain the correction information, and establish and save the correspondence between the service call request and the correction information; Upon receiving subsequent identical service call requests, the correct user service content after processing is fed back to the client according to the aforementioned correspondence, so as to achieve immediate correction of abnormal content; The annotation data corresponding to the user service content is extracted from the correction information, and training sample data is constructed based on the annotation data and the user service content; The user service model is trained using the training sample data; The step of generating corresponding correction guidance information based on the service anomaly information includes: The specific type of error is determined based on the service exception information; wherein, the specific type includes at least one of location error, identification error, and meaning interpretation error; Based on the specific type, identify the abnormal content in the user service content, and add annotations to the abnormal content to guide the user to correct it, so as to generate correction guidance information.
2. The method as described in claim 1, characterized in that, The user service content includes image data; the step of generating corresponding correction guidance information based on the service anomaly information includes: Based on the service anomaly information, abnormal element information is located in the image data; the abnormal element information includes at least one of the following: the location of the abnormal element in the image data, the content of the abnormal element, and the meaning of the abnormal element. Correction guidance information is generated based on the abnormal element information; the correction guidance information is used to instruct the abnormal element information to be corrected.
3. The method as described in claim 2, characterized in that, The correction information includes at least one of the following: the correct location information of the abnormal element in the image data, the correction content of the abnormal element, and the correct meaning information of the abnormal element.
4. The method as described in claim 1, characterized in that, The correction information is also used to send the user service content and the correction information back to the client together after receiving the service call request sent by the client.
5. The method as described in claim 1, characterized in that, The correction guidance information includes a prompt; the prompt is used to remind the user to correct the abnormal content.
6. A user service model training method, applied to a client; characterized in that, The method includes: Send a service call request to the server; Receive and display user service content fed back by the server; the user service content is the original content generated by the server in response to the service call request using the user service model, or the correct content after processing determined by the server according to the correspondence between the saved service call request and correction information, so as to realize the immediate correction of abnormal content; Upon receiving a service error command input by the user, a service error message is sent to the server. The service error message indicates that there is an error in the user's service content, so that the server can determine the specific type of error based on the service error message. The specific type includes at least one of location error, identification error, and meaning interpretation error. Receive and display correction guidance information from the server; the correction guidance information is generated by the server by adding correction guidance labels to the abnormal content according to the determined specific type, and is used to guide the user to correct the abnormal content in the user service content; Obtain correction information corresponding to the abnormal content input by the user; The correction information is sent to the server so that the server can establish and save the correspondence between the service call request and the correction information, and the server can extract labeled data from the correction information to construct training sample data for training the user service model.
7. A user service model training device, applied to a server; characterized in that, The device includes: The service call request receiving module is used to receive service call requests sent by the client; The user service content generation module is used to call the user service model to generate user service content corresponding to the service call request. The user service content feedback module is used to feed back the user service content to the client. The correction guidance information generation module is used to generate corresponding correction guidance information based on the service exception information sent by the client when receiving service exception information; the service exception information is used to indicate that there is an exception in the user service content; the correction guidance information is used to guide the user to correct the exception content in the user service content. A correction guidance information sending module is used to send the correction guidance information to the client so that the user can input correction information corresponding to the abnormal content based on the client; The user service model training module is used to obtain the correction information and establish and save the correspondence between the service call request and the correction information; Upon receiving subsequent identical service call requests, the correct user service content after processing is fed back to the client according to the aforementioned correspondence, so as to achieve immediate correction of abnormal content; The annotation data corresponding to the user service content is extracted from the correction information, and training sample data is constructed based on the annotation data and the user service content; The user service model is trained using the training sample data; The step of generating corresponding correction guidance information based on the service anomaly information includes: The specific type of error is determined based on the service exception information; wherein, the specific type includes at least one of location error, identification error, and meaning interpretation error; Based on the specific type, identify the abnormal content in the user service content, and add annotations to the abnormal content to guide the user to correct it, so as to generate correction guidance information.
8. A user service model training device, characterized in that, Applied to a client; characterized in that the device comprises: The service call request sending module is used to send service call requests to the server; The user service content display module is used to receive and display user service content fed back by the server. The user service content is the original content generated by the server in response to the service call request using the user service model, or the correct content after processing determined by the server according to the correspondence between the saved service call request and correction information, so as to realize the real-time correction of abnormal content. The service exception information sending module is used to send service exception information to the server when a user inputs a service exception command; the service exception information is used to indicate that there is an exception in the user's service content, so that the server can determine the specific type of error based on the service exception information; the specific type includes at least one of location error, identification error, and meaning interpretation error; The correction guidance information display module is used to receive and display correction guidance information fed back by the server; the correction guidance information is generated by the server by adding guidance correction labels to the abnormal content according to the determined specific type, and is used to guide users to correct the abnormal content in the user service content. The correction information acquisition module is used to acquire correction information corresponding to the abnormal content input by the user; The correction information sending module is used to send the correction information to the server, so that the server can establish and save the correspondence between the service call request and the correction information, and the server can extract labeled data from the correction information to construct training sample data for training the user service model.
9. An electronic device comprising a memory and a processor; characterized in that, The memory is used to store computer programs / instructions; the processor is used to execute the computer programs / instructions to implement the steps of the method as described in any one of claims 1-6.
10. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1-6.
11. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1-6.