Problem processing method, device and equipment based on large model and storage medium

Abstract

The invention discloses a problem processing method and device based on a large model, equipment and a storage medium, and relates to the technical field of artificial intelligence. The method comprises the steps of generating a first decision scheme according to problem log data of a to-be-processed problem and current account data associated with the to-be-processed problem based on a pre-trained large model; according to the problem log data, matching the to-be-processed problem with each standard problem in a knowledge base to obtain a successfully matched target problem, and determining a second decision scheme according to the target problem; and generating a screening prompt text according to the first decision scheme and the second decision scheme, and inputting the screening prompt text into the large model to obtain an output target decision scheme. According to the technical scheme, intelligent decision making of the to-be-processed problem is achieved, and the accuracy and efficiency of abnormal positioning of the to-be-processed problem are improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, in particular to the field of artificial intelligence technology, and specifically to a problem processing method, apparatus, device and storage medium based on a large model. Background Art

[0002] Large models refer to machine learning models with large-scale parameters and powerful computing power. They are usually built by deep neural networks and have billions or even hundreds of billions of parameters. These models can understand and process complex input information by learning large amounts of data, thereby generating content such as text and images.

[0003] In the banking business field, once an abnormal problem is encountered during the processing of a certain account business, traditional troubleshooting methods such as manual review of logs and comparison of transaction records one by one are often time-consuming, inefficient, and difficult to quickly identify the root cause of the problem. How to apply big model technology to abnormal problem location scenarios is very important. Summary of the invention

[0004] The present application provides a problem handling method, apparatus, device and storage medium based on a large model to improve the accuracy of locating abnormal problems.

[0005] According to one aspect of the present application, a problem processing method based on a large model is provided, comprising:

[0006] Based on the pre-trained large model, a first decision solution is generated according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed;

[0007] According to the problem log data, the problem to be processed is matched with each standard problem in the knowledge base to obtain a successfully matched target problem, and a second decision plan is determined according to the target problem;

[0008] A screening prompt text is generated according to the first decision plan and the second decision plan, and the screening prompt text is input into the large model to obtain an output target decision plan.

[0009] According to another aspect of the present application, a problem processing device based on a large model is provided, the device comprising:

[0010] A first decision module, configured to generate a first decision solution based on a pre-trained large model and according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed;

[0011] A second decision module is used to match the problem to be processed with each standard problem in the knowledge base according to the problem log data to obtain a successfully matched target problem, and determine a second decision solution according to the target problem;

[0012] The decision screening module is used to generate screening prompt text according to the first decision plan and the second decision plan, and input the screening prompt text into the large model to obtain the output target decision plan.

[0013] According to another aspect of the present application, an electronic device is provided, the electronic device comprising:

[0014] one or more processors;

[0015] A memory for storing one or more programs;

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement any one of the large model-based problem processing methods provided in the embodiments of the present application.

[0017] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, any large model-based problem processing method provided in the embodiments of the present application is implemented.

[0018] According to another aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements any large model-based problem processing method provided in the embodiments of the present application.

[0019] This application uses a pre-trained large model combined with problem log data, current account data associated with the problem to be processed, and various standard questions in the knowledge base to generate and screen decision-making plans for the problems to be processed. It can quickly and accurately locate the problems, that is, reduce manual intervention through intelligent and automated methods, and improve processing efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flowchart of a problem processing method based on a large model provided according to the first embodiment of the present application;

[0021] Figure 2a It is a flowchart of a problem processing method based on a large model provided according to the second embodiment of the present application;

[0022] Figure 2b It is a specific schematic diagram of a problem processing method based on a large model provided in Example 2 of the present application;

[0023] Figure 3It is a structural schematic diagram of a problem processing device based on a large model provided according to the third embodiment of the present application;

[0024] Figure 4 It is a structural diagram of an electronic device that implements the large model-based problem-solving method of an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] In addition, it should be noted that in the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of relevant data such as email address blacklists and emails to be detected are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0028] Embodiment 1

[0029] Figure 1 This is a flowchart of a problem handling method based on a large model provided in accordance with the first embodiment of the present application. This embodiment is applicable to the situation of locating abnormal problems and can be executed by a problem handling device based on a large model. The problem handling device based on a large model can be implemented in the form of hardware and / or software. The problem handling device based on a large model can be configured in a computer device. Figure 1 As shown, the method includes:

[0030] S110, based on the pre-trained large model, generating a first decision solution according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed;

[0031] S120, according to the problem log data, matching the problem to be processed with each standard problem in the knowledge base to obtain a successfully matched target problem, and determining a second decision solution according to the target problem;

[0032] S130, generating a screening prompt text according to the first decision plan and the second decision plan, and inputting the screening prompt text into the large model to obtain an output target decision plan.

[0033] The knowledge base can be a general domain knowledge base or an enterprise-level private domain knowledge base. The knowledge base can include standard questions and standard decision-making solutions corresponding to standard questions. Taking the construction of a private domain knowledge base within a bank as an example, the operation log data of the bank's business system and the interface documents of each business interface in the bank's business system can be obtained, and the knowledge base can be constructed through natural language processing (NLP) technology.

[0034] The operation log data can record the operation status of the banking business system, abnormal details, and operations performed by users. The interface document can include structured information such as the functional description, parameter list, and return value structure of the business interface. By using natural language processing, key fields such as timestamp, error code, and detailed error description can be extracted from the operation log data; and the interface document can be parsed to obtain key fields such as interface name, function overview, request parameter details, and response parameter structure. Based on the extracted key fields, the standard problem set related to banking business, the standard problem causes and standard decision plans corresponding to each standard problem in the standard problem set are determined to obtain a knowledge base. The standard decision plan is used to process standard problems, restore the banking business system to normal, and reduce the impact on the business system. A standard problem can correspond to multiple standard problem causes, and a standard problem and a standard problem cause combination can correspond to a standard decision plan.

[0035] Moreover, a training set can be constructed based on the historical production problems of the banking business system and the corresponding decision-making solutions, as well as the standard problems, standard problem causes, and standard decision-making solutions in the knowledge base. The training set includes at least sample problems and decision-making solutions annotated with the sample problems. The training set can be used to fine-tune the general large model to obtain a pre-trained large model for abnormal location of the problem to be processed. Exemplarily, the program running log information within a period of production and all the data in the knowledge base are collected; unnecessary information is removed from the knowledge base data and formatted for pre-training; logs unrelated to transactions are removed from the logs and formatted for model fine-tuning; by setting parameters such as training batches, learning rates, and number of rounds, and adjusting the parameters of the general large model for multiple rounds through actual trials, a pre-trained large model is obtained for abnormal location.

[0036] Exemplarily, a user can input a pending issue of a business system in the interactive page of a pre-trained big model, the big model calls the log analysis tool to extract the problem log data of the pending issue from the log file of the business system, and the big model calls the account system corresponding to the business system to query the current account associated with the pending issue. For example, if the pending issue is the failure of Zhang San's transfer transaction, the current account refers to Zhang San's specific account used for financial transactions in the banking business system; and the current account data such as basic information, transaction records, and account status of the current account are queried from the account system. The pre-trained big model also generates at least one first decision plan based on the problem log data of the pending issue and the current account data.

[0037] Moreover, the problem to be processed can be matched with each standard problem in the knowledge base to obtain a successfully matched target problem, and at least one second decision solution can be determined based on the standard problem cause and standard decision solution associated with the target problem. Among them, the prompt text (Prompt) refers to the initial text input provided by the user to guide the model to generate a specific output, and the screening prompt text is used to guide the large model to further screen each first decision solution and each second decision solution. The screening prompt text can be generated based on the first decision solution and the second decision solution, and the screening prompt text can be input into the pre-trained large model to obtain the target decision solution selected from each first decision solution and each second decision solution.

[0038] The technical solution provided in the embodiment of the present application generates at least one first decision plan based on the problem log data of the problem to be processed and the associated current account data based on a pre-trained big model, and obtains at least one second decision plan by matching the problem to be processed with the standard problem in the knowledge base, and integrates the two to generate a screening prompt text and uses the big model for optimization again to obtain a more accurate target decision plan, thereby realizing intelligent decision-making on the problem to be processed and improving the accuracy and efficiency of abnormal location of the problem to be processed.

[0039] In an optional implementation, the method of matching the problem to be processed with each standard problem in a knowledge base based on the current diagnostic data of the problem to be processed to obtain a successfully matched target problem includes: extracting at least one current problem attribute of the current occurrence time of the problem to be processed, the current problem phenomenon or the current problem classification from the problem log data of the problem to be processed; semantically encoding the current problem attribute to obtain current problem features; determining the semantic similarity between the current problem features and the standard problem features of each standard problem in the knowledge base, and determining the successfully matched target problem based on the semantic similarity.

[0040] In an embodiment of the present application, at least one current problem attribute such as the current occurrence time, current problem phenomenon or current problem classification of the problem to be processed can be extracted from the problem log data of the problem to be processed; the current problem attribute is input into the semantic encoder to obtain the current problem feature; and the standard problem feature of each standard problem in the knowledge base is obtained, and the semantic similarity between the current problem feature and each standard problem feature, such as cosine similarity, is determined, and the standard problem with relatively high similarity is used as the target problem with successful matching. Among them, each standard problem feature is obtained by pre-semantically encoding the standard occurrence time, standard problem phenomenon or at least one standard problem attribute in the standard problem classification of each standard problem. By extracting the current problem attribute from the problem log of the problem to be processed and performing semantic encoding to obtain the current problem feature, and performing semantic similarity matching between the current problem feature and the standard problem feature pre-generated for each standard problem in the knowledge base, the target problem with successful matching can be quickly and accurately located, further improving the efficiency and accuracy of problem diagnosis.

[0041] Embodiment 2

[0042] Figure 2a is a flowchart of a problem processing method based on a large model provided according to the second embodiment of the present application. Figure 2a Based on the above embodiment, the problem processing method based on the large model of this embodiment may include:

[0043] S210, extracting at least one current problem attribute of the current occurrence time of the problem to be processed, the current problem phenomenon or the current problem classification from the problem log data of the problem to be processed;

[0044] S220, extracting at least one current account attribute of the current region, current account level, current asset size, historical transaction records, or business usage habits of the current account from the current account data;

[0045] S230, based on the pre-trained large model, generating a first decision solution according to the current problem attribute, the current account attribute and the problem log data;

[0046] S240, according to the problem log data, matching the problem to be processed with each standard problem in the knowledge base to obtain a successfully matched target problem, and determining a second decision solution according to the target problem;

[0047] S250, generating a screening prompt text according to the first decision plan and the second decision plan, and inputting the screening prompt text into the large model to obtain an output target decision plan.

[0048] Among them, the current occurrence time can be obtained by extracting the timestamp from the problem log data; the current problem phenomenon is the specific symptoms or conditions exhibited when the problem to be processed occurs, such as transaction abnormalities, excessive CPU utilization, insufficient memory, system crashes, interface freezes, slow responses, server downtime, etc. The current problem is classified into abnormal data, server downtime, hardware abnormalities, program abnormalities, network environment abnormalities, poor communication, etc.

[0049] In addition, the current account associated with the pending issue can also be obtained, and at least one of the current account attributes of the current region, current account level, current asset size, historical transaction records or business usage habits of the current account can be queried from the user database in the business system according to the current account. It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with the relevant laws, regulations and standards of the relevant regions.

[0050] Exemplarily, the current problem attributes such as the current occurrence time of the problem to be processed, the current problem phenomenon or the current problem classification, the current account attributes such as the current region of the current account, the current account level, the current asset size, the historical transaction records or business usage habits, and the problem log data can be input into the pre-trained big model to obtain at least one first decision plan output by the big model. Different current problem attributes and associated current account attributes of the problem to be processed may lead to different causes of the problem to be processed, and thus different decision plans need to be taken to deal with it. By introducing the current problem attributes and the current account attributes, the root cause of the problem can be analyzed more accurately and the solution can be customized, thereby effectively improving the efficiency and effectiveness of problem handling.

[0051] In an optional implementation, the pre-trained large model generates a first decision plan according to the current problem attributes, the current account attributes and the problem log data, including: determining the current business type preferred by the current account according to the historical transaction records and the business usage habits; determining the current business interface used for the problem to be processed according to the current region, the current account level, the current asset size and the current business type; generating a first decision plan based on the problem log data and the current business interface.

[0052] Among them, historical transaction records are generated based on the historical behavior data of the current account, such as deposits, withdrawals, transfers, fund purchases, financial product investments, credit card consumption, etc.; business usage habits can be the most frequently used service channels, preferred transaction time periods, preferred service types, etc. For example, some users prefer fund business, and some users prefer insurance business; combined with the current region, current account level, current asset size and current business type, determine the current transaction process to which the problem to be processed belongs, and locate the current transaction link from the various transaction links included in the current transaction process, and then determine the current business interface that may be used for the problem to be processed based on the current transaction link, and use the current business interface as context information of the problem log data, input it into the pre-trained large model, and obtain the first decision plan.

[0053] For the same pending issue, the transaction process and its current link to which the pending issue belongs may be different depending on the current region, current account level, current asset size, and current business type preferred by the current account. The current business interface that may be used for the pending issue is determined accordingly, thereby improving the reliability of the current business interface; and, by inputting the current business interface as context information of the problem log data into the pre-trained large model to guide the generation of the first decision plan, the generation efficiency of the first decision plan can be further improved.

[0054] In an optional implementation, a first decision plan is generated based on the problem log data and the current business interface, including: generating normal log prompt text based on the current business interface and the problem log data, and inputting the normal log prompt text into the big model to obtain normal log data of the problem to be processed; inputting the normal log data and the problem log data into the big model to obtain the first decision plan.

[0055] Exemplarily, the functions, input and output parameters, processing logic, expected results and other descriptive information of the current business interface and the problem log data are added to the preset normal log prompt template to obtain the normal log prompt text; the normal log prompt text is input into the pre-trained large model to obtain the normal log data corresponding to the problem to be processed. By introducing the functions, input and output parameters, processing logic, expected results, etc. of the current business interface into the normal log prompt text, it is convenient to further understand the missing, wrong or inconsistent fields or values ​​in the problem log data, and process the corresponding abnormal data to generate normal log data, thereby improving the efficiency of generating normal log data. Moreover, the normal log data and the problem log data are also input into the large model, and the differences between the normal log data and the problem log data are identified by performing semantic comparison analysis on the normal log data and the problem log data, and the current problem cause to which the problem to be processed belongs is determined according to the difference, and at least one first decision plan is generated for the current problem cause, thereby further improving the accuracy of the first decision plan.

[0056] The technical solution provided in the embodiment of the present application determines the current business interface that may be used for the problem to be processed based on the current region, current account level, current asset size, current account preferences, etc., and uses the current business interface as context information of the problem log data to guide the generation of the first decision plan, thereby improving the generation efficiency and accuracy of the first decision plan, thereby further improving the efficiency of problem handling.

[0057] In an optional embodiment, the method also includes: obtaining a third decision plan input by a user; generating a filtering prompt text according to the first decision plan and the second decision plan, and inputting the filtering prompt text into the large model to obtain an output target decision plan, including: obtaining a decision cost of at least one of the first decision plan, the second decision plan and the third decision plan, the decision cost being used to characterize the corresponding resource consumption or system interruption duration; generating a filtering prompt text according to the first decision plan, the second decision plan and the third decision plan, and the decision cost, and inputting the filtering prompt text into the large model to obtain an output target decision plan.

[0058] refer to Figure 2b For the problem to be processed, the problem to be processed is input into a pre-trained large model for intelligent decision-making processing to obtain at least one first decision plan, and the problem to be processed is matched with the knowledge base to obtain at least one second decision plan. At least one third decision plan input by the user based on personal experience and judgment can also be obtained; the decision cost corresponding to the first decision plan, the second decision plan or the third decision plan can also be obtained, such as the corresponding resource consumption, system interruption duration, etc.; combined with the corresponding decision cost, a screening prompt text is generated.

[0059] Take the following situation as an example: The screening prompt is: Your current role is a decision maker with rich experience in handling various corporate issues. You are good at handling complex affairs in various companies and giving decision-making opinions. Now we have encountered a corporate affair that requires decision-making. Please make decision opinions based on similar problems you have encountered in various companies on the external network, combined with the internal data analysis of the company, and combined with your past decision-making experience. As a decision, please do not diverge or guess at will, and you need to give accurate opinions. The problem to be processed is: There is now a problem in production where users occasionally cannot access functions. The problem affects one-tenth of the total number of users, and customer complaints have already been generated. The first decision plan is to shut down and repair the problem immediately. The corresponding decision cost is that all users cannot access the system within 1 hour; the second decision plan is to temporarily maintain the status quo in production and wait for shutdown and repair when the number of users is low at night; the third decision plan is to restart the system. The generated screening prompt text can be as follows:

[0060] Your current role is a decision maker with rich experience in handling various corporate issues. You are good at handling complex affairs in various companies and giving decision-making opinions. Now we have encountered a corporate matter that needs to be decided. Please give decision-making opinions based on your solutions to similar problems encountered by various companies on the external network, as well as the analysis of internal data of this company, and your past decision-making experience. As a decision-maker, please do not diverge or guess at will, and give accurate opinions.

[0061] Problem description: Currently, there is a problem in production where users are unable to access functions occasionally. The problem affects one-tenth of all users. Customer complaints have been generated. The problem needs to be fixed by shutting down all servers for 1 hour. The project team currently has three solutions: the first is to shut down and fix the problem immediately, but this will cause all users to be unable to access the system for 1 hour; the second is to temporarily maintain the status quo of production and wait for the night when the number of users is low to shut down and fix it; the third is to restart the business system. Which solution will have the least impact and the best effect?

[0062] The above-mentioned processing generates a screening prompt text and inputs it into the big model by combining the first decision plan obtained by the intelligent decision of the big model, the second decision plan obtained by matching the knowledge base, the third decision plan based on human personal experience judgment, and the decision cost corresponding to the first decision plan, the second decision plan or the third decision plan. This can effectively screen out the optimal target decision plan, thereby optimizing resource allocation and reducing system impact while ensuring effective problem solving.

[0063] Embodiment 3

[0064] Figure 3This is a schematic diagram of a problem handling device based on a large model according to the third embodiment of the present application, which can be applied to the situation of locating abnormal problems and can be executed by a problem handling device based on a large model. The problem handling device based on a large model can be implemented in the form of hardware and / or software, and the problem handling device based on a large model can be configured in a computer device. Figure 3 As shown, the device comprises:

[0065] A first decision module 310 is used to generate a first decision solution based on a pre-trained large model according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed;

[0066] The second decision module 320 is used to match the problem to be processed with each standard problem in the knowledge base according to the problem log data to obtain a successfully matched target problem, and determine a second decision solution according to the target problem;

[0067] The decision screening module 330 is used to generate screening prompt text according to the first decision plan and the second decision plan, and input the screening prompt text into the large model to obtain the output target decision plan.

[0068] In an optional implementation, the first decision module 310 includes:

[0069] A problem attribute unit, used to extract at least one current problem attribute of the current occurrence time of the problem to be processed, the current problem phenomenon or the current problem classification from the problem log data of the problem to be processed;

[0070] An account attribute unit, used to extract at least one current account attribute of the current region, current account level, current asset size, historical transaction records or business usage habits of the current account from the current account data;

[0071] The first decision unit is used to generate a first decision solution based on a pre-trained large model according to the current problem attribute, the current account attribute and the problem log data.

[0072] In an optional implementation, the first decision unit includes:

[0073] A business type subunit, used to determine the current business type preferred by the current account according to the historical transaction records and the business usage habits;

[0074] A business interface subunit, configured to determine a current business interface used for the problem to be processed according to the current region, the current account level, the current asset size and the current business type;

[0075] The first decision subunit is used to generate a first decision solution according to the problem log data and the current business interface.

[0076] In an optional implementation manner, the first decision subunit is specifically configured to:

[0077] Generate a normal log prompt text according to the current business interface and the problem log data, and input the normal log prompt text into the large model to obtain the normal log data of the problem to be processed;

[0078] The normal log data and the problem log data are input into the large model to obtain a first decision solution.

[0079] In an optional implementation, the second decision module 320 includes:

[0080] A problem attribute unit, used to extract at least one current problem attribute of the current occurrence time of the problem to be processed, the current problem phenomenon or the current problem classification from the problem log data of the problem to be processed;

[0081] A question feature unit, used for semantically encoding the current question attribute to obtain the current question feature;

[0082] The second decision unit is used to determine the semantic similarity between the current question feature and the standard question features of each standard question in the knowledge base, and determine the successfully matched target question according to the semantic similarity.

[0083] In an optional implementation, the above-mentioned problem processing device based on a large model further includes a third decision module, which is used to obtain a third decision solution input by a user; the decision screening module includes:

[0084] A decision cost unit, used to obtain a decision cost of at least one of the first decision solution, the second decision solution and the third decision solution, wherein the decision cost is used to characterize the corresponding resource consumption or system interruption duration;

[0085] A decision screening unit is used to generate a screening prompt text according to the first decision plan, the second decision plan, the third decision plan, and the decision cost, and input the screening prompt text into the large model to obtain an output target decision plan.

[0086] The big model-based problem processing device provided in the embodiment of the present application can execute the big model-based problem processing method provided in any embodiment of the present application, and has the corresponding functional modules and beneficial effects for executing each big model-based problem processing method.

[0087] According to an embodiment of the present application, the present application also provides an electronic device, a readable storage medium and a computer program product.

[0088] Embodiment 4

[0089] Figure 4 It is a structural diagram of an electronic device 410 that implements a large model-based problem-solving method of an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0090] like Figure 4 As shown, the electronic device 410 includes at least one processor 411, and a memory connected to the at least one processor 411 in communication, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 411 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 to the random access memory (RAM) 413. In RAM413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, ROM412 and RAM413 are connected to each other via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0091] Multiple components in the electronic device 410 are connected to the I / O interface 415, including: an input unit 416, such as a keyboard, a mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a disk, an optical disk, etc.; and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0092] Processor 411 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. Processor 411 performs the various methods and processes described above, such as a large model-based problem processing method.

[0093] In some embodiments, the large model-based problem processing method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the large model-based problem processing method described above can be performed. Alternatively, in other embodiments, the processor 411 can be configured as a large model-based problem processing method in any other appropriate manner (for example, by means of firmware).

[0094] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0095] The computer programs for implementing the methods of the present application may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable large-model-based problem-solving device, so that when the computer programs are executed by the processor, the functions / operations specified in the flow charts and / or block diagrams are implemented. The computer programs may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0096] In the context of the present application, a computer readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device or equipment. A computer readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer readable storage medium may be a machine readable signal medium. A more specific example of a machine readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0097] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0098] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0099] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0100] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution of this application can be achieved, and this document is not limited here.

[0101] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A problem-solving method based on a large model, characterized in that: include: Based on the pre-trained large model, a first decision solution is generated according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed; According to the problem log data, the problem to be processed is matched with each standard problem in the knowledge base to obtain a successfully matched target problem, and a second decision plan is determined according to the target problem; A screening prompt text is generated according to the first decision plan and the second decision plan, and the screening prompt text is input into the large model to obtain an output target decision plan.

2. The method according to claim 1, characterized in that The method of generating a first decision solution based on the pre-trained large model according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed includes: Extracting at least one current problem attribute of the current occurrence time, current problem phenomenon or current problem classification of the problem to be processed from the problem log data of the problem to be processed; Extracting at least one current account attribute of the current account, including the current region, current account level, current asset size, historical transaction records, or business usage habits, from the current account data; Based on the pre-trained large model, a first decision solution is generated according to the current problem attributes, the current account attributes and the problem log data.

3. The method according to claim 2, characterized in that The method of generating a first decision solution based on the pre-trained large model according to the current problem attribute, the current account attribute and the problem log data includes: Determine the current business type preferred by the current account based on the historical transaction records and the business usage habits; Determine the current business interface used for the problem to be processed according to the current region, the current account level, the current asset size and the current business type; A first decision solution is generated according to the problem log data and the current business interface.

4. The method according to claim 3, characterized in that Generating a first decision solution according to the problem log data and the current business interface includes: Generate a normal log prompt text according to the current business interface and the problem log data, and input the normal log prompt text into the large model to obtain the normal log data of the problem to be processed; The normal log data and the problem log data are input into the large model to obtain a first decision solution.

5. The method according to claim 1, characterized in that The step of matching the problem to be processed with each standard problem in the knowledge base according to the current diagnostic data of the problem to be processed to obtain a successfully matched target problem includes: Extracting at least one current problem attribute of the current occurrence time, current problem phenomenon or current problem classification of the problem to be processed from the problem log data of the problem to be processed; Semantically encoding the current question attributes to obtain current question features; Determine the semantic similarity between the current question feature and the standard question features of each standard question in the knowledge base, and determine the successfully matched target question based on the semantic similarity.

6. The method according to claim 1, characterized in that The method further includes obtaining a third decision solution input by a user; The step of generating a screening prompt text according to the first decision plan and the second decision plan, and inputting the screening prompt text into the large model to obtain an output target decision plan includes: Obtaining a decision cost of at least one of the first decision solution, the second decision solution, and the third decision solution, where the decision cost is used to characterize corresponding resource consumption or system interruption duration; A screening prompt text is generated according to the first decision plan, the second decision plan, the third decision plan, and the decision cost, and the screening prompt text is input into the large model to obtain an output target decision plan.

7. A problem-solving device based on a large model, characterized in that: include: A first decision module, configured to generate a first decision solution based on a pre-trained large model and according to the problem log data of the problem to be processed and the current account data associated with the problem to be processed; A second decision module is used to match the problem to be processed with each standard problem in the knowledge base according to the problem log data to obtain a successfully matched target problem, and determine a second decision solution according to the target problem; The decision screening module is used to generate screening prompt text according to the first decision plan and the second decision plan, and input the screening prompt text into the large model to obtain the output target decision plan.

8. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the large model-based problem processing method as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the large model-based problem processing method as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the large model-based problem processing method according to any one of claims 1 to 6.

Images