Business processing method and device, equipment, medium and program product
Through a large language model, the problem of high business processing costs and inability to automate in the existing technology is solved, and the analysis accuracy and automation level are improved, and the cost is reduced.
Patent Information
- Application Number
- CN202411111056.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-26
AI Technical Summary
In the prior art, business processing costs are high and cannot be automated. Engineers have high requirements for mathematical modeling and processing complex business problems, and cannot effectively analyze and automate processing.
By calling the large language model to extract business key information, generate structured fields, and use pre-built storage database to query and solve data, convert formats to generate executable commands, call the solution tool to solve, output business processing results, and reduce the large language model. Computing power and corpus processing difficulty to improve automation.
It improves the accuracy of business problem analysis, reduces costs, reduces dependence on engineers, and realizes automation of business processing.
Smart Images

Figure CN119025649A_ABST
Abstract
Description
Background Art
[0002] With the development of the Internet, business is expanding faster and faster. Often, there are some complex problems that need to be dealt with in different businesses. Many of these problems require mathematical modeling through operations research methods, and are very complex.
[0003] In related technologies, engineers are required to analyze various business problems and build mathematical models to solve the problems. This places high demands on engineers, is costly, and cannot be automated.
[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention
[0005] The purpose of the present disclosure is to provide a business processing method, device, equipment, medium and program product, which at least to some extent solve the problems of high cost and non-automation in the prior art.
[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.
[0007] According to a first aspect of the present disclosure, an embodiment of the present disclosure provides a service processing method, the method comprising:
[0008] Receive business questions input by users;
[0009] Call the large language model to obtain the structured fields corresponding to the key business information;
[0010] Call the pre-built storage database to query the solution data corresponding to the structured field; the solution data includes numerical information;
[0011] Convert the format of structured fields according to the solved data and generate executable commands;
[0012] Call the solving tool, solve according to the executable command, and output the business processing results.
[0013] In a possible embodiment, a large language model is called to obtain structured fields corresponding to key business information, including:
[0014] Call the large language model to extract key business information from business problems;
[0015] Map business-critical information to the middle layer; the middle layer is built based on programming languages;
[0016] The business-critical information is converted to obtain a structured field; the information included in the structured field is a programming language field obtained based on the programming language.
[0017] In a possible embodiment, extracting key business information from a business problem includes:
[0018] Based on the business problem, identify the type of business problem;
[0019] According to the type of business problem, business key information is extracted from the business problem; the business key information includes at least the following information: problem processing conditions and problem constraints.
[0020] In a possible embodiment, the training process of the large language model includes:
[0021] Randomly generate structured training sets;
[0022] Generate a business problem training set corresponding to the structured training set based on the structured training set;
[0023] Construct input-output pair training sets based on structured training sets and business problem training sets;
[0024] According to the input and output pairs of the training set, the large language model to be trained is trained to obtain a trained large language model.
[0025] In a possible embodiment, after receiving the business question input by the user, the method further includes:
[0026] Call the text classification model to determine the domain to which the business problem belongs and determine the target domain corresponding to the business problem;
[0027] Call the large language model of the target domain to obtain the structured fields corresponding to the key business information.
[0028] In a possible embodiment, format conversion is performed on structured fields according to the solution data to generate executable commands, including:
[0029] The structured fields are formatted according to the solved data to obtain intermediate fields;
[0030] According to the pre-built conversion program, the intermediate fields are converted to obtain executable commands.
[0031] In a possible embodiment, the large language model is a small large language model.
[0032] According to another aspect of the present disclosure, there is provided a service processing device, including:
[0033] A receiving unit, used for receiving a business question input by a user;
[0034] The extraction unit is used to call the large language model to obtain the structured fields corresponding to the key business information;
[0035] A query unit, used to call a pre-built storage database to query the solution data corresponding to the structured field; the solution data includes numerical information;
[0036] A conversion unit, used to convert the format of structured fields according to the solution data and generate executable commands;
[0037] The processing unit is used to call the solving tool, solve according to the executable command, and output the business processing results.
[0038] According to yet another aspect of the present disclosure, there is provided an electronic device, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform any one of the methods in the first aspect by executing the executable instructions.
[0039] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, any one of the methods in the first aspect is implemented.
[0040] According to another aspect of the present disclosure, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method in the first aspect above.
[0041] The embodiments of the present disclosure provide a business processing method, device, equipment, medium and program product, the method comprising: receiving a business problem input by a user, calling a large language model, obtaining a structured field corresponding to business key information, calling a pre-built storage database, querying the solution data corresponding to the structured field, the solution data including numerical information, formatting the structured field according to the solution data, generating an executable command, calling a solution tool, solving according to the executable command, and outputting the business processing result. In the present disclosure, the business problem is processed by a large language model, and a mathematical model is not directly constructed by a large language model to solve the problem. Instead, the structured field corresponding to the business key information is first extracted, and then part or all of the information in the structured field is converted into numerical information, and then an executable command is obtained. Not only can the accuracy of parsing business problems be improved, but also the requirements for the computing power of the large language model and the difficulty of corpus processing can be reduced. Finally, the executable command corresponding to the numerical information is solved to obtain the business processing result, improve the degree of automation, and reduce costs.
[0042] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0044] Figure 1 A flow chart showing a business processing method in an embodiment of the present disclosure is shown;
[0045] Figure 2 A flowchart showing a process of handling a business problem in an embodiment of the present disclosure is shown;
[0046] Figure 3 A flowchart of obtaining a structured field in an embodiment of the present disclosure is shown;
[0047] Figure 4 A flowchart showing a large language model training process in an embodiment of the present disclosure is shown;
[0048] Figure 5 A schematic diagram showing a preparation process and a use process in an embodiment of the present disclosure;
[0049] Figure 6 A schematic diagram showing the structure of a large language model agent in an embodiment of the present disclosure is shown;
[0050] Figure 7 A flowchart showing another business processing method in an embodiment of the present disclosure;
[0051] Figure 8 A schematic diagram showing the structure of a service processing device in an embodiment of the present disclosure is shown;
[0052] Fig. 9 A schematic structural diagram of an electronic device in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0053] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
[0054] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.
[0055] The method is described in the embodiments of the present disclosure, and is specifically described through the following embodiments.
[0056] Figure 1 A flowchart of a service processing method in an embodiment of the present disclosure is shown. The method can be executed by any electronic device with computing and processing capabilities. In the following process, the electronic device is taken as an example of a terminal device. Figure 1 As shown, the steps include:
[0057] S102: Receive a business question input by a user.
[0058] In a possible embodiment, when a user inputs a business problem, he or she may refer to a given business problem template for input, or may output the problem without referring to the business problem template. The business problem needs to contain the information necessary to solve the problem, that is, the problem processing conditions. For example, in a route planning problem, the problem processing conditions necessary to solve the problem need to include information such as the address and vehicle model of the shipping point and the receiving point, otherwise the problem cannot be constructed and solved.
[0059] It may also include: limiting requirements, which may include, for example, processing conditions, constraints, etc. If there are no limiting requirements, you may not enter them, and only enter the question.
[0060] Business problems can be problems in various fields, such as: path planning problems and vehicle scheduling problems in the logistics field; packing problems in the supply chain field; production problems in the manufacturing field; and home appliance work scheduling problems in the artificial intelligence field.
[0061] For example, taking the vehicle route planning problem as an example, the user's input can be: I want to ship goods from point A to outlets B, C, D, E, F, and G. There are 12.5-meter trucks at the delivery point, which cannot be overloaded and need to be delivered according to the delivery time required by the outlet. One vehicle can deliver to a maximum of 5 points. Please help me plan the delivery route.
[0062] For example, in the case of a production problem in the manufacturing industry, the user input may be: 1,000 pieces of product A and 1,500 pieces of product B need to be produced, and three types of equipment, A, B, and C, are needed to produce product A, and two types of equipment, C and D, are needed to produce product B. Please calculate the types and quantities of equipment to be configured for product A and product B respectively based on the current equipment configuration of the workshop.
[0063] For example, taking the household appliance work scheduling problem as an example, the user's input can be: I arrive home at 8 o'clock in the evening, and the indoor temperature needs to be 25 degrees. Please help me calculate the start-up time and power of the air conditioner.
[0064] S104: Call the large language model to obtain structured fields corresponding to the key business information.
[0065] In a possible embodiment, after receiving the above input question, a large language model may be called to identify the business question and obtain structured fields corresponding to the key business information.
[0066] The role of the large language model is to convert business problems input by users from a business perspective into expressions from a technical perspective.
[0067] S106: calling a pre-built storage database to query the solution data corresponding to the structured field, where the solution data includes numerical information.
[0068] S108: Convert the format of the structured fields according to the solved data to generate executable commands.
[0069] In a possible embodiment, the format of the structured fields is converted according to the solution data to obtain the intermediate fields, and the intermediate fields are converted according to a pre-built conversion program to obtain executable commands.
[0070] The function of the storage database is to store information that can be directly used by the solver, that is, the solver data, which includes numerical information. Through the solver data, some descriptions in the structured fields of the expression form of the technical perspective are converted into numerical information that can be used by the solver, and the intermediate fields are obtained. According to the pre-built conversion program, the intermediate fields are converted to obtain executable commands. The content of the executable command is in a format that the solver can support. The solver can directly solve according to the executable command.
[0071] In a possible embodiment, the structured fields are converted into executable commands that can be recognized by the solving tool, the addresses of various shipping points and receiving points and the loads of different types of vehicles are queried in the storage database, the above information is converted into numerical information, and then the executable commands are obtained for final processing of the business problem.
[0072] Exemplarily, the middle field may include: shipping location: A; receiving points: longitude and latitude information of locations such as B, C, D, E, F, G, etc., giving the specific location on the map, and querying the load capacity of a 12.5-meter truck at the shipping location at point A, which is used to allocate goods to the vehicle.
[0073] S110: calling a solution tool, performing a solution according to executable commands, and outputting a business processing result.
[0074] In a possible embodiment, the solving tool may include: a commercial solver, a custom solving program, a heuristic algorithm, etc.
[0075] The final service processing results may include: path 1: AEC; path 2: ABFG; path 3: AD.
[0076] The business is processed and the goods are delivered according to the above business processing results.
[0077] Through the method disclosed in the present invention, for various complex business problems that can only be solved by using operations research, the business problems can be processed through a large language model, which can reduce the dependence on engineers and reduce the cost of use. It should be noted that operations research algorithms are generally used in, for example, warehouse network planning, vehicle scheduling, replenishment allocation, etc., which belong to operations optimization problems. Appropriate strategies are obtained through operations research algorithms. This type of problem requires engineers to abstract practical problems into mathematical problems, build mathematical models, and solve them.
[0078] The algorithm disclosed in the present invention uses the semantic analysis capability of a large language model for text to extract valid business-critical information and identify relevant constraints. In addition, the algorithm disclosed in the present invention first extracts the structured fields corresponding to the business-critical information, converts part of the information in the structured fields into numerical information, and then converts the information into executable commands that can be processed by the solving tool. This not only improves the accuracy of analyzing business problems, but also reduces the requirements for the computing power of the large language model and the difficulty of corpus processing.
[0079] It should be noted that there are necessary conditions for processing and solving business problems. These conditions can be extracted based on user input or queried based on user input as clues.
[0080] For example, when a user inputs a business problem, he or she may not have entered all the necessary conditions for solving the problem. For example, for a production problem in the manufacturing industry, based on the current configuration of the workshop, if the user has not entered any information, it is necessary to query the database based on the user's input. For example, based on the historical information previously entered by the user, the query can be made to find which workshop the production problem is for. If the query cannot be found, a prompt message can be given to ask the user to enter which production workshop it is, or to enter the equipment currently available in the workshop.
[0081] In a possible embodiment, an input interface can be displayed in the terminal device in the form of an application, and the input interface includes an area for inputting business questions. After the user inputs the business question, the terminal device calls the large language model to be used from the cloud to process the business question, obtains the business processing result, and transmits it back to the application of the terminal device, and displays the output interface, and displays the business processing result on the output interface. It brings convenience to users from the perspective of visualization and use, and the large language model is waiting to be called in the cloud, which can reduce the memory requirements of the terminal device.
[0082] In a possible embodiment, since various business problems may correspond to different fields, the domain knowledge involved in the business problems may be different. In order to focus more on the accuracy of business processing in different fields, Figure 2 A flowchart for processing business problems is shown, such as Figure 2 As shown, including:
[0083] S202: Calling a text classification model to judge the domain to which the business problem belongs, and determining a target domain corresponding to the business problem.
[0084] S204: Call the large language model of the target domain to obtain structured fields corresponding to the key business information.
[0085] In a possible embodiment, the text classification model can be used to determine the domain to which the business problem belongs, and the corresponding large language model can be called to obtain structured fields. The fields used in the training of large language models in different fields and the content corresponding to the fields are different, so that more professional processing methods can be obtained for each different field and type of business problem, and problem processing conditions and constraints can be more accurately obtained from actual problems.
[0086] In a possible embodiment, the method of calling the large language model in S104 to obtain the structured fields corresponding to the business-critical information may include: extracting the business-critical information from the business problem and outputting the structured fields.
[0087] Exemplarily, textual structured processing may be performed based on key business information in the business problem to obtain structured fields, thereby obtaining structured fields of natural language type.
[0088] Taking the input of the above-mentioned path planning problem as an example, the structured fields obtained may include: Problem type: path planning. Shipping location: A. Receiving outlets: B, C, D, E, F, G. Vehicle: 12.5-meter truck. Constraints: Delivery time constraint: delivery time required by outlets; Cargo capacity constraint: no overweight; Maximum delivery point constraint: 5. Problem optimization: return to the shipping point after delivery.
[0089] Exemplarily, other methods may also be used to obtain structured fields. Figure 3 A flow chart for obtaining a structured field is shown, comprising the following steps:
[0090] S302: Calling a large language model to extract key business information from business problems.
[0091] S304: Mapping business-critical information to an intermediate layer, where the intermediate layer is constructed based on a programming language.
[0092] The middle layer includes key fields related to business issues, and has corresponding field definitions for each business issue.
[0093] S306: converting the business-critical information to obtain a structured field, where the information included in the structured field is a programming language field obtained based on the programming language.
[0094] The programming language can include multiple programming languages, such as json format, Markdown, CSV, etc. Among them, the programming language in json format has the clearest structure and is the most convenient to use.
[0095] In a possible embodiment, the business-critical information is mapped to the middle layer, and the plain text type business-critical information is converted into structured fields corresponding to programming language information according to key fields related to the business problem included in the middle layer.
[0096] The resulting structured fields in JSON format can be as follows:
[0097] {'type':VRP,'Depots':['A'],'Customers':['B','C','D','E','F','G'],'Vehicles':['12.5 meter truck'],Constraints' :{'TimeCons':1,'CapacityCons':1,'MaxnodeCons':1,'maxnodes':5,'Objective':{'return':1,'obj_type':3}}.
[0098] Among them, type indicates the type of problem; VRP means vehicle routing planning. Depots refers to the shipping point, Customers is the receiving point, Vehicles is the vehicle type, Constraints is the constraint of the problem, TimeCons is the delivery time constraint, CapacityCons is the cargo capacity constraint, MaxnodeCons is the maximum distribution point constraint, maxnodes is the maximum number of distribution points, Objective is the problem optimization goal, return means returning to the shipping point after delivery, and obj_type is the type of optimization goal.
[0099] There are many types of optimization objectives, such as minimizing the number of vehicles used, minimizing transportation time, minimizing transportation costs, etc. 3 represents the type of optimization objective of returning to the shipping point after delivery.
[0100] It should be noted that the structured fields in the above JSON format are only examples, not the full text.
[0101] As can be seen from the above example, the structured fields can include the type of problem, the addresses of the shipping point and each receiving point, the vehicle type, and various constraints.
[0102] In a possible embodiment, extracting key business information from a business problem may include the following methods:
[0103] For example, based on the semantic analysis capability of the large language model, key business information can be extracted directly from the full text of the business question.
[0104] Exemplarily, the business problem type can be identified based on the business problem, and business-critical information can be extracted from the business problem based on the business problem type. The business-critical information includes at least the following information: problem processing conditions, and may also include problem constraint conditions and problem optimization conditions, etc.
[0105] Since the corresponding field contents of different types of business problems may be quite different, we first define the business problem type through the large language model. After defining the business problem type, we can call out the field template of the corresponding structured field. Different field templates have different corresponding field definitions. In this way, we can better obtain the structured fields of different types of business problems.
[0106] For example, for a route planning problem, the field template must include the shipping point and the receiving point. For a packing problem, the field template must include the total amount of goods to be packed, and so on.
[0107] In a possible implementation, the large language model used above may be obtained through training. The specific training method may be as follows: Figure 4 A flowchart of a large language model training process is shown, such as Figure 4 As shown, the following steps are included:
[0108] S402: Randomly generate a structured training set.
[0109] S404: Generate a business problem training set corresponding to the structured training set according to the structured training set.
[0110] In a possible embodiment, a structured training set is randomly generated in the above manner and input into a corresponding business problem template to obtain a business problem training set corresponding to the structured training set.
[0111] In another possible embodiment, the structured training set may be processed according to other models with strong natural language processing capabilities to generate a corresponding business problem training set.
[0112] S406: Construct an input-output pair training set according to the structured training set and the business problem training set.
[0113] S408: Train the large language model to be trained according to the input and output pairs of the training set to obtain a trained large language model.
[0114] After the large language model is trained, it can be stored in the cloud and wait for call.
[0115] Since the method in the present disclosure converts the large language model into structured fields according to the business problem. Therefore, during the training process, the strong correlation between the structured fields and the business problem is ensured to complete the training. The structured training set is randomly generated in the above manner to obtain the corresponding business problem training set, so as to construct the input-output pair training set, which can ensure the high degree of automation of the input-output pair construction, without too much manual intervention, and the expected acquisition source of the training is relatively simple. In addition, the matching focus between the input and the output is strong, the training difficulty is low during training, and the ability requirements for the large language model do not need to be too high. A small large language model can be used for training and put into subsequent use. The small large language model has low computing resource requirements and usually requires less computing resources, such as memory and processor power, and is easier to deploy and use on devices with limited computing power. Better generalization ability, lower energy consumption, easy maintenance and updating, and better interpretability.
[0116] Small large language models can include: ChatGLM model.
[0117] It should be noted that for different technical fields, input-output pair training sets for different technical fields can be constructed, and several small large language models can be trained separately. This can improve the accuracy of the large language model in identifying key information in business problems and obtaining structured fields.
[0118] In one possible embodiment, Figure 5 A schematic diagram of a business processing method in the preparation process and the use process is shown, such as Figure 5 As shown, the preparation process includes the training process of the large language model, the design process of the storage database and the design process of the format conversion program.
[0119] The design process of the storage database may include: storing the numerical information corresponding to various structured fields in the database to construct the storage database.
[0120] For example, for the problem of route planning, the addresses of all outlets involved in logistics are stored in the database, and all vehicles involved are stored in the database. For the problem of box packing, all models of all cargo boxes in the system are stored.
[0121] The design process of the storage database may include: designing a language conversion program for the structured field. If the structured field is programming language information obtained based on the programming language, then designing a language conversion program that converts the programming language into numerical information and then obtains executable commands.
[0122] Usage process: After receiving the business problem input by the user, the language parsing capability of the large language model is used to extract key business information and map it to the middle layer to obtain structured fields. Next, the Agent extracts the solution information from the storage database to complete the information in the middle layer, generates executable commands, and then uses the solution tool to generate business processing results. Solution information is the information used by the solution tool when solving business problems through mathematical models. The business processing result is the solution to the problem involved in the mathematical expression in the mathematical model.
[0123] Among them, there is a constructed mathematical model in the solving tool, and the executable command is processed through the constructed mathematical model.
[0124] In a possible embodiment, a solution tool for solving corresponding business problems is called to solve the business problems.
[0125] In a possible embodiment, for a business problem, the field to which the business problem belongs can be identified, and a solving tool in the corresponding field can be called to solve the problem. The business problem can be solved using a mathematical model that has been built in the solving tool.
[0126] In a possible embodiment, the service processing method in the present disclosure is implemented based on a large language model agent (LLM-Agent). Figure 6 A schematic diagram of the structure of a large language model agent is shown, Figure 6 As shown in the figure, it includes a large language model, an intermediate layer, a storage database, and a solving tool. The intermediate layer is built based on a programming language, which includes various key fields related to business problems.
[0127] The present disclosure is implemented based on LLM-Agent, which not only fully utilizes the powerful language capabilities of the large language model, but also transforms the solution algorithm into a tool that it can use. By reasonably designing the Agent architecture and adding an intermediate layer, complex tasks are decomposed to achieve efficient conversion from business requirements described from a business perspective to clearly structured and concise field content, and then to a mathematical model, and solve the corresponding problems.
[0128] Directly generating codes that can be used by solving tools through a large language model has very high requirements for the training corpus of the large language model. Not only does it require a large amount of high-quality corpus, but the content of the corpus generally needs to be relevant academic papers and high-quality documents. Otherwise, the output accuracy of the trained large language model is low. In addition, the large language model is used to directly recognize and convert multiple languages such as natural language, mathematical expressions, and codes. The training difficulty is also relatively large, and the format requirements of mathematical expressions and codes are relatively strict. Slight errors will cause the final code to fail to run and fail to give correct results. The above problems can be avoided by the method disclosed in the present invention.
[0129] In a possible embodiment, the present disclosure also provides a flowchart of another business processing method, such as Figure 7 As shown, the following steps are included:
[0130] S702: Receive a business question input by a user.
[0131] S704: Calling a text classification model to judge the domain to which the business problem belongs, and determining a target domain corresponding to the business problem.
[0132] S706: Call the large language model of the target domain to extract key business information in the business problem.
[0133] S708: Mapping business-critical information to an intermediate layer, where the intermediate layer is built based on a programming language.
[0134] S710: Convert the business-critical information to obtain a structured field, where information included in the structured field is a programming language field obtained based on the programming language.
[0135] S712: calling a pre-built storage database to query the solution data corresponding to the structured field; the solution data includes numerical information.
[0136] S714: Convert the format of the structured fields according to the solved data to generate executable commands.
[0137] S716: Calling a solving tool, solving according to the executable command, and outputting the business processing result.
[0138] Based on the same inventive concept as the above method embodiment, the embodiment of the present application also provides a business processing device. Figure 8 A schematic diagram of the structure of a service processing device provided in an embodiment of the present application is shown.
[0139] The device 80 includes: a receiving unit 801, which is used to receive a business problem input by a user; an extraction unit 802, which is used to call a large language model to obtain structured fields corresponding to business-critical information; a query unit 803, which is used to call a pre-built storage database to query solution data corresponding to the structured fields; a conversion unit 804, which is used to convert the format of the structured fields according to the solution data and generate executable commands; the executable commands are numerical information; a processing unit 805, which is used to call a solution tool, solve according to the executable command, and output the business processing result.
[0140] It will be appreciated by those skilled in the art that various aspects of the present invention may be implemented as a system, method or program product. Therefore, various aspects of the present invention may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as a "circuit", "module" or "system".
[0141] Refer to the following Fig. 9 An electronic device 900 according to this embodiment of the present invention will be described. Fig. 9 The electronic device 900 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.
[0142] like Fig. 9 As shown, the electronic device 900 is in the form of a general computing device. The components of the electronic device 900 may include but are not limited to: at least one processing unit 910, at least one storage unit 920, and a bus 930 connecting different system components (including the storage unit 920 and the processing unit 910).
[0143] The storage unit stores program codes, which can be executed by the processing unit 910, so that the processing unit 910 executes the steps according to various exemplary embodiments of the present invention described in the above “Exemplary Method” section of this specification.
[0144] The storage unit 920 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 9201 and / or a cache storage unit 9202 , and may further include a read-only storage unit (ROM) 9203 .
[0145] The storage unit 920 may also include a program / utility 9204 having a set (at least one) of program modules 9205, such program modules 9205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.
[0146] Bus 930 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.
[0147] The electronic device 900 may also communicate with one or more external devices 940 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 900, and / or may communicate with any device that enables the electronic device 900 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 950. Furthermore, the electronic device 900 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 960. As shown, the network adapter 960 communicates with other modules of the electronic device 900 via a bus 930. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 900, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0148] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0149] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible implementations, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary embodiments of the present invention described in the above "Exemplary Method" section of the present specification.
[0150] A program product for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.
[0151] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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 above.
[0152] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0153] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
[0154] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).
[0155] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.
[0156] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.
[0157] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0158] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.
Claims
1. A business processing method, It is characterized in that The method comprises: Receive business questions input by users; Call the large language model to obtain the structured fields corresponding to the key business information; Calling a pre-built storage database to query the solution data corresponding to the structured field; the solution data includes numerical information; Performing format conversion on the structured fields according to the solution data to generate executable commands; The solving tool is called, the solving is performed according to the executable command, and the business processing result is output.
2. The method according to claim 1, It is characterized in that The calling of the large language model to obtain structured fields corresponding to the business key information includes: Calling a large language model to extract key business information from the business problem; Mapping business-critical information to an intermediate layer; the intermediate layer is built based on a programming language; The business-critical information is converted to obtain a structured field; the information included in the structured field is a programming language field obtained based on a programming language.
3. The method according to claim 2, It is characterized in that The extracting of key business information from the business problem includes: Based on the business problem, identify the type of business problem; According to the type of the business problem, business key information is extracted from the business problem; the business key information at least includes the following information: problem processing conditions.
4. The method according to claim 1, It is characterized in that The training process of the large language model includes: Randomly generate structured training sets; According to the structured training set, generating a business problem training set corresponding to the structured training set; Constructing an input-output pair training set according to the structured training set and the business problem training set; According to the input-output pair training set, the large language model to be trained is trained to obtain the trained large language model.
5. The method according to claim 1, It is characterized in that After receiving the business question input by the user, the method further includes: Calling a text classification model to judge the domain to which the business problem belongs, and determining the target domain corresponding to the business problem; The large language model of the target domain is called to obtain structured fields corresponding to the key business information.
6. The method according to claim 1, It is characterized in that The format conversion of the structured field according to the solution data to generate an executable command includes: Performing format conversion on the structured field according to the solved data to obtain an intermediate field; According to a pre-built conversion program, the intermediate fields are converted to obtain executable commands.
7. A service processing device, It is characterized in that include: A receiving unit, used for receiving a business question input by a user; The extraction unit is used to call the large language model to obtain the structured fields corresponding to the key business information; A query unit, used to call a pre-built storage database to query the solution data corresponding to the structured field; the solution data includes numerical information; A conversion unit, used for performing format conversion on the structured field according to the solution data to generate an executable command; The processing unit is used to call the solving tool, solve according to the executable command, and output the business processing result.
8. An electronic device, It is characterized in that include: processor; and a memory for storing executable instructions for the processor; The processor is configured to perform the method of any one of claims 1 to 6 by executing the executable instructions.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.
10. A computer program product, include: A computer program or instruction, characterized in that when the computer program or instruction is executed by a processor, the method described in any one of claims 1 to 6 is implemented.
Citation Information
Patent Citations
Large language model-based operation planning optimization method and system, electronic equipment and storage medium
CN117273005A
Large language model-based operation planning optimization method and device, medium and program product
CN118132708A
