Work order settlement generation method and device, storage medium and electronic equipment
By acquiring business data and utilizing prompt word prediction models and work order summary generation models, the system addresses the issues of insufficient accuracy and intelligence in work order summary generation in real-time interactive scenarios, achieving efficient and accurate work order summary generation.
Patent Information
- Application Number
- CN202511069164.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-31
- Publication Date
- 2025-11-07
AI Technical Summary
Modern customer service systems cannot accurately generate work order summaries in real-time interactive scenarios, resulting in high risks of information omissions, long customer waiting times, and high labor costs. Existing automation solutions cannot overcome the limitations of insufficient intelligence and scenario adaptability.
By acquiring business data and utilizing a pre-trained prompt word prediction model and work order summary generation model, a work order summary is generated. This includes acquiring call data, work order information, and environmental information, using the prompt word prediction model to generate business scenario prompt words, combining them with general prompt words, inputting them into the work order summary generation model, and finally generating an accurate work order summary.
It enables accurate work order summary generation in real-time interactive scenarios, improving the level of intelligence and scenario adaptability, and reducing the risk of information omission and labor costs.
Smart Images
Figure CN120911484A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of computers, and in particular relates to a work order summary generation method and device, a storage medium and an electronic device. BACKGROUND
[0002] Modern customer service systems face the dual challenges of diversified service needs and efficiency improvement, and need to balance response accuracy and service experience optimization. Traditional work order processing relies on manual call recording transcription and form filling, which has inherent defects such as high risk of information omission, long customer waiting time, and high labor costs.
[0003] Although automatic speech recognition and natural language processing technologies have been applied to form filling assistance, existing solutions are still limited to historical form information matching and static content recommendation, and cannot break through the automation bottleneck in real-time interaction scenarios, resulting in a serious lack of intelligence and scene adaptability, and cannot accurately generate work order summaries. SUMMARY
[0004] The technical problem to be solved by the present application is to provide a work order summary generation method, device, storage medium and electronic device, which can accurately generate a work order summary. The specific scheme is as follows:
[0005] A work order summary generation method comprises:
[0006] In the case of receiving a work order generation instruction, obtaining business data corresponding to the work order generation instruction; the business data includes call data, work order information and environment information;
[0007] Inputting the business data into a pre-trained prompt word prediction model to obtain a business scenario prompt word;
[0008] Combining the business scenario prompt word with a general prompt word to obtain a target prompt word; the general prompt word includes a dialogue rule;
[0009] Inputting the target prompt word into a work order summary generation model to generate a work order summary.
[0010] The above method can optionally include a training process of the prompt word model, comprising:
[0011] Obtaining a training data set and an initial prediction model to be trained; the training data set includes a plurality of training samples, each training sample includes historical business data and a sample label, and the sample label includes a reference prompt word;
[0012] Determining a target training sample to be used for training in each training sample;
[0013] inputting the target training sample into an initial prediction model to obtain a prediction prompt word output by the initial prediction model;
[0014] updating model parameters of the initial prompt word according to a difference between the prediction prompt word and the reference prompt word;
[0015] in a case where the updated initial prediction model does not meet a training completion condition, returning to performing the step of determining a target training sample to be currently used for training in each of the training samples;
[0016] in a case where the updated initial prediction model does not meet a training completion condition, taking the updated initial prediction model as a trained prompt word prediction model.
[0017] The method can further include:
[0018] obtaining a general prompt word in a preset configuration file, wherein the general prompt word includes a dialogue rule, a model random number, and a work order summary generation requirement;
[0019] splicing the business scenario prompt word and the general prompt word to obtain a target prompt word.
[0020] The business scenario prompt word can include at least one of the following:
[0021] dialogue intent information, a business processing flow description, a customer feature label, a historical interaction record, and date feature information;
[0022] The date feature information can include at least one of the following: a holiday date, a financial business interest settlement date, a bill generation date, and a rate adjustment effective date.
[0023] The method can further include:
[0024] outputting the work order summary to enable a user to review the work order summary;
[0025] in a case where a work order submitted by the user is detected, detecting whether the work order summary deviates from the work order submitted by the user;
[0026] if the work order summary deviates from the work order submitted by the user, optimizing the work order summary generation model according to the deviation between the work order summary and the work order submitted by the user.
[0027] A work order summary generation device can include:
[0028] The receiving unit is configured to, in response to receiving a work order generation instruction, acquire service data corresponding to the work order generation instruction; the service data comprises call data, work order information and environment information;
[0029] The prediction unit is configured to input the service data into a pre-trained prompt prediction model to obtain a service scenario prompt;
[0030] The execution unit is configured to combine the service scenario prompt with a general prompt to obtain a target prompt; the general prompt comprises a dialogue rule;
[0031] The generation unit is configured to input the target prompt into a work order summary generation model to generate a work order summary.
[0032] The device described above, optionally, the prediction unit comprises:
[0033] The first acquisition subunit is configured to acquire a training data set and an initial prediction model to be trained; the training data set comprises a plurality of training samples, each of the training samples comprises historical service data and a sample label, and the sample label comprises a reference prompt;
[0034] The first determination subunit is configured to determine a target training sample to be currently used for training in each of the training samples;
[0035] The output subunit is configured to input the target training sample into the initial prediction model to obtain a prediction prompt output by the initial prediction model;
[0036] The update subunit is configured to update model parameters of the initial prediction model according to a difference degree between the prediction prompt and the reference prompt;
[0037] The second determination subunit is configured to, in a case where the updated initial prediction model does not meet a training completion condition, return to perform the step of determining a target training sample to be currently used for training in each of the training samples;
[0038] In a case where the updated initial prediction model does not meet the training completion condition, the updated initial prediction model is taken as a trained prompt prediction model.
[0039] The device described above, optionally, the execution unit comprises:
[0040] The second acquisition subunit is configured to acquire a general prompt in a preset configuration file; the general prompt comprises a dialogue rule, a model random number and a work order summary generation requirement;
[0041] The splicing sub-unit is configured to splice the business scenario prompt word and the general prompt word to obtain a target prompt word.
[0042] A storage medium includes stored instructions, wherein the instructions, when executed, control a device in which the storage medium is located to perform the work order summary generation method as described above.
[0043] An electronic device includes a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by one or more processors to perform the work order summary generation method as described above.
[0044] The application provides a work order summary generation method, device, storage medium and electronic device. The method comprises: in the case of receiving a work order generation instruction, obtaining business data corresponding to the work order generation instruction; the business data comprises call data, work order information and environment information; inputting the business data into a pre-trained prompt word prediction model to obtain a business scenario prompt word; combining the business scenario prompt word with a general prompt word to obtain a target prompt word; the general prompt word comprises a dialogue rule; inputting the target prompt word into a work order summary generation model to generate a work order summary. The method provided by the application can accurately generate a work order summary. BRIEF DESCRIPTION OF DRAWINGS
[0045] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of the provided drawings.
[0046] Figure 1 A method flowchart of a work order summary generation method provided by the application;
[0047] Figure 2 A structural schematic diagram of an intelligent work order generation system provided by the application;
[0048] Figure 3 A data processing flowchart provided by the application;
[0049] Figure 4 Another data processing flowchart provided by the application;
[0050] Figure 5 Another data processing flowchart provided by the application;
[0051] Figure 6 Another data processing flowchart provided by the application;
[0052] Figure 7 Another data processing flowchart provided for the present application;
[0053] Figure 8 A structural schematic diagram of a work order summary generation device provided for the present application;
[0054] Figure 9 A structural schematic diagram of an electronic device provided for the present application. DETAILED DESCRIPTION
[0055] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.
[0056] In the present application, the term "comprising", "containing" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the sentence "including a…" does not exclude the presence of other identical elements in the process, method, article or device including the element.
[0057] The embodiments of the present application provide a work order summary generation method, which can be applied to an electronic device, and the electronic device can be a smart phone, a tablet device, a smart wearable device, etc. The method flowchart of the method is as shown in Figure 1 The method specifically includes the following steps.
[0058] S101: In the case of receiving a work order generation instruction, acquiring business data corresponding to the work order generation instruction; the business data includes call data, work order information and environment information.
[0059] In the present embodiment, the call data can include at least one of call date, call duration and conversation text information.
[0060] Optionally, the work order information can include at least one of business attribute, traffic type, subdivision type, note information, etc.
[0061] Optionally, the environment information can include at least one of holiday, interest settlement day, billing day, housing interest rate reduction day.
[0062] S102: input the service data into a pre-trained prompt prediction model to obtain a service scenario prompt.
[0063] Optionally, the service scenario prompt comprises at least one of the following:
[0064] dialog intent information, service processing flow description, customer feature label, historical interaction record, and date feature information.
[0065] The date feature information comprises at least one of the following: holiday date, financial business interest settlement date, bill generation date, and interest rate adjustment effective date.
[0066] In some embodiments provided in the present application, optionally, the training process of the prompt model comprises:
[0067] obtaining a training data set and an initial prediction model to be trained; the training data set comprises a plurality of training samples, each of the training samples comprises historical service data and a sample label, and the sample label comprises a reference prompt;
[0068] determining a target training sample to be currently used for training in each of the training samples;
[0069] inputting the target training sample into the initial prediction model to obtain a predicted prompt output by the initial prediction model;
[0070] updating model parameters of the initial prompt according to a difference degree between the predicted prompt and the reference prompt;
[0071] if the updated initial prediction model does not meet a training completion condition, returning to the step of determining a target training sample to be currently used for training in each of the training samples;
[0072] if the updated initial prediction model does not meet a training completion condition, taking the updated initial prediction model as a trained prompt prediction model.
[0073] In the present embodiment, the difference degree between the predicted prompt and the reference prompt can be calculated by a loss function.
[0074] Optionally, the training completion condition can be that the prediction accuracy of the initial prediction model is greater than an accuracy threshold, or the number of training times is greater than a training times threshold, etc.
[0075] S103: combining the service scenario prompt with a general prompt to obtain a target prompt; the general prompt comprises a dialog rule.
[0076] Optionally, the general prompt word can further include a work order summary generation requirement and a model random number.
[0077] In some embodiments provided in the present application, optionally, the combining the business scenario prompt word with the general prompt word to obtain a target prompt word comprises:
[0078] obtaining a general prompt word in a preset configuration file; the general prompt word includes a dialogue rule, a model random number and a work order summary generation requirement;
[0079] splicing the business scenario prompt word and the general prompt word to obtain a target prompt word.
[0080] S104: inputting the target prompt word into a work order summary generation model to generate a work order summary.
[0081] In the present embodiment, the work order summary generation model can be a generative model, and the work order summary can be a closed-loop record generated after the work order is closed.
[0082] By applying the method provided in the present application, the work order summary can be accurately generated.
[0083] In some embodiments provided in the present application, optionally, after the inputting the target prompt word into the work order summary generation model to generate the work order summary, the method further comprises:
[0084] outputting the work order summary to enable a user to audit the work order summary;
[0085] in the case of detecting a work order submitted by a user, detecting whether there is a deviation between the work order summary and the work order submitted by the user;
[0086] if there is a deviation between the work order summary and the work order submitted by the user, optimizing the work order summary generation model according to the deviation between the work order summary and the work order submitted by the user.
[0087] Referring to Figure 2 A structural schematic diagram of an intelligent work order generation system provided in the present application includes a user interaction unit, a data acquisition and processing unit, a business prompt word prediction unit, a prompt word management unit and an intelligent generation model intelligent interaction and feedback optimization unit.
[0088] The user interaction unit is used for receiving multi-modal consultation, complaint, transaction support and other interactive data input by a client, and includes a data receiver, a data identifier and a data display. The interactive data processing flow of the user interaction unit is as shown in Figure 3As shown, in the case of customer incoming line, the agent clicks to answer, the data receiver receives the interaction data of the agent and the customer, which is divided into voice data and text data. When the interaction data type is voice data, the obtained voice stream data is sent to the data recognizer for processing into text data, and then sent to the data display unit after structured processing. When the interaction data type is text data, the obtained text data of the agent and the customer conversation can be directly sent to the data display unit for data display after structured processing.
[0089] The data acquisition and processing unit is used for collecting, cleaning and feature engineering of multi-source heterogeneous data. The data processing flow of the data acquisition and processing unit is as shown in Figure 4 As shown, first, the incoming agent call basic data such as call date, call duration, conversation text information (text data of the agent and the customer conversation processed by the user interaction unit), work order basic data such as business attribute, traffic type, subdivision type, note information, and environment data such as holiday, interest day, billing day, and housing interest rate reduction day are collected. Secondly, the test data is integrated with data sources, sliced according to time sequence, and periodic features are extracted to form a structured time series data set. Finally, the data is dynamically updated by using a sliding window mechanism, and after eliminating the dimension difference by normalization processing, it is stored in the database to complete the data standardization and storage.
[0090] The business prompt word prediction unit is used for predicting output business scene prompt words. The data processing flow of the business prompt word prediction unit is as shown in Figure 5 The business prompt word prediction unit collects business data, inputs the prompt word prediction model for training to obtain business scene prompt words, which are used to assist the intelligent generation model to generate work order summaries, and are combined with the prompt word deviation data (prompt words involved in the modification part of the work order summary output by the intelligent generation model and the corresponding deviation rate) returned by the corresponding intelligent generation model intelligent interaction and feedback optimization unit to judge. If it does not meet the expectation, new prompt words are obtained according to the deviation content of the returned work order summary, which are input into the prediction model together with the collected business data and unbiased prompt words for retraining and optimization of the model, so as to improve the fitting degree of the business prompt word prediction and continuously improve the accuracy of the work order summary content generated by the intelligent generation model.
[0091] The prompt word management unit is used for prompt word assembly processing, and the processed prompt word is used for inputting an intelligent generation model to generate a personalized work order summary. The prompt word unit is divided into a general prompt word management unit and a business scenario special prompt word management unit. The general prompt word includes three parts of basic requirements, model random numbers and dialogue rules, and the general prompt word remains unchanged in different business scenarios. The business scenario special prompt word includes five parts of dialogue intent, specific business process, customer label, historical dialogue information and special date (such as holidays, interest payment dates, billing dates and housing interest rate reduction dates), and the special prompt word content is obtained by a business prompt word prediction unit according to different business scenarios. The data processing flow of the prompt word management unit is as shown in Figure 6 , the general prompt word and the business scenario special prompt word can be combined and then transmitted to the intelligent generation model as input data.
[0092] The intelligent generation model intelligent interaction and feedback optimization unit is used for forming data backflow in real time during a call process, so as to complete closed-loop optimization and iterative improvement. As shown in Figure 7 , a work order summary is generated in real time by clicking a filling button during a seat call process, the content of the work order summary supports modification, the generation result and the final submission result of the work order summary are input into an intelligent generation model prompt word deviation identifier, the deviation content and the corresponding deviation rate are calculated, and the deviation content and the corresponding deviation rate are input into a business prompt word prediction unit to form data backflow, the business prompt word prediction model is retrained, the work order summary generation result is cyclically optimized, and the work order summary accuracy is continuously improved.
[0093] By applying the method provided in the embodiment of the application, the business scenario prompt word is predicted through call data, and is combined with the general prompt word for use, the accuracy of the intelligent generation model in generating intelligent filling data is effectively improved, the scene adaptability is stronger, the corresponding work order summary content can be flexibly generated according to different scenes, the seat can modify and evaluate the intelligent filling result, the deviation data of the submission content and the generation content is obtained by comparison, is re-input into the business prompt word prediction model for training after processing, the prompt word prediction scene with a larger deviation is corrected, the data acquisition, audit annotation and data backflow are integrated into the daily work process, the cyclic optimization closed-loop construction is completed, the deviation rate is continuously reduced, and the work order summary accuracy is continuously improved.
[0094] Corresponding to the method of Figure 1 , the embodiment of the application further provides a work order summary generation device for specific implementation of the method in Figure 1 , a structure diagram is as shown in Figure 8 , and the device includes:
[0095] The receiving unit 801 is configured to, in response to receiving a work order generation instruction, acquire service data corresponding to the work order generation instruction; the service data comprises call data, work order information and environment information;
[0096] The prediction unit 802 is configured to input the service data into a pre-trained prompt word prediction model to obtain a service scenario prompt word;
[0097] The execution unit 803 is configured to combine the service scenario prompt word with a general prompt word to obtain a target prompt word; the general prompt word comprises a dialogue rule;
[0098] The generation unit 804 is configured to input the target prompt word into a work order summary generation model to generate a work order summary.
[0099] In an embodiment provided in the present application, optionally, the prediction unit 802 comprises:
[0100] The first acquisition subunit is configured to acquire a training data set and an initial prediction model to be trained; the training data set comprises a plurality of training samples, and each training sample comprises historical service data and a sample label; the sample label comprises a reference prompt word;
[0101] The first determination subunit is configured to determine a target training sample to be currently used for training in each training sample;
[0102] The output subunit is configured to input the target training sample into the initial prediction model to obtain a prediction prompt word output by the initial prediction model;
[0103] The update subunit is configured to update model parameters of the initial prediction model according to a difference degree between the prediction prompt word and the reference prompt word;
[0104] The second determination subunit is configured to, in a case where the updated initial prediction model does not satisfy a training completion condition, return to perform the step of determining a target training sample to be currently used for training in each training sample;
[0105] In a case where the updated initial prediction model does not satisfy the training completion condition, the updated initial prediction model is taken as a trained prompt word prediction model.
[0106] In an embodiment provided in the present application, optionally, the execution unit 803 comprises:
[0107] The second acquisition subunit is configured to acquire a general prompt word in a preset configuration file; the general prompt word comprises a dialogue rule, a model random number and a work order summary generation requirement;
[0108] The splicing sub-unit is configured to splice the service scenario prompt word and the general prompt word to obtain a target prompt word.
[0109] The specific principles and execution processes of each unit and module in the work order summary generation apparatus disclosed in the embodiments of the present application are the same as those of the work order summary generation method disclosed in the embodiments of the present application, and can be referred to the corresponding parts in the work order summary generation method provided by the embodiments of the present application.
[0110] The embodiments of the present application further provide a storage medium, which comprises stored instructions, wherein when the instructions are executed, the device where the storage medium is located performs the work order summary generation method.
[0111] The embodiments of the present application further provide an electronic device, a structural schematic diagram of which is shown in Figure 9 The electronic device specifically comprises a memory 901 and one or more than one instruction 902, wherein the one or more than one instruction 902 is stored in the memory 901 and is configured to be executed by one or more than one processor 903 to execute the work order summary generation method.
[0112] It can be understood that, before using the technical solutions disclosed in the embodiments of the present application, the type, use range, use scenario and the like of the personal information involved in the present application should be informed to the user and the authorization of the user should be obtained through appropriate means according to relevant laws and regulations.
[0113] For example, in response to receiving the active request of the user, prompt information is sent to the user to explicitly prompt the user that the operation requested to be executed will need to obtain and use the personal information of the user. Thus, the user can voluntarily choose whether to provide the personal information to the software or hardware such as the electronic device, the application program, the server or the storage medium that executes the technical solutions of the present application according to the prompt information.
[0114] As an optional but not limited implementation manner, in response to receiving the active request of the user, the manner of sending the prompt information to the user may, for example, be the manner of pop-up window, and the prompt information may, for example, be presented in the form of text in the pop-up window. In addition, the pop-up window may further carry a selection control for the user to select "agree" or "disagree" to provide the personal information to the electronic device.
[0115] It should be noted that each embodiment in the present specification adopts a progressive manner for description, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between each embodiment can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.
[0116] Finally, it should be noted that the terminology used herein, such as first and second, is merely used to differentiate one entity or operation from another entity or operation and does not necessarily require or imply any such actual relationship or order between the entities or operations.
[0117] For the convenience of description, the above apparatus is described in various units by function respectively. Of course, the functions of the units can be implemented in one or more software and / or hardware in the implementation of the present application.
[0118] From the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in various embodiments or some parts of the embodiments of the present application.
[0119] The above provides a detailed description of a work order summary generation method provided by the present application, and the principles and implementation manners of the present application are described by applying specific examples; the above embodiment descriptions are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manner and application range will have changes; in summary, the content of the specification should not be understood as a limitation of the present application.
Claims
1. A method for generating a work order closure, the method comprising: The method comprises the following steps: In the case where a work order generation instruction is received, business data corresponding to the work order generation instruction is acquired; the business data comprises call data, work order information and environmental information; The business data is input into a pre-trained prompt word prediction model to obtain a business scenario prompt word; The business scenario prompt word is combined with a general prompt word to obtain a target prompt word; The general prompt word comprises a dialogue rule; The target prompt word is input into a work order summary generation model to generate a work order summary.
2. The method of claim 1, wherein, The training process of the prompt word model comprises the following steps: A training data set and an initial prediction model to be trained are acquired; the training data set comprises a plurality of training samples, each of which comprises historical business data and a sample label, and the sample label comprises a reference prompt word; A target training sample to be currently used for training is determined in each of the training samples; The target training sample is input into the initial prediction model to obtain a prediction prompt word output by the initial prediction model; Model parameters of the initial prompt word are updated according to a difference degree between the prediction prompt word and the reference prompt word; In the case where the updated initial prediction model does not meet a training completion condition, the step of determining the target training sample to be currently used for training in each of the training samples is performed again; In the case where the updated initial prediction model does not meet the training completion condition, the updated initial prediction model is taken as a trained prompt word prediction model.
3. The method of claim 1, wherein, The combination of the business scenario prompt word and the general prompt word to obtain the target prompt word comprises the following steps: A general prompt word is acquired in a preset configuration file; the general prompt word comprises a dialogue rule, a model random number and a work order summary generation requirement; The business scenario prompt word and the general prompt word are spliced to obtain the target prompt word.
4. The method of claim 1, wherein, The business scenario prompt word comprises at least one of the following: dialogue intent information, business processing flow description, customer feature label, historical interaction record and date feature information; The date feature information comprises at least one of the following: holiday date, financial business interest settlement date, bill generation date and interest rate adjustment effective date.
5. The method of claim 1, wherein, After the target prompt word is input into the work order summary generation model to generate a work order summary, the following steps are further included: The work order summary is output to enable a user to audit the work order summary; In the case where a user-submitted work order is detected, it is detected whether the work order summary and the user-submitted work order are deviated; If the work order summary and the user-submitted work order are deviated, the work order summary generation model is optimized according to the deviation between the work order summary and the user-submitted work order.
6. A work order closure generation apparatus, comprising: The method comprises the following steps: A receiving unit is configured to acquire business data corresponding to a work order generation instruction in the case where the work order generation instruction is received; the business data comprises call data, work order information and environmental information; A prediction unit is configured to input the business data into a pre-trained prompt word prediction model to obtain a business scenario prompt word; An execution unit is configured to combine the business scenario prompt word with a general prompt word to obtain a target prompt word. The general prompt word comprises a dialogue rule; The generating unit is configured to input the target prompt word into a work order summary generation model to generate a work order summary.
7. The apparatus of claim 6, wherein, The prediction unit comprises: The first obtaining sub-unit is configured to obtain a training data set and an initial prediction model to be trained, wherein the training data set comprises a plurality of training samples, each of the training samples comprises historical business data and a sample label, and the sample label comprises a reference prompt word. The first determining sub-unit is configured to determine a target training sample to be trained in each of the training samples. The output sub-unit is configured to input the target training sample into the initial prediction model to obtain a predicted prompt word output by the initial prediction model. The updating sub-unit is configured to update model parameters of the initial prediction model according to a difference between the predicted prompt word and the reference prompt word. The second determining sub-unit is configured to return to the step of determining the target training sample to be trained in each of the training samples if the updated initial prediction model does not meet a training completion condition. If the updated initial prediction model does not meet the training completion condition, the updated initial prediction model is used as a trained prompt word prediction model.
8. The apparatus of claim 6, wherein, The executing unit comprises: The second obtaining sub-unit is configured to obtain a general prompt word in a preset configuration file, wherein the general prompt word comprises a dialogue rule, a model random number, and a work order summary generation requirement. The splicing sub-unit is configured to splice the business scenario prompt word and the general prompt word to obtain a target prompt word.
9. A storage medium, characterized by The storage medium comprises a storage instruction, wherein the storage instruction is configured to control a device in which the storage medium is located to perform the work order summary generation method according to any one of claims 1 to 5.
10. An electronic device, comprising: The device comprises a memory and one or more instructions, wherein the one or more instructions are stored in the memory and are configured to be executed by one or more processors to perform the work order summary generation method according to any one of claims 1 to 5.