Method and related device for acquiring mathematical model, and operations optimization method

Abstract

A method and device for acquiring a mathematical model, and an operations optimization method. The method allows artificial intelligence technology to be used for acquiring a mathematical model corresponding to an operations optimization problem. The method comprises: outputting at least one question, and acquiring an answer to each question, wherein the answer to each question is used for obtaining first description information, the first description information is description information that is used for describing an operations optimization problem, and a first question is a question used to acquire the description information about the operations optimization problem; obtaining first information on the basis of the first description information; and inputting the first information into a machine learning model, so as to obtain a mathematical model, wherein the mathematical model is used for solving the operations optimization problem, and the mathematical model comprises an objective function and constraint conditions. By means of the aforementioned solution, the labor cost consumed during the process of acquiring the aforementioned mathematical model is greatly reduced; and the solution can be adapted to acquiring mathematical models in various fields and has a high generalization performance.

Description

A method for obtaining a mathematical model, related equipment, and operations optimization method

[0001] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office on October 31, 2023, with application number 202311440915.X and application name “A method for obtaining a mathematical model and related equipment and an operations optimization method”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the field of artificial intelligence, and in particular to a method for acquiring a mathematical model and related equipment as well as an operations optimization method. Background Art

[0003] Operational optimization problems refer to making the best decision that meets business goals while considering certain constraints. When solving such problems in practice, it is necessary to establish a mathematical model corresponding to the operational optimization problem and solve the aforementioned mathematical model through a computer.

[0004] At present, in order to obtain the mathematical model corresponding to the operations optimization problem, specifically, all operations optimization problems can be divided into multiple fields, that is, multiple fields to which the mathematical model to be established can belong are obtained. In each field included in the aforementioned multiple fields, all objective functions of the field, the text description corresponding to each objective function, all constraints of the field, and the text description corresponding to each constraint must be abstracted. After obtaining the text description corresponding to the operations optimization problem input by the user, the text description input by the user is matched with the text description of each objective function for similarity, and the text description input by the user is matched with the text description of each constraint for similarity, so as to obtain an objective function and at least one constraint that meet the similarity conditions with the text description input by the user. The mathematical model corresponding to the operations optimization problem includes the aforementioned objective functions and constraints.

[0005] However, since the labor cost of generating "all objective functions, text descriptions corresponding to each objective function, all constraints, and text descriptions corresponding to each constraint" corresponding to a single field is very high, the labor cost of establishing "all objective functions, text descriptions corresponding to each objective function, all constraints, and text descriptions corresponding to each constraint" corresponding to multiple fields is even higher, which means that the generalization of this solution is relatively poor.

[0006] Summary of the Invention

[0007] The embodiments of the present application provide a method for obtaining a mathematical model, related equipment, and an operations optimization method. In this solution, for different fields, there is no need to pre-generate multiple objective functions, text descriptions corresponding to each objective function, multiple constraints, and text descriptions corresponding to each constraint in each field, which greatly reduces the manpower cost consumed in the process of obtaining the mathematical model, and can be adapted to obtain mathematical models in various fields, with high generalization. In addition, in this solution, descriptive information is obtained in the form of answers to at least one of the aforementioned questions, which is conducive to obtaining more accurate descriptive information for the operations optimization problem, and thus is conducive to improving the adaptability between the mathematical model output by the machine learning model and the operations optimization problem, that is, it is conducive to obtaining a more accurate mathematical model.

[0008] The embodiments of this application provide the following technical solutions:

[0009] In the first aspect, an embodiment of the present application provides a method for obtaining a mathematical model, which can use artificial intelligence technology to obtain a mathematical model corresponding to an operations optimization problem. The method includes: in order to obtain the descriptive information of the first operations optimization problem, the first device can output at least one first question to the user, use the content of the user's reply to each of the at least one first question as at least one answer, and obtain the at least one answer. The answer to each of the at least one question is used to obtain the first descriptive information, and the first descriptive information is used to describe the first operations optimization problem. The first device determines the first information based on the first descriptive information, and then inputs the first information into a machine learning model (hereinafter referred to as the "first machine learning model" for the convenience of description) to obtain a first mathematical model. The first mathematical model includes an objective function and constraints. The first mathematical model is generated by using the first machine learning model in order to solve the first operations optimization problem, that is, the solution of the first mathematical model can solve the first operations optimization problem.

[0010] In this application, operations optimization problems can be replaced by other descriptions, such as mathematical programming problems and decision optimization problems.

[0011] Exemplarily, the first information may include the first descriptive information and other information, or the first information and the first descriptive information may include the same information. Exemplarily, the other information may include prompt information, such as the prompt information may include "Please generate a mathematical model corresponding to the operations optimization problem based on the subsequent descriptive information." Another example is the prompt information may include "Please generate a mathematical model based on the first descriptive information. The first descriptive information includes:" and so on. The prompt information may also be other information. The first device may be a client device, such as a terminal device or an edge device; or the first device may be a cloud device, such as a server or a server cluster.

[0012] For example, when the first device is a client device, if the first machine learning model is not deployed in the first device, then "the first device inputs the first information into the machine learning model" can be understood as the first device sending the first information to the device on which the first machine learning model is deployed, and "the first device obtains the first mathematical model" can be understood as the first device receiving the first mathematical model sent by the device on which the first machine learning model is deployed.

[0013] In this implementation, in order to obtain the descriptive information of the operations optimization problem, at least one problem is abstracted, and the answer to the at least one problem is obtained. After obtaining the first information based on the first descriptive information, the first information is input into the machine learning model to obtain the first mathematical model corresponding to the first operations optimization problem. In this solution, for different fields, there is no need to pre-generate multiple objective functions, text descriptions corresponding to each objective function, multiple constraints, and text descriptions corresponding to each constraint in each field, which greatly reduces the manpower cost consumed in the process of obtaining the mathematical model corresponding to the operations optimization problem, and can be adapted to obtain mathematical models corresponding to operations optimization problems in various fields, and has high generalization. In addition, in this solution, the descriptive information of the operations optimization problem is obtained by obtaining the answer to the aforementioned at least one question, which to a certain extent alleviates the problem that the user cannot accurately determine the first operations optimization problem, and is conducive to obtaining more accurate descriptive information, and thus is conducive to improving the adaptability between the mathematical model output by the machine learning model and the operations optimization problem, that is, it is conducive to obtaining a more accurate mathematical model.

[0014] In one possible implementation, the method also includes: the first device displays the answer options corresponding to each first question through a display screen. When the first device obtains a selection operation input for the first answer option, the correct answer corresponding to the first answer option can be obtained, that is, the correct answer to at least one of the first questions is obtained.

[0015] In one possible implementation, after the first device outputs the first question to the user, the method further includes: the first device obtaining a first answer entered by the user via a text box, and then obtaining a first matching result between the first answer and all correct answers to the first question; the first matching result indicating which correct answers to the first question satisfy a first matching condition between the first answer and the first answer, or the first matching result indicating a similarity between the first answer and each correct answer among all correct answers to the first question. The first device may determine a correct answer to the first question to display to the user based on the first matching result.

[0016] In one possible implementation, a first question set is pre-stored in the first device, and the at least one first question is determined based on the pre-stored first question set. In this implementation, the pre-stored first question set enables the determination of which first questions to output to the user, thereby speeding up the process of determining which first questions to output and facilitating a more efficient acquisition of the first mathematical model.

[0017] In one possible implementation, the first question set is represented by a directed graph data format, which indicates the order of appearance of at least two different questions in the first question set. In this implementation, the questions included in the first question set are stored in the directed graph data format, which can indicate the order of appearance of at least two different questions in the first question set. Since different questions in at least one question corresponding to the first operations optimization problem may have a correlation relationship, the use of the directed graph data format can better reflect the logical relationship between different questions, making the process of asking questions to the user more logical and also helping to avoid outputting useless questions to the user, thereby improving the user stickiness of this solution.

[0018] In one possible implementation, in order to obtain the descriptive information of the first operations research optimization problem, the first problem set used may include multiple target problems, and the aforementioned multiple target problems may include at least one of the first problems output to the user. Optionally, the aforementioned multiple target problems may also include a second problem other than the first problem, that is, the second problem is also the problem used to obtain the descriptive information of the first operations research optimization problem, and the aforementioned second problem corresponds to the second descriptive information input by the user.

[0019] In one possible implementation, before the first device outputs the first question to the user for the first time, or in the process of the first device outputting the first question to the user multiple times, it can also obtain second descriptive information input by the user, and the second descriptive information carries the descriptive information of the first operations optimization problem; then the first device can also determine which second questions of the multiple target questions included in the first question set carry the answers in the second descriptive information, and then determine which questions of the multiple target questions included in the first question set have not yet been answered based on the multiple target questions and the second questions that have already been answered, and the aforementioned questions that have not yet been answered are used as the first questions that need to be output to the user. Correspondingly, the first device can determine the first information based on the first descriptive information, which may include: the first device can merge the first descriptive information and the second descriptive information to obtain the union of the first descriptive information and the second descriptive information (hereinafter referred to as "updated first descriptive information" for the convenience of description), and then can determine the first information based on the aforementioned updated first descriptive information, the first information includes the updated first descriptive information, that is, the first information includes the aforementioned first descriptive information and the second descriptive information.

[0020] In one possible implementation, after outputting at least one first question to the user, the first device may also obtain second descriptive information input by the user, and the second descriptive information may carry descriptive information of the first operations optimization problem; then the first device determines the first information based on the first descriptive information, which may include: the first device determines the first information based on the first descriptive information and the second descriptive information; optionally, the first device may merge the first descriptive information and the second descriptive information to obtain updated first descriptive information, and then determine the first information based on the aforementioned updated first descriptive information, the first information including the updated first descriptive information, that is, the first information includes the aforementioned first descriptive information and the second descriptive information.

[0021] In one possible implementation, the method is applied to a first device, in which at least one case is stored, and the target case is any one of the at least one case, and the target case includes descriptive information and a mathematical model corresponding to the operations optimization problem; it should be understood that the concept of the aforementioned "descriptive information" is similar to the concept of the aforementioned "first descriptive information", both of which are descriptive information used to describe the operations optimization problem, and the concept of the aforementioned "mathematical model" is similar to the concept of the aforementioned "first mathematical model", except that the descriptive information and mathematical model included in the target case are already stored in the first device, the first descriptive information is being obtained, and the first mathematical model is generated by the first machine learning model. Wherein, the first device determines the first information based on the first descriptive information, including: the first device determines at least one first case from the at least one case based on the similarity between the first descriptive information and the descriptive information included in each case in the at least one case; and then the first device can obtain the first information based on the first descriptive information and the at least one first case.

[0022] In which, the similarity between the first descriptive information and the descriptive information included in the first case satisfies a preset condition; exemplarily, at least one first case includes K first cases, and the preset condition may include: the similarity between the first descriptive information and the descriptive information included in each first case is greater than or equal to a similarity threshold, and / or, the K descriptive information included in the K first cases are the K descriptive information most similar to the first descriptive information among all the descriptive information included in at least one case.

[0023] In this implementation, at least one first case that best matches the first descriptive information is obtained from at least one pre-stored case, and the similarity between the first descriptive information and the descriptive information included in each first case meets the preset conditions, and then each first case and the first descriptive information are used as inputs of the first machine learning model, that is, each first case is used as a reference case for the first machine learning model, which is beneficial to assisting the first machine learning model to generate a more accurate first mathematical model, and is also beneficial to improving the efficiency of the first machine learning model in the process of generating the first mathematical model.

[0024] In one possible implementation, the method further includes: the first device obtains second information, the second information including at least one of the summary and keywords of the first descriptive information; and then determines the similarity between the second information and each descriptive information included in the at least one case, wherein the similarity between the second information and each descriptive information included in the at least one case is used as the similarity between the first descriptive information and each descriptive information included in the at least one case. In this implementation, since the first descriptive information may include text information input by the user, and the text information input by the user may carry invalid information, first obtaining the summary and / or keywords of the first descriptive information, that is, filtering out the invalid information in the first descriptive information, and then generating the similarity between the summary and / or keywords of the first descriptive information and the fourth descriptive information is conducive to obtaining more accurate similarity information, and thus is conducive to obtaining a more matching first case.

[0025] In one possible implementation, the method may further include: the first device may further obtain third information corresponding to each descriptive information included in at least one case (hereinafter referred to as "fourth descriptive information" for the convenience of description), that is, obtaining at least one third information corresponding one-to-one to at least one fourth descriptive information, and each third information includes a summary and / or keywords of the fourth descriptive information.

[0026] The first device determines the similarity between the second information and each fourth descriptive information included in at least one case, including: the first device obtains initial feature information of the second information and initial feature information of each third information, and then generates third similarity information; wherein the third similarity information includes the similarity between the initial feature information of the second information and the initial feature information of each third information, and the similarity between the initial feature information of the second information and the initial feature information of each third information can be used as the similarity between the second information and each fourth descriptive information, that is, it can be used as the similarity between the first descriptive information and each fourth descriptive information.

[0027] In one possible implementation, the first device obtains the first information based on the first descriptive information and at least one first case, which may include: the first device may merge at least one first case and the first descriptive information to obtain the first information, that is, the first information includes the first descriptive information and each first case; exemplarily, the first information may include the first descriptive information and prompt information, then the first device may combine at least one first case into the aforementioned prompt information to obtain the first information.

[0028] In one possible implementation, before the first device obtains the first descriptive information corresponding to at least one question, the method further includes: the first device may determine the field to which the mathematical model to be established belongs from at least one field to which the mathematical model to be established can belong (for the convenience of description, the "field to which the mathematical model to be established belongs" will be referred to as the "first field" hereinafter); and then, based on the first field to which the mathematical model to be established belongs. The first device may pre-store at least one set of questions corresponding to the aforementioned at least one field. After determining the first field to which the mathematical model to be established belongs, a first set of questions corresponding to the first field may be determined from the at least one set of questions, and then the at least one question may be determined based on the first set of questions.

[0029] In this implementation, all operations research and optimization problems are divided into multiple fields, that is, multiple fields to which the mathematical model to be established can belong are obtained. After determining the field to which the mathematical model to be established belongs, based on the field to which the aforementioned mathematical model to be established belongs, at least one problem used to obtain detailed description information of the first operations research and optimization problem is determined, that is, in order to obtain description information of operations research and optimization problems in different fields, different problems are used. It can be seen from the above description that in this solution, more refined management of the problems used to obtain description information of operations research and optimization problems is carried out, which is conducive to obtaining more accurate description information, and further conducive to obtaining more accurate mathematical models.

[0030] In one possible implementation, the first device determines the field to which the mathematical model to be established belongs, including: the first device provides the user with multiple fields to which the operations optimization problem can belong, that is, provides the user with multiple fields to which the mathematical model to be established can belong, and after obtaining the user's selection operation input for the first field among the aforementioned multiple fields, the first field to which the mathematical model to be established belongs can be determined.

[0031] In one possible implementation, the first device determines the field to which the mathematical model to be established belongs, including: the first device obtains third descriptive information input by the user, and based on the aforementioned third descriptive information, uses the second machine learning model to determine the first field to which the mathematical model to be established belongs; the "third descriptive information" can be understood as the background description information of the first operations optimization problem.

[0032] In one possible implementation, the domain to which the mathematical model to be established belongs includes any of the following: location selection, scheduling, order fulfillment, supply chain, packing, transportation, resource allocation, revenue management, or production planning. This implementation lists multiple domains to which the mathematical model to be established can belong, significantly expanding the application scenarios of this solution and increasing its implementation flexibility.

[0033] In a second aspect, an embodiment of the present application provides a method for obtaining a mathematical model, which can use artificial intelligence technology to obtain a mathematical model corresponding to an operations optimization problem. At least one case is stored in a first device, and any one of the at least one case includes descriptive information and a mathematical model corresponding to the operations optimization problem. In this method, the first device obtains first descriptive information, which is descriptive information used to describe the first operations optimization problem; based on the similarity between the first descriptive information and each descriptive information included in the at least one case, at least one second mathematical model is determined from at least one mathematical model included in the at least one case, the second mathematical model belongs to the first case in the at least one case, and the similarity between the first descriptive information and the descriptive information included in the first case meets a preset condition; based on the first descriptive information and the at least one second mathematical model, first information is obtained; the first information is input into a machine learning model to obtain a first mathematical model, the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations optimization problem.

[0034] In one possible implementation, the first device obtains first description information, including: the first device obtains first description information corresponding to at least one first question, the first description information includes an answer to each first question in the at least one first question, and the answer to each first question in the at least one first question is used to obtain the first description information, wherein the first question is a question used to obtain the description information of the first operations optimization problem.

[0035] In the second aspect of this application, the first device can also be used to execute the steps performed by the first device in the first aspect and various possible implementation methods of the first aspect. The specific implementation methods of the steps in the second aspect, the meanings of the terms and the beneficial effects brought about can all be referred to the first aspect and will not be repeated here.

[0036] In a third aspect, an embodiment of the present application provides a device for acquiring a mathematical model, which can use artificial intelligence technology to acquire a mathematical model corresponding to an operations optimization problem. The device for acquiring a mathematical model includes: an output module for outputting at least one first question; a processing module for acquiring at least one answer based on at least one first question, and the at least one answer is used to obtain first descriptive information, wherein the first descriptive information is descriptive information used to describe the first operations optimization problem, and the first question is a problem used to obtain the descriptive information of the first operations optimization problem; a determination module for determining the first information based on the first descriptive information; the processing module is also used to input the first information into a machine learning model to obtain a first mathematical model, the first mathematical model including an objective function and constraints, and the first mathematical model is used to solve the first operations optimization problem.

[0037] In the third aspect of this application, the mathematical model acquisition device is also used to execute the steps performed by the first device in the first aspect and various possible implementation methods of the first aspect. The specific implementation methods of the steps in the third aspect, the meanings of the terms and the beneficial effects brought about can all be referred to the first aspect and will not be repeated here.

[0038] In a fourth aspect, an embodiment of the present application provides a device for acquiring a mathematical model, which can use artificial intelligence technology to acquire a mathematical model corresponding to an operations optimization problem. The device for acquiring a mathematical model is applied to a first device, and at least one case is stored in the first device. Any one of the at least one case includes descriptive information and a mathematical model corresponding to the operations optimization problem. The device for acquiring a mathematical model includes: an acquisition module for acquiring first descriptive information, where the first descriptive information is descriptive information used to describe the first operations optimization problem; a determination module for determining at least one second mathematical model from at least one mathematical model included in at least one case based on the similarity between the first descriptive information and each descriptive information included in the at least one case, where the second mathematical model belongs to the first case in the at least one case, and the similarity between the first descriptive information and the descriptive information included in the first case meets a preset condition; a processing module for obtaining first information based on the first descriptive information and the at least one second mathematical model; the processing module is also used to input the first information into a machine learning model to obtain a first mathematical model, where the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations optimization problem.

[0039] In the fourth aspect of this application, the mathematical model acquisition device is also used to execute the steps performed by the first device in the second aspect and various possible implementation methods of the second aspect. The specific implementation methods of the steps in the fourth aspect, the meanings of the terms and the beneficial effects brought about can all be referred to the second aspect and will not be repeated here.

[0040] In the fifth aspect, an embodiment of the present application provides a device including a processor and a memory, wherein the processor is coupled to the memory, the memory is used to store programs; and the processor is used to execute the programs in the memory, so that the device executes the method for obtaining the mathematical model of the first aspect or the second aspect mentioned above.

[0041] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer executes the method of the first or second aspect mentioned above.

[0042] In a seventh aspect, an embodiment of the present application provides a computer program product, which includes a program. When the program is run on a computer, it enables the computer to execute the method of the first aspect or the second aspect mentioned above.

[0043] In the eighth aspect, an embodiment of the present application provides an operations research optimization method, which includes: after obtaining the first mathematical model, the second device can solve the first mathematical model to obtain a solution result of the first mathematical model, wherein the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem. The first mathematical model is obtained based on the method provided in the first or second aspect above.

[0044] In a ninth aspect, the present application provides a chip system, which includes a processor for supporting the implementation of the functions involved in the above aspects, such as sending or processing the data and / or information involved in the above methods. In one possible design, the chip system also includes a memory, which is used to store program instructions and data necessary for the terminal device or communication device. The chip system can be composed of a chip or can include a chip and other discrete devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG1 is a schematic diagram of the structure of an artificial intelligence main framework provided in an embodiment of the present application;

[0046] FIG2 is a system architecture diagram of a data processing system provided in an embodiment of the present application;

[0047] FIG3 is a schematic diagram of a method for obtaining a mathematical model provided in an embodiment of the present application;

[0048] FIG4 is another flow chart of a method for obtaining a mathematical model according to an embodiment of the present application;

[0049] FIG5 is a schematic diagram of an interface for obtaining the “field to which the mathematical model to be established belongs” provided in an embodiment of the present application;

[0050] FIG6 is a schematic diagram of obtaining second prediction information using a large model according to an embodiment of the present application;

[0051] FIG7 is a schematic diagram of multiple target problems in the form of a directed graph provided by an embodiment of the present application;

[0052] FIG8 is a schematic diagram of “displaying a first question” and “obtaining an answer to the first question” provided in an embodiment of the present application;

[0053] FIG9 is a schematic diagram of obtaining first description information corresponding to at least one question according to an embodiment of the present application;

[0054] FIG10 is a schematic diagram of a process for determining at least one first case from at least one case according to an embodiment of the present application;

[0055] FIG11 is a schematic diagram of a process for obtaining a first mathematical model based on first information according to an embodiment of the present application;

[0056] FIG12 is a schematic diagram of outputting a first mathematical model through a display screen according to an embodiment of the present application;

[0057] FIG13 is another schematic diagram of a method for obtaining a mathematical model provided in an embodiment of the present application;

[0058] FIG14 is a schematic structural diagram of a device for acquiring a mathematical model according to an embodiment of the present application;

[0059] FIG15 is another schematic diagram of the structure of the device for obtaining a mathematical model provided in an embodiment of the present application;

[0060] FIG16 is a schematic structural diagram of a device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0062] 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 sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, and this is merely a way of distinguishing the objects of the same attributes when describing them in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0063] In the embodiments of the present application, "sending" and "receiving" indicate the direction of signal transmission. For example, "sending information to XX device" can be understood as the destination of the information being XX device, which can include direct sending through the air interface, as well as indirect sending through the air interface by other units or modules. "Receiving information from YY device" can be understood as the source of the information being YY device, which can include direct receiving from YY device through the air interface, as well as indirect receiving from YY device through the air interface from other units or modules. "Sending" can also be understood as the "output" of the chip interface, and "receiving" can also be understood as the "input" of the chip interface. In other words, sending and receiving can be performed between devices or within a device, for example, between components, modules, chips, software modules or hardware modules within the device through a bus, trace or interface. It is understandable that information may undergo necessary processing, such as encoding, modulation, etc., between the source and destination of the information, but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated.

[0064] In the embodiments of the present application, "indication" may include direct indication and indirect indication, and may also include explicit indication and implicit indication. The information indicated by a certain information (such as the indication information described below) is called information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, the information to be indicated can be directly indicated, such as the information to be indicated itself or the index of the information to be indicated. The information to be indicated can also be indirectly indicated by indicating other information, wherein there is an association between the other information and the information to be indicated; it is also possible to indicate only a part of the information to be indicated, while the other parts of the information to be indicated are known or agreed in advance, for example, the indication of specific information can be achieved with the help of the arrangement order of each information agreed in advance (such as predefined by the protocol), thereby reducing the indication overhead to a certain extent. The present application does not limit the specific method of indication. It is understandable that, for the sender of the indication information, the indication information can be used to indicate the information to be indicated, and for the receiver of the indication information, the indication information can be used to determine the information to be indicated.

[0065] First, let's describe the overall workflow of an AI system. See Figure 1, which shows a schematic diagram of the AI ​​framework. This framework will be explained from two perspectives: the "intelligent information chain" (horizontal axis) and the "IT value chain" (vertical axis). The "intelligent information chain" reflects the entire process from data acquisition to processing. For example, it could be the general process of intelligent information perception, intelligent information representation and formation, intelligent reasoning, intelligent decision-making, and intelligent execution and output. Throughout this process, data undergoes a condensed journey from "data-information-knowledge-wisdom." The "IT value chain," spanning the underlying infrastructure of human intelligence, information (provided and processed by technology), and the system's industrial ecosystem, reflects the value that AI brings to the information technology industry.

[0066] (1) Infrastructure

[0067] The infrastructure provides computing power for AI systems, enabling communication with the outside world and providing support through a basic platform. Communication with the outside world is achieved through sensors; computing power is provided by smart chips, which can specifically adopt hardware acceleration chips such as central processing units (CPUs), embedded neural network processing units (NPUs), graphics processing units (GPUs), application-specific integrated circuits (ASICs), or field programmable gate arrays (FPGAs); the basic platform includes related platform guarantees and support such as distributed computing frameworks and networks, and can include cloud storage and computing, and interconnected networks. For example, sensors communicate with the outside world to obtain data, which is then provided to the smart chips in the distributed computing system provided by the basic platform for calculation.

[0068] (2) Data

[0069] Data above the infrastructure layer represents data sources for AI. This data includes graphics, images, voice, and text, as well as IoT data from traditional devices. This includes business data from existing systems and sensor data such as force, displacement, liquid level, temperature, and humidity.

[0070] (3) Data processing

[0071] Data processing generally includes data training, machine learning, deep learning, search, reasoning, decision-making, etc.

[0072] Among them, machine learning and deep learning can symbolize and formalize data for intelligent information modeling, extraction, preprocessing, and training.

[0073] Reasoning refers to the process of simulating human intelligent reasoning in computers or intelligent systems, using formalized information to perform machine thinking and solve problems based on reasoning control strategies. Typical functions are search and matching.

[0074] Decision-making refers to the process of making decisions after intelligent information is reasoned, and usually provides functions such as classification, sorting, and prediction.

[0075] (4) General ability

[0076] After the data has undergone the data processing mentioned above, some general capabilities can be further formed based on the results of the data processing, such as algorithms or a general system, for example, translation, text analysis, computer vision processing, speech recognition, image recognition, etc.

[0077] (5) Smart products and industry applications

[0078] Smart products and industry applications refer to the products and applications of artificial intelligence systems in various fields. They are the encapsulation of the overall artificial intelligence solution, which productizes intelligent information decision-making and realizes practical application. Its application areas mainly include: smart terminals, smart manufacturing, smart transportation, smart homes, smart medical care, smart security, autonomous driving, smart cities, etc.

[0079] The method provided in this application can be applied to various application fields of artificial intelligence technology. Specifically, it can be used to automatically generate mathematical models corresponding to operational optimization problems when there are operational optimization problems in various application fields; the mathematical model acquisition method provided in this application is used to automatically generate mathematical models corresponding to operational optimization problems using machine learning models.

[0080] Exemplarily, the above mathematical model may include an objective function and constraints, and the aforementioned mathematical model may also include variables involved in the objective function and the constraints. In order to further understand the relationship between "operational optimization problems" and "mathematical models", the following is an example of a specific scenario. Here, the operational optimization problem is taken as an example of a production planning problem. The production planning problem is also called a high-level scheduling problem, which is a problem faced by manufacturing companies when conducting production and processing. When companies are producing, they need to decide how to optimally use resources to meet customer needs. The main task is to match supply with demand, and output the recommended processing decisions for each factory every day within a period of time, the recommended transportation decisions between factories, and the delivery decisions for demand. When making decisions, the goals are to maximize the delivery level and minimize the cost, and to consider constraints such as production capacity not exceeding the upper limit, inventory not exceeding the upper limit, and raw material substitutability.

[0081] For example, the key components (raw materials) involved in this generation planning problem (also known as a type of operations optimization problem) include structural component 1 and structural component 2. The generation planning problem corresponds to three demand codes: inverter A, inverter B, and inverter C. The corresponding demands include 50 inverters A, 100 inverters B, and 150 inverters C. The virtual revenue of inverter A is 1.3, the virtual revenue of inverter B is 2.1, and the virtual revenue of inverter C is 1.8. The inventory of structural component 1 is 100 pieces, and the inventory of structural component 2 is 90 pieces. Structural component 1 can be used to produce inverters A and B. Two structural components 1 are required to produce one unit of inverter A or inverter B. One unit of inverter C requires one structural component 2. Structural component 2 can replace structural component 1, but structural component 1 cannot replace structural component 2. How should we decide the production quantity of inverters A, B, and C and the allocation of structural parts 1 and 2 so as to maximize the total virtual profit of the satisfied demand?

[0082] The mathematical model corresponding to the above generation plan problem can be as follows:

[0083] variable:

[0084] (Production quantity of inverter A) x

[0085] (Production quantity of inverter B)y

[0086] (Production quantity of inverter C)

[0087] (Number of replacements where structural component 2 replaces structural component 1) r

[0088] Objective function:

[0089] (Maximization of total virtual benefits) Maximization: 1.3x+2.1y+1.8z

[0090] Constraints:

[0091] (Maximum production quantity constraint for inverter A) x <= 50

[0092] (Maximum production quantity constraint for inverter B) y <= 100

[0093] (Maximum production quantity constraint for inverter C) z <= 150

[0094] (The consumption quantity of structural component 1 cannot exceed the inventory) 2x+2y-r<=100

[0095] (The consumption quantity of structural part 2 cannot exceed the inventory) z+r<=90

[0096] (non-negative integer decision variable constraints)x,y,z,r\in\mathbb_{N}

[0097] It should be noted that the examples of "operations optimization problems" and "mathematical models" described above are only for the purpose of facilitating understanding of this solution and are not intended to limit this solution. Since the method provided in this application can be used in intelligent manufacturing, intelligent transportation, or other fields, the following examples illustrate application scenarios in multiple application fields of this application.

[0098] Application field 1: Intelligent manufacturing

[0099] For example, there may be production planning problems in the field of intelligent manufacturing. For explanations and examples of generation planning problems, please refer to the above description, which will not be repeated here. The method provided in this application can be used to obtain a mathematical model corresponding to the generation planning problem in the field of intelligent manufacturing.

[0100] Application field 2: Intelligent transportation

[0101] For example, in the field of intelligent transportation, traffic signals in the transportation network can be optimized in order to improve the efficiency of the transportation system under the constraint of ensuring the safety of the transportation system. There may be an operational optimization problem corresponding to the traffic signal, and the method provided in this application can be used to obtain a mathematical model corresponding to the operational optimization problem corresponding to the traffic signal.

[0102] It should be noted that the method provided in this application can also be applied to other application scenarios. The above examples of various application scenarios of this application are only for the convenience of understanding this solution and are not used to limit this solution.

[0103] Before describing the method provided in the present application in detail, please refer to Figure 2, which is a system architecture diagram of a data processing system provided in an embodiment of the present application. In Figure 2, the data processing system 200 includes a training device 210, a database 220, an execution device 230, a data storage system 240 and a client device 250, and the execution device 230 includes a computing module 231.

[0104] Among them, the database 220 stores a training data set. During the training phase of the first machine learning model 201, the training device 210 generates the first machine learning model 201 and iteratively trains the first machine learning model 201 using the training data set to obtain the first machine learning model 201 that has performed the training operation. The first machine learning model 201 can be specifically expressed as a neural network, or as a non-neural network model. In the embodiment of the present application, only the first machine learning model 201 expressed as a neural network is used as an example for explanation. Furthermore, when the first machine learning model 201 is expressed as a neural network, the first machine learning model 201 can be a large model, or can be other types of neural networks, etc., which are not limited in the embodiment of the present application.

[0105] The first machine learning model 201 that has been trained and obtained by the training device 210 can be deployed in the computing module 231 of the execution device 230. The execution device 230 can call the data, code, etc. in the data storage system 240, or store the data, instructions, etc. in the data storage system 240. The data storage system 240 can be placed in the execution device 230, or the data storage system 240 can be an external memory relative to the execution device 230. It should be noted that in the application stage of the first machine learning model 201, the first machine learning model 201 is used to generate a mathematical model corresponding to the operations optimization problem. The concepts of "operations optimization problem" and "mathematical model" can be referred to in the above examples and will not be described in detail here.

[0106] In some embodiments of the present application, please refer to Figure 2, the execution device 230 and the client device 250 are independent devices, and the execution device 230 is configured with an input / output (I / O) interface to interact with the client device 250 for data. After obtaining the first description information corresponding to at least one problem, the client device 250 can obtain the first information based on the first description information, and then send the first information to the execution device 230 through the I / O interface. The first description information includes the description information of the first operations optimization problem; after receiving the first information, the execution device 230 can generate a first mathematical model corresponding to the first information through the first machine learning model 201 in the computing module 231, and then send the aforementioned first mathematical model to the client device 250 through the I / O interface.

[0107] For example, the client device 250 can be a terminal device or an edge device, and the execution device 230 can be a cloud device, such as a server or a server cluster. The product form of the client device 250 and the execution device 230 is not limited in the embodiments of the present application.

[0108] It is worth noting that Figure 2 is only an architectural diagram of two data processing systems provided by an embodiment of the present invention, and the positional relationship between the devices, components, modules, etc. shown in the figure does not constitute any limitation. For example, in other embodiments of the present application, the execution device 230 and the client device 250 may be integrated into the same device, and the user can interact directly with the execution device 230. Exemplarily, when the client device 250 is a mobile phone or a tablet, the execution device 230 may be a module in the main processor (Host CPU) of the mobile phone or tablet that uses the first machine learning model to perform data processing, and the execution device 230 may also be a neural network processor (NPU) in the mobile phone or tablet. The NPU is mounted on the main processor as a coprocessor, and the main processor assigns tasks.

[0109] Specifically, please refer to Figure 3, which is a schematic diagram of a method for obtaining a mathematical model corresponding to an operations optimization problem provided in an embodiment of the present application. In step 301, a first device outputs at least one first question, and obtains at least one answer based on the at least one first question. The at least one answer is used to obtain first descriptive information, wherein the first descriptive information is descriptive information used to describe the first operations optimization problem, and the first question is the question used to obtain the descriptive information of the first operations optimization problem.

[0110] In an embodiment of the present application, exemplarily, the first device may be the client device 250 in FIG2 . In order to obtain detailed first description information of the first operations optimization problem, the first device may first determine at least one question used to obtain the description information of the first operations optimization problem (hereinafter referred to as the “target question” for the convenience of distinction), and then determine at least one question output to the user (hereinafter referred to as the “first question” for the convenience of distinction), use the content of the user's reply to each of the aforementioned at least one first question as at least one answer, and obtain the aforementioned at least one answer.

[0111] For example, when the operations optimization problem is a production scheduling problem, the at least one target problem may include: Is the demand delivery mode a back order or a lost sale? Should a single-layer or multi-layer structure be generated? Should one item or multiple items be produced? Or other questions, etc. For another example, when the operations optimization problem is a factory scheduling problem, the at least one target problem may include: What is the number of machines? What is the relationship between tasks and machines? Or other questions, etc. It should be noted that the examples given here for the at least one target problem in conjunction with specific operations optimization problems are only for the convenience of understanding this solution. The at least one target problem may also be oriented towards any operations optimization problem. The specific setting of the at least one target problem can be flexibly determined in combination with the actual application scenario, and is not limited in the embodiments of this application.

[0112] 302. The first device determines first information according to the first description information.

[0113] In an embodiment of the present application, the first information may include the first descriptive information and other information, or the first information and the first descriptive information may include the same information; exemplarily, the other information may include prompt information, for example, the prompt information may include "Please generate a mathematical model corresponding to the operations optimization problem based on the subsequent description information", and for example, the prompt information may include "Please generate a mathematical model corresponding to the operations optimization problem based on the first descriptive information, the first descriptive information includes: " and so on. The prompt information may also be expressed as other information, etc.

[0114] 303. The first device inputs the first information into a machine learning model to obtain a first mathematical model. The first mathematical model includes an objective function and constraints. The first mathematical model is used to solve a first operations optimization problem.

[0115] In an embodiment of the present application, the above-mentioned machine learning model (hereinafter referred to as the "first machine learning model" for the convenience of distinction) can be a large model. For example, the large model can be specifically expressed as a Pangu large model or other types of large models, etc.; or the first machine learning model can also be a neural network in other forms. For example, the first machine learning model can be a neural network based on an attention mechanism, etc., which is not limited in this application.

[0116] The first mathematical model is a mathematical model generated using the first machine learning model and corresponding to the first operations optimization problem. The purpose of generating the first mathematical model using the aforementioned first machine learning model is to solve the first operations optimization problem. That is, the solution of the first mathematical model is capable of solving the aforementioned first operations optimization problem. Exemplarily, the first mathematical model includes an objective function and constraints corresponding to the first operations optimization problem. Optionally, the first mathematical model may also include variables corresponding to the first operations optimization problem. For examples of the first mathematical model, please refer to the above description and will not be given here.

[0117] In one case, "the first device inputs the first information into the first machine learning model" can be understood as the first device sending the first information to the device deploying the first machine learning model; "the first device obtains the first mathematical model corresponding to the first operations optimization problem" can be understood as the first device receiving the first mathematical model sent by the device deploying the first machine learning model.

[0118] In another case, step 303 can be understood as the first device inputting the first information into a locally deployed first machine learning model to obtain a first mathematical model output by the first machine learning model.

[0119] It should be noted that the "operations optimization problem" in this application can also be called "decision optimization problem" or other names, and the "mathematical model" in this application can also be called "mathematical programming model" or other names.

[0120] In an embodiment of the present application, in order to obtain descriptive information of an operations optimization problem, at least one problem is abstracted, and an answer to the at least one problem is obtained. After obtaining the first information based on the first descriptive information, the first information is input into the machine learning model to obtain a first mathematical model corresponding to the first operations optimization problem. In this solution, for different fields, there is no need to generate multiple objective functions, text descriptions corresponding to each objective function, multiple constraints, and text descriptions corresponding to each constraint in advance in each field, which greatly reduces the manpower cost consumed in the process of obtaining the mathematical model corresponding to the operations optimization problem, and can be adapted to obtain mathematical models corresponding to operations optimization problems in various fields, and has high generalization. In addition, in this solution, the descriptive information of the operations optimization problem is obtained by obtaining the answer to the aforementioned at least one question, which to a certain extent alleviates the problem that the user cannot accurately determine the first operations optimization problem, and is conducive to improving the adaptability between the mathematical model output by the machine learning model and the operations optimization problem, that is, it is conducive to obtaining a more accurate mathematical model.

[0121] In conjunction with the above description, the specific implementation process of the method provided in the embodiment of the present application is described in detail below. Specifically, please refer to Figure 4, which is another flow chart of the method for obtaining a mathematical model provided in the embodiment of the present application. The method for obtaining a mathematical model provided in the embodiment of the present application may include:

[0122] 401. Determine the field to which the mathematical model to be established belongs.

[0123] In the embodiment of the present application, step 401 is an optional step. For example, the first device in the embodiment corresponding to FIG4 can be the client device 250 shown in FIG2. Before executing the step of generating the mathematical model corresponding to the first operational optimization problem, the first device can first determine the field to which the mathematical model to be established belongs (hereinafter referred to as the "first field" for the convenience of description), and then determine which questions (hereinafter referred to as the "target question" for the convenience of description) need to be answered based on the first field to which the mathematical model to be established belongs, so as to obtain detailed first description information of the first operational optimization problem; it should be understood that the relationship between the two concepts of "first description information" and "first operational optimization problem" can be referred to in the above description, and will not be elaborated here. Since the establishment of the mathematical model is used to solve the first operational optimization problem, the "field to which the mathematical model to be established belongs" can also be understood as the field to which the first operational optimization problem belongs.

[0124] Exemplarily, the first field to which the mathematical model to be established belongs can be any of the following: the field of location problem, the field of scheduling problem, the field of order fulfillment, the field of supply chain, the field of packing problem, the field of transportation, the field of resource assignment, the field of revenue management, the field of production planning problem, or other fields. It should be noted that the specific fields to which the mathematical model to be established can belong can be flexibly determined in combination with the actual application scenario. The examples here are only for the convenience of understanding the concept of "the field to which the mathematical model to be established belongs" and are not used to limit this solution. The multiple fields to which the mathematical model to be established can belong are listed, which greatly expands the application scenarios of this solution and is conducive to improving the implementation flexibility of this solution.

[0125] For example, the "site selection problem domain" mainly involves selecting one or more optimal locations from a given set of candidate locations to meet specific goals and constraints. The site selection problem domain can include multiple sub-problems, such as facility site selection problems, warehouse site selection problems, network site selection problems, or other site selection problems. For example, the facility site selection problem refers to the need to determine which facility locations should be opened to meet the demand, minimize overall costs, or maximize overall benefits, given a set of potential facility locations and a set of demand points. This type of problem is common in application areas such as retail and logistics.

[0126] The "scheduling problem domain" refers to the problem of rationally arranging the sequence and timing of tasks or work to maximize efficiency or meet specific constraints given limited resources. The scheduling problem domain can include multiple sub-problems, such as job scheduling, vehicle scheduling, project scheduling, or other scheduling problems. For example, a job scheduling problem involves determining the start and finish time of each job given a set of jobs and a set of available resources in order to minimize the overall completion time or maximize the overall profit. This type of problem is common in application areas such as manufacturing and project management.

[0127] The "order fulfillment domain" refers to the rational arrangement of order processing and delivery, given a set of orders and available resources, to meet customer needs and maximize efficiency. Order fulfillment involves multiple aspects, including order receipt, order processing, inventory management, and logistics distribution. Its goal is to minimize order processing time, reduce inventory costs, and improve on-time delivery, all while meeting customer needs. Common challenges in order fulfillment include order priority, resource constraints, delivery time windows, and inventory management. The key to addressing these challenges lies in rationally arranging the order processing sequence, allocating resources, and optimizing logistics routes. Methods and techniques from operations research can be applied to solving order fulfillment problems. For example, linear programming can be used to optimize resource allocation and order processing sequence, dynamic programming can be used to optimize logistics routes, and simulated annealing and genetic algorithms can be used to solve complex scheduling problems.

[0128] The "supply chain" refers to the optimization of logistics, inventory, production, and other aspects of the supply chain network, encompassing multiple links and participants, through rational planning and decision-making to maximize efficiency and profitability across the entire supply chain. Supply chain issues involve multiple links, including suppliers, manufacturers, distributors, retailers, as well as logistics, inventory, and order management. The goal is to achieve optimal resource allocation, inventory control, production planning, and logistics distribution by optimizing decisions across each link in the supply chain to meet customer needs and reduce costs.

[0129] The "packing problem domain" refers to placing a group of items into as few containers as possible to maximize the use of the container space. Packing problems are often used to optimize the fields of logistics and transportation to reduce transportation costs and improve efficiency. The packing problem domain can include multiple sub-problems, such as one-dimensional packing problems and two-dimensional packing problems. For example, the one-dimensional packing problem refers to placing a group of items into containers on a straight line so that there is no overlap between items. Each item has its own length, and the container has a certain length limit. The goal is to find a placement plan that minimizes the number of containers used. The two-dimensional packing problem refers to placing a group of items into a container on a two-dimensional plane so that there is no overlap between items. Each item has its own length and width, and the container has a certain length and width limit. The goal is to find a placement plan that minimizes the number of containers used.

[0130] The "transportation field" refers to the problem of optimizing resource allocation and path selection in the transportation network. Operational optimization problems in the transportation field involve aspects such as traffic flow management, route planning, and traffic signal optimization, aiming to improve the efficiency and safety of the transportation system. The transportation field can include multiple sub-problems, such as traffic flow allocation problems, path selection problems, and traffic signal optimization problems. For example, the traffic flow allocation problem refers to how to reasonably allocate traffic flow to different paths to reduce congestion and improve traffic efficiency. This problem can be solved by establishing a traffic flow model and using optimization algorithms. The path selection problem refers to how to choose the best path to reach the destination in a given transportation network. This problem can be optimized by considering factors such as traffic flow, road conditions, and travel time.

[0131] The "resource allocation field" refers to how to reasonably allocate limited resources to maximize benefits or meet specific constraints. Resources can be manpower, materials, funds, equipment, etc., and operational optimization problems in the resource allocation field can involve different fields, such as production, logistics, project management, etc. The resource allocation field can include multiple sub-problems, such as production resource allocation problems, logistics resource allocation problems, and project resource allocation problems. For example, the production resource allocation problem refers to how to reasonably allocate resources in the production process to maximize output or profit. This problem involves aspects such as production line optimization, job scheduling, and equipment configuration. The logistics resource allocation problem refers to how to reasonably allocate resources in the logistics process to minimize costs or improve efficiency. This problem involves aspects such as the transportation, warehousing, and distribution of goods. The project resource allocation problem refers to how to reasonably allocate resources during the project execution process to maximize project completion or meet specific constraints. This problem involves aspects such as the allocation of project tasks, resource scheduling, and progress control.

[0132] The field of revenue management refers to maximizing revenue for a business or organization through strategies such as pricing, inventory management, and capacity control. Revenue management issues are often applied to service industries such as aviation, hospitality, and tourism. The revenue management field can encompass multiple sub-problems, such as pricing, inventory management, and capacity control. Pricing, for example, involves determining the price of a product or service to maximize revenue, a problem that involves factors such as market demand, the competitive landscape, and consumer behavior.

[0133] The "generative planning problem domain" refers to how to reasonably arrange resources and tasks in the production process to maximize production efficiency and meet customer needs. Production planning problems involve production scheduling, task allocation, resource utilization, etc., aiming to improve the efficiency and flexibility of the production system. The production planning domain can include multiple sub-problems, such as production scheduling problems, task allocation problems, and resource allocation problems. For example, the production scheduling problem refers to how to reasonably arrange the execution order and time of production tasks to minimize production time and cost. This problem involves aspects such as task priority, equipment utilization, and production line balance. The task allocation problem refers to how to reasonably allocate production tasks to different workstations or employees to maximize production efficiency and balance workload. This problem involves aspects such as the characteristics of the task, the capabilities of the workstation, and the skills of the employees. The resource allocation problem refers to how to reasonably allocate resources in the production process to maximize resource utilization and meet production needs. This problem involves aspects such as equipment scheduling, raw material procurement, and human resource allocation.

[0134] It should be noted that, in practical applications, the mathematical model to be established can also be divided into other fields. The above explanations of the multiple problem fields to which the mathematical model to be established belongs are only for the convenience of understanding this solution and are not used to limit this solution.

[0135] The first device can implement step 401 in multiple ways. In one implementation, the first device can provide the user with multiple fields to which the operations optimization problem can belong, that is, provide the user with multiple fields to which the mathematical model to be established can belong. After obtaining the user's selection operation input for the first field among the aforementioned multiple fields, the first field to which the mathematical model to be established belongs can be determined.

[0136] For example, the first device may display multiple fields to which the operations optimization problem may belong through a display screen. Upon receiving a click operation input by the user for one of the multiple fields, it may be determined that a selection operation input by the user for the first field is obtained.

[0137] In order to understand this solution more intuitively, please refer to Figure 5, which is a schematic diagram of the interface for obtaining the "field to which the mathematical model to be established belongs" provided in an embodiment of the present application. As shown in Figure 5, after the first device determines that a mathematical model needs to be automatically constructed for the operations optimization problem, it can display to the user through the display screen a plurality of fields to which the operations optimization problem can belong, namely, the site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field and planning problem field shown in Figure 5. When the first device receives a click operation input by the user for one of the aforementioned multiple fields, it can be determined that the selection operation input by the user for the first field (that is, the "planning problem field") is obtained. It should be understood that the example in Figure 5 is only for the convenience of understanding this solution and is not used to limit this solution.

[0138] For example, in another case, the first device can play multiple fields to which the operations optimization problem can belong to to the user in the form of voice, and can obtain the selection operation input by the user in the form of voice for the first field among the multiple fields; illustratively, the first device can play in the form of voice "The fields to which operations optimization problems belong include: site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field and production planning problem field. For which field of operations optimization problems do you want to generate a mathematical model?" It should be understood that the examples given here are not used to limit this solution.

[0139] In another case, the first device can display the multiple fields to which the operations optimization problem can belong through the display screen, and also play the multiple fields to which the operations optimization problem can belong to to the user in the form of voice. After obtaining the user's click operation input for a certain field among the aforementioned multiple fields, it can determine the first field to which the mathematical model to be established belongs, and so on. It should be noted that the first device can also use other methods to achieve "providing the user with multiple fields to which the operations optimization problem can belong, and then obtaining the user's selection operation input for the first field among the aforementioned multiple fields". The example here is only to prove the feasibility of this solution and is not used to limit this solution.

[0140] In another implementation, the first device can obtain third descriptive information input by the user, and based on the aforementioned third descriptive information, use the second machine learning model to determine the first field to which the mathematical model to be established belongs; the "third descriptive information" can be understood as the background descriptive information of the first operations optimization problem.

[0141] For example, the first device may display a text box on a display screen and receive the third descriptive information in the form of text input by the user through the text box. Alternatively, the first device may receive the third descriptive information in the form of voice input by the user. Alternatively, the first device may use other methods to obtain the third descriptive information. This application does not limit the form in which the first device obtains the third descriptive information. For example, the content of the third descriptive information may be "Please generate a mathematical model for operational optimization problems in the supply chain field," "Operational optimization problems in the field of packing problems," or other content. It should be understood that the examples here are only for the convenience of understanding this solution and are not used to limit this solution.

[0142] For example, in one case, the second machine learning model can be specifically expressed as a large model. It should be noted that when the second machine learning model is expressed as a large model, the second machine learning model and the above-mentioned first machine learning model can be the same machine learning model or different machine learning models, which is not limited in this application.

[0143] After obtaining the third descriptive information, the first device can input multiple fields to which the operations optimization problem can be attributed and the third descriptive information into the large model to obtain second prediction information corresponding to the third descriptive information. The second prediction information indicates the first field corresponding to the third descriptive information, that is, the first field corresponding to the first operations optimization problem.

[0144] Exemplarily, the first device may determine the first prompt word based on "multiple fields to which operations optimization problems can belong"; exemplarily, the first prompt word may include content such as "for the given background description, select a field that best matches the background description of the operations optimization problem from the site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field, and production planning problem field." It should be understood that the example here is only for the convenience of understanding the concept of "first prompt word" and is not used to limit this solution.

[0145] "The first device inputs multiple fields to which the operations optimization problem can be attributed and the third descriptive information into the second machine learning model" can be understood as the first device can send the first prompt word and the third descriptive information to the device where the second machine learning model is deployed. After receiving the first prompt word and the third descriptive information, the device where the second machine learning model is deployed calls the second machine learning model to process the first prompt word and the third descriptive information to obtain the above-mentioned second prediction information output by the second machine learning model. The device where the second machine learning model is deployed sends the second prediction information corresponding to the third descriptive information to the first device. Correspondingly, the first device can obtain the above-mentioned second prediction information corresponding to the third descriptive information.

[0146] To understand this solution more intuitively, please refer to Figure 6, which is a schematic diagram of obtaining the second prediction information using a large model provided in an embodiment of the present application. As shown in Figure 6, the first prompt word can be determined based on the multiple fields to which the mathematical model to be established can belong (that is, the site selection problem field, scheduling problem field, ... shown in Figure 6). The first prompt word can be the "prompt word" shown in Figure 6: "For the given background description, select a field that best matches the background description of the operations optimization problem from {site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field, and production planning problem field}". The aforementioned operations optimization problem background description is also the third description information.

[0147] The first prompt word and the third descriptive information are input into the second machine learning model in the form of a large model to obtain second prediction information output by the second machine learning model in the form of a large model. The second prediction information indicates the first field corresponding to the third descriptive information, that is, indicates the first field corresponding to the first operations optimization problem. It should be understood that the example in Figure 6 is only for the convenience of understanding this solution and is not used to limit this solution.

[0148] In another scenario, a second machine learning model that has been trained may be deployed in the first device. The second machine learning model is a machine learning model for performing text processing tasks. Exemplarily, the second machine learning model is used to determine a domain that best matches the input descriptive information among multiple domains to which the operations optimization problem can be assigned. The first device may input the first descriptive information into the second machine learning model, and obtain second prediction information corresponding to the first descriptive information output by the second machine learning model. The second prediction information indicates the first domain corresponding to the first descriptive information.

[0149] It should be noted that the first device can also determine the field to which the mathematical model to be established belongs through other methods. The examples in the embodiments of this application are only for the convenience of understanding this solution and are not used to limit this solution.

[0150] In an embodiment of the present application, all operations research and optimization problems are divided into multiple fields, that is, multiple fields to which the mathematical model to be established can belong are obtained. After determining the field to which the mathematical model to be established belongs, based on the field to which the aforementioned first operations research and optimization problem belongs, at least one problem used to obtain detailed description information of the first operations research and optimization problem is determined, that is, when obtaining description information of operations research and optimization problems in different fields, different problems are used. It can be seen from the above description that in this solution, more refined management of the problems used to obtain description information of operations research and optimization problems is carried out, which is conducive to obtaining more accurate description information, and further conducive to obtaining a more accurate mathematical model.

[0151] 402. Output at least one first question, and obtain at least one answer based on the at least one first question, where the at least one answer is used to obtain first descriptive information, wherein the first descriptive information is descriptive information used to describe the first operations research optimization problem, and the first question is a question used to obtain the descriptive information of the first operations research optimization problem.

[0152] In the embodiment of the present application, step 402 is an optional step. It should be understood that the first descriptive information is descriptive information used to describe the first operational optimization problem, and the first mathematical model is used to solve the first operational optimization problem. In the present application, the first descriptive information is used to replace the specific first operational optimization problem in the prior art, which to a certain extent alleviates the problem that the user cannot accurately determine the first operational optimization problem. Before the first device executes the step of generating the mathematical model corresponding to the first operational optimization problem, in order to obtain the detailed first descriptive information of the first operational optimization problem, it can obtain the answer to each target problem in at least one problem corresponding to the first operational optimization problem (for the convenience of description, it will be referred to as "target problem" later). The first descriptive information can include the answer to the aforementioned at least one target problem, and then the at least one target problem corresponding to the first operational optimization problem can be determined first.

[0153] For example, the at least one target question may include at least one first question output to the user. Optionally, the at least one target question may also include a second question. For example, the first question refers to a question output to the user by the first device, and the second question refers to a question other than the first question in the at least one target question. For example, the answer to the second question may be obtained based on second descriptive information actively input by the user to the first device.

[0154] Regarding the specific implementation method for the first device to determine at least one problem corresponding to the first operations optimization problem, since step 401 is an optional step, in one implementation method, if step 401 is executed, the first device can determine at least one target problem corresponding to the first field to which the first operations optimization problem belongs based on the first field obtained in step 401, that is, obtain at least one target problem corresponding to the first operations optimization problem.

[0155] In one embodiment, the first device may store at least one problem set, including a problem set corresponding to each of at least one field to which an operations optimization problem can be assigned, and each problem set in the at least one problem set includes at least one problem. Optionally, the at least one problem set corresponds one-to-one with the at least one field to which the operations optimization problem can be assigned. The first device may then obtain, from the at least one problem set, a first problem set corresponding to the first field to which the first operations optimization problem belongs, i.e., the at least one problem set includes a first problem set corresponding to the first operations optimization problem, and the first problem set includes at least one target problem.

[0156] In another case, after determining the first domain to which the first operational optimization problem belongs, the first device may generate a first problem set corresponding to the first domain, that is, obtain a first problem set corresponding to the first operational optimization problem.

[0157] In another implementation, if step 401 is not performed, the first device may store a second problem set including at least one problem, or the first device may generate a second problem set including at least one problem after determining that a mathematical model corresponding to the operations optimization problem needs to be established. The second problem set may correspond to operations optimization problems belonging to any field, that is, no matter which field the first device needs to establish a mathematical model for operations optimization problems, the problems in the second problem set may be used. The at least one problem included in the second problem set may also be understood as at least one target problem corresponding to the first operations optimization problem.

[0158] Optionally, the first question set (or the second question set) may further include a correct answer to each target question, that is, the first question set (or the second question set) may further include an answer template for each target question.

[0159] For example, the data format used by each question set in the at least one question set mentioned above can be any of the following: directed graph, list, table, undirected graph or other data forms, etc. The specific data form used can be flexibly determined in combination with the actual application scenario, and is not limited in the embodiments of this application.

[0160] Exemplarily, when the first question set (or the second question set) is expressed in the data form of a directed graph, each target question in the first question set (or the second question set) is used as a point in the directed graph, and the correct answer to each target question in the first question set (or the second question set) is used as an edge in the directed graph. The connection relationship between different points in the directed graph is determined based on the association relationship between multiple target questions in the first question set (or the second question set).

[0161] In which, when the first question set (or the second question set) adopts the data form of a directed graph, the arrows in the directed graph can indicate the order of appearance between at least two different target questions among the multiple target questions when multiple target questions in the first question set (or the second question set) are provided to the user.

[0162] For a more intuitive understanding of this solution, please refer to Figure 7, which is a schematic diagram of multiple target questions in the form of a directed graph provided in an embodiment of the present application. In Figure 7, the first set of questions (or the second set of questions) including seven target questions P1 to P7 is taken as an example. In Figure 7, each target question is regarded as a point in the directed graph, and the correct answer to each target question is regarded as an edge in the directed graph.

[0163] When presenting the seven target questions P1 to P7 to the user in the order indicated by the arrows in Figure 7, there are constraints on the presentation order of the four target questions P1, P3, P4, and P6, and on the presentation order of the three target questions P2, P5, and P7. There are no constraints on the presentation order of the four target questions P1, P3, P4, and P6 and the three target questions P2, P5, and P7. As shown in Figure 7, P1 is presented earlier than P3 and P4, and P4 is presented earlier than P6. When the correct answer 1 is obtained for the target question P1, P3 is displayed to the user, and P4 and P6 are no longer required. When the correct answer 2 is obtained for the target question P1, P4 is displayed to the user, and P3 is no longer required. The presentation order of P2 needs to be earlier than that of P5, and P5 needs to be presented earlier than that of P7. It should be understood that the example in Figure 7 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0164] In an embodiment of the present application, a first question set is pre-stored, and the pre-stored first question set can be used to determine which first questions are output to the user, which is conducive to improving the speed of the process of "determining which first questions are output" and is conducive to more efficiently obtaining a mathematical model corresponding to the operations optimization problem.

[0165] The questions included in the first question set are stored in the data form of a directed graph, which can indicate the order of appearance between at least two different questions in the first question set. Since different questions in at least one question corresponding to the first operations optimization problem may have a correlation relationship, the use of the data form of a directed graph can better reflect the logical relationship between different questions, making the process of asking questions to the user more logical, and is also conducive to avoiding outputting useless questions to the user, so as to improve the user stickiness of this solution.

[0166] Exemplarily, step 402 may include: the first device may output a first question to the user, then determine the user's response to each of the at least one first question as at least one answer, and determine first descriptive information based on the at least one answer. Optionally, if the first device also obtains second descriptive information input by the user, and the second descriptive information carries descriptive information of the first operations optimization problem, the first device may further determine which of the at least one target question's second questions are answered in the second descriptive information, and then determine which first questions to output to the user based on the at least one target question and the second questions for which answers have been obtained.

[0167] Furthermore, in one implementation, after the first device determines the first problem set (or second problem set) corresponding to the first operations optimization problem, that is, after determining at least one target problem included in the first problem set (or second problem set), it can output at least one first question to the user for the first time to obtain an answer input by the user for each first question output for the first time.

[0168] For example, the specific implementation method of the above-mentioned "outputting the first question to the user" can be through displaying on a display screen, playing in voice form or other output methods, etc. The specific output method can be flexibly determined in combination with the actual application scenario, and is not limited in this application. In the embodiment of this application, only "displaying through a display screen" is used as an example to illustrate the detailed time process of "displaying the first question and obtaining the answer to the first question". For the specific implementation methods using other output methods, please refer to them for understanding, and they will not be described one by one in the embodiment of this application.

[0169] In one case, if the data format of the first question set (or the second question set) is a directed graph, the first device may determine each first question to be presented to the user for the first time via the display screen based on the order of presentation indicated by the directed graph. Alternatively, if the data format of the first question set (or the second question set) is other data format capable of indicating the order of presentation between at least two different target questions, the first device may also determine at least one first question to be presented to the user for the first time based on the order of presentation indicated by the other data format.

[0170] In order to understand this solution more intuitively, an example is given here in conjunction with Figure 7. The at least one first question displayed to the user for the first time through the display screen may include two target questions P1 and P2, or the at least one first question displayed to the user through the display screen may be a single target question P1, or the at least one first question displayed to the user through the display screen may be a single target question P2. It should be understood that the example given here in conjunction with Figure 7 is only for the convenience of understanding this solution and is not used to limit this solution.

[0171] In another case, if the data format of the first question set (or the second question set) is a list, a table, an undirected graph, or other data format that does not indicate the order of appearance between different target questions, the at least one first question displayed to the user for the first time may include any one or more target questions in the first question set (or the second question set).

[0172] Optionally, in one scenario, if the first device obtains second description information input by the user before determining the first question to be presented to the user for the first time, and the second description information carries description information of the first operations optimization problem, the first device may determine which of the second questions among the at least one target question are answered in the second description information, and further determine the first question to be presented to the user for the first time based on the at least one target question and the aforementioned second question. And / or, in another scenario, if the first device obtains second description information input by the user while presenting the first question to the user multiple times, the first device may also determine which of the second questions among the at least one target question are answered in the second description information, and further determine the first question to be presented to the user based on the at least one target question and the aforementioned second question.

[0173] Specifically, the first device can implement "outputting the first question and obtaining the answer to the first question" in a variety of implementation methods. For example, in one implementation method, the first device not only displays each first question to the user through a display screen, but also displays the answer options corresponding to each first question. When the first device obtains the user's selection operation input through the display screen for a certain answer option (hereinafter referred to as the "first answer option" for convenience of description), the correct answer corresponding to the aforementioned first answer option can be obtained, that is, the correct answer to a certain first question (hereinafter referred to as the "third question" for convenience of distinction) among all the first questions displayed for the first time is obtained.

[0174] For a more intuitive understanding of this solution, please refer to Figure 8, which is a schematic diagram of "displaying the first question" and "obtaining the answer to the first question" provided in an embodiment of the present application. As shown in Figure 8, after the first device determines that the field to which the first operations optimization problem belongs is the "planning problem field", it displays three first questions to the user through the display screen. The three first questions respectively relate to the granularity of the plan, the number of codes included in the plan, and the total length of the plan period. When the user selects a first question (for example, by hovering the cursor over a first question), the first device can be triggered to display the answer options for the aforementioned first question. As shown in Figure 8, when the first device determines that the cursor is hovering over the first question "This is a plan with a granularity of {day}", it can trigger the display of three answer options for the first question "This is a plan with a granularity of {day}" (i.e., "day", "half day" and "week" in Figure 8). The user can click on one of the three answer options to enter the selection operation for the answer option.

[0175] Optionally, referring to FIG8 , after the first device determines the first field, it may also display at least one historical case in the planning problem field to the user, where the at least one historical case is a mathematical model established for an operations optimization problem in the planning problem field.

[0176] It should be noted that the example given here in conjunction with Figure 8 is only for the convenience of understanding this solution. "Triggering the display of answer options" and "inputting a selection operation for a certain answer option through the display screen" can also be implemented in other ways. For example, the first device can directly display all answer options for each first question while displaying each first question; for example, the user can drag a certain answer option to the area where the first question is located to implement the input selection operation for a certain answer option, etc. The specific implementation method can be flexibly determined in combination with the actual application scenario, and is not limited in this application.

[0177] In another implementation, after the first device displays at least one first question to the user for the first time via a display screen, it may obtain a first answer entered by the user via a text box, and the first device may obtain a first matching result between the first answer and all correct answers to all first questions displayed for the first time. Based on the first matching result, the first device may determine the correct answers to one or more third questions among all first questions displayed for the first time, or the first device may determine, based on the first matching result, that the correct answer to any first question among all first questions displayed for the first time was not obtained.

[0178] Optionally, the first device may obtain a first matching result between the first answer and all correct answers to all first questions displayed for the first time through a third machine learning model. For example, in one case, if the third machine learning model is specifically expressed as a large model, the third machine learning model and the above-mentioned second machine learning model and first machine learning model may be the same machine learning model or different machine learning models. The first device may send the first answer, all first questions displayed for the first time, and all correct answers to all first questions displayed for the first time to a device deployed with a third machine learning model in the form of a large model, and the aforementioned device calls the third machine learning model in the form of a large model to process the first answer, all first questions displayed for the first time, and all answers to all first questions displayed for the first time, and obtain a first matching result generated by the third machine learning model in the form of a large model. The device deployed with the third machine learning model in the form of a large model sends the first matching result to the first device.

[0179] For example, the above-mentioned device can input all the first questions displayed for the first time and all the answers to all the first questions displayed for the first time into the third machine learning model in the form of a large model, and input "Please ask which questions among all the first questions displayed for the first time are included in the first answer and which correct answers are included" into the third machine learning model in the form of a large model, triggering the third machine learning model in the form of a large model to process the first answer, all the first questions displayed for the first time, and all the answers to all the first questions displayed for the first time to obtain a first matching result. It should be understood that the examples here are only to prove the feasibility of this solution and are not used to limit this solution.

[0180] If the first matching result is used to inform the first device that the first answer meets the first matching condition with the correct answer to the one or more third questions (hereinafter referred to as the "first correct answer" for ease of description), the first device may determine that the first answer is the correct answer to the one or more third questions. If the first matching result is used to inform the first device that the first answer does not meet the first matching condition with the correct answer to any of the first questions among all the first questions displayed for the first time, the first device may determine that the correct answer to any of the first questions among all the first questions displayed for the first time has not been obtained.

[0181] In another case, a third machine learning model that has performed training operations may be deployed in the first device, and the first device inputs the first answer and all correct answers to all first questions displayed for the first time into the third machine learning model to obtain a first matching result output by the third machine learning model, where the first matching result indicates the similarity between the first answer and each correct answer among all correct answers to all first questions displayed for the first time.

[0182] The first device may determine, based on the first matching result, whether the similarity between the aforementioned first answer and a correct answer to at least one of all first questions displayed for the first time satisfies a first similarity. If the first device determines that the similarity between the first answer and the first correct answer to the one or more third questions satisfies the first similarity, the first device may determine that the first answer is the correct answer to the one or more third questions. Alternatively, if the first device determines that the similarity between the aforementioned first answer and all correct answers to all first questions displayed for the first time does not satisfy the first similarity, the first device may determine that the correct answer to any of the first questions displayed for the first time has not been obtained.

[0183] Alternatively, the first device may also adopt other algorithms to calculate the similarity between the first answer and all answers to all first questions displayed for the first time, for example, by calculating the cosine similarity, Euclidean distance or L1 distance between the first answer and all answers to all first questions displayed for the first time, to determine the similarity between the first answer and all answers to all first questions displayed for the first time, and then determine which answers to which questions are included in the first answer. The above-mentioned examples of various implementation methods are only to prove the feasibility of this solution and are not used to limit this solution.

[0184] In another implementation, the first device not only displays at least one first question to the user for the first time through the display screen, but also displays the answer options corresponding to each first question, and displays a text box for obtaining the answer to the first question to the user through the display screen. The user can then input a selection operation for a certain answer option, or the user can also input the answer through the text box, and then the first device obtains the answer to at least one third question among all the first questions displayed for the first time based on the user's operation; it should be noted that the specific implementation methods of "determining the answer to the third question based on the selection operation input by the user for the answer option" or "determining the answer to the third question based on the text description input by the user" can refer to the description of the above two implementation methods, which will not be repeated here.

[0185] After the first device outputs at least one first question to the user for the first time and then determines that the answer to at least one third question has been obtained (or determines that the answer to any question has not been obtained), it can determine which of all the target questions included in the first question set (or second question set) have not yet obtained the correct answer, and then determine which first questions need to continue to be output to the user; the first device repeats the above steps at least once until the correct answers to all the target questions in the first question set (or second question set) are obtained, and stops outputting the first question to the user; that is, the first device obtains the first description information, and the first description information includes the correct answer to at least one target question. Among them, "first question" and "third question" are both questions among multiple target questions, the difference being that "first question" represents the question output to the user, and "third question" represents the question for which the correct answer has been obtained.

[0186] It should be noted that the specific implementation methods of "the first device determines the first question to be output to the user each time", "how to output the first question to the user" and "determine which answers to the third questions have been obtained" can be found in the above description of the specific implementation methods of "outputting the first question to the user and obtaining the answer for the first time", and will not be repeated here.

[0187] For a more intuitive understanding of this solution, please refer to Figure 9, which is a schematic diagram of obtaining first descriptive information corresponding to at least one problem provided in an embodiment of the present application. As shown in Figure 9, in stage one: after the first device determines that a first mathematical model corresponding to the first operations optimization problem needs to be generated, it can first determine the first field to which the mathematical model to be established belongs, and then determine at least one target problem corresponding to the operations optimization problem in the first field.

[0188] In stage two: the first device can obtain the answer to the aforementioned at least one target question, that is, obtain the first description information. In Figure 9, taking the use of a directed graph to store the aforementioned at least one target question as an example, the first device can determine the at least one first question output to the user for the first time based on the indication of the directed graph to obtain the correct answer to at least one third question. The first device determines which third questions in the at least one target question have been answered (that is, which correct answers to the third questions have been obtained), and then determines which answers to the at least one target question have not yet been obtained, and continues to determine the at least one first question to be output to the user again based on the indication of the directed graph. The first device can repeat the aforementioned operation at least once to obtain the first description information. It should be understood that the example in Figure 9 is only for the convenience of understanding this solution and is not used to limit this solution.

[0189] In an embodiment of the present application, at least one first question is output to the user to obtain an answer to each question, and then the first descriptive information is determined based on the answer to each first question. That is, a guided approach is adopted to obtain the answer to at least one question corresponding to the first operations optimization problem. This is conducive to more efficiently obtaining the descriptive information of the first operations optimization problem, and is also conducive to obtaining more accurate first descriptive information, thereby facilitating obtaining a more accurate mathematical model.

[0190] Optionally, before the first device outputs the first question to the user for the first time, or during the process of the first device outputting the first question to the user multiple times, it can also obtain second description information actively input by the user; or, after the first device outputs the first question to the user, it obtains second description information actively input by the user, and the second description information carries description information of the first operations optimization problem.

[0191] The first device can also merge the first description information and the second description information to obtain the union of the first description information and the second description information (for the convenience of description, hereinafter referred to as "updated first description information"), and then determine the first information based on the aforementioned updated first description information in subsequent steps. The first information includes the updated first description information, that is, the first information includes the aforementioned first description information and the second description information.

[0192] 403. Determine at least one first case from the at least one case based on the similarity between the first descriptive information and each descriptive information included in the at least one case, wherein the first device stores at least one case, any one of the at least one case includes descriptive information and a mathematical model corresponding to the operations optimization problem, and the similarity between the first descriptive information and the descriptive information included in the first case satisfies a preset condition.

[0193] In the embodiment of the present application, step 403 is an optional step. It should be understood that the concept of "descriptive information" included in the target case is similar to the concept of "first descriptive information", both of which are descriptive information used to describe the operations optimization problem. The concept of "mathematical model" included in the target case is similar to the concept of "first mathematical model", except that the descriptive information and mathematical model included in the target case are already stored in the first device, the first descriptive information is being obtained, and the first mathematical model requires the first machine learning model to generate it.

[0194] After obtaining the first description information (or the updated first description information), the first device can also obtain the description information included in each case in at least one case stored by the first device (hereinafter referred to as "fourth description information" for ease of description). That is, it can obtain at least one fourth description information corresponding to the at least one case, and further obtain the similarity between the first description information (or the updated first description information) and the fourth description information included in each case in the at least one case. Exemplarily, each case in the aforementioned at least one case (that is, including the target case) can be expressed in the form of <description information, mathematical model>, and the "description information" and "mathematical model" included in each case correspond to the same operations optimization problem.

[0195] Since steps 401 and 402 are optional, if steps 401 and 402 are performed, the first description information used in step 403 can be obtained through steps 401 and 402. If steps 401 and 402 are not performed, there may not be at least one target problem corresponding to the first operations research problem, and the first description information may also include a piece of description information input by the user, and the first description information at least indicates that a mathematical model corresponding to the operations research optimization problem needs to be generated.

[0196] The first device can use multiple methods to implement "obtaining the similarity between the first description information and the fourth description information included in each case in at least one case." In one implementation method, the first device can obtain the initial feature information of the first description information (or the updated first description information) and the initial feature information of the fourth description information included in each case in at least one case; for example, the initial feature information of the first description information is obtained by vectorizing (embedding) the first description information (or the updated first description information), and the initial feature information of the fourth description information is obtained by vectorizing the fourth description information. The first device can generate first similarity information, which includes the similarity between the initial feature information of the first description information and the initial feature information of each fourth description information in at least one fourth description information.

[0197] Optionally, the initial feature information of the first description information and the initial feature information of the fourth description information can be obtained through a fourth machine learning model. In one case, if the fourth machine learning model is specifically manifested as a large model, the fourth machine learning model, the third machine learning model, the third machine learning model and the first machine learning model can be the same machine learning model, or they can also be different machine learning models. Exemplarily, the first device can send the first description information (or the updated first description information) and all the fourth description information included in at least one case to a device deployed with a fourth machine learning model in the form of a large model, so as to process the first description information (or the updated first description information) and each fourth description information through the fourth machine learning model in the form of a large model, and obtain the initial feature information of the first description information output by the fourth machine learning model in the form of a large model and the initial feature information of each fourth description information.

[0198] For example, a device deployed with a fourth machine learning model in the form of a large model can input "Please vectorize the first description information" to the aforementioned fourth machine learning model, triggering the fourth machine learning model in the form of a large model to process the first description information, and obtain the initial feature information of the aforementioned first description information; the method of obtaining the "initial feature information of the fourth description information" can refer to the aforementioned description and will not be repeated here. It should be understood that the examples here are only for demonstrating the feasibility of this solution and are not used to limit this solution.

[0199] In another case, a fourth machine learning model may also be deployed in the first device. The first device inputs the first description information (or the updated first description information) and each fourth description information into the fourth machine learning model respectively, and vectorizes the first description information (or the updated first description information) and each fourth description information through the fourth machine learning model to obtain the initial feature information of the first description information and the initial feature information of each fourth description information.

[0200] The first device can obtain the first similarity information in a variety of ways. For example, the first device can determine the first similarity information based on the cosine similarity between the initial feature information of the first description information and the initial feature information of each fourth description information. For another example, the aforementioned "cosine similarity" can be replaced with "Euclidean distance," "Mahalanobis distance," "L1 distance," or other similarity calculation algorithms. The specific similarity can be determined based on the actual application scenario and is not exhaustive in the embodiments of this application.

[0201] In another implementation, the first device may obtain second information, and the second information includes at least one of the summary and keywords of the first descriptive information (or the updated first descriptive information); exemplarily, a fifth machine learning model may be deployed in the first device, and the first device inputs the first descriptive information into the fifth machine learning model, and extracts the summary and / or keywords of the first descriptive information through the fifth machine learning model to obtain the second information output by the fifth machine learning model. It should be noted that the examples given here are only to prove the feasibility of this solution, and the aforementioned steps can also be implemented through a large model, and other implementation methods will not be listed one by one here.

[0202] After obtaining the second information, the first device can determine the similarity between the second information and the fourth descriptive information included in each case in at least one case, wherein the similarity between the second information and the fourth descriptive information included in each case in at least one case is used as the similarity between the first descriptive information and the fourth descriptive information included in each case in at least one case.

[0203] For example, in one case, the first device can obtain the initial feature information of the second information and the initial feature information of each fourth description information, and then generate second similarity information, the second similarity information including the similarity between the initial feature information of the second information (i.e., the summary and / or keywords of the first description information) and the initial feature information of each fourth description information, that is, the similarity between the first description information and the fourth description information included in each case in at least one case is obtained. It should be noted that the specific method for obtaining the "initial feature information of the second information" can refer to the above description of the specific method for obtaining the "initial feature information of the first description information" and the "initial feature information of the fourth description information", and the specific method for obtaining the "second similarity information" can be generated by using the above description of the specific method for obtaining the "first similarity information", which will not be repeated here.

[0204] In another scenario, the first device may also obtain third information corresponding to each fourth descriptive information, that is, obtain at least one third information corresponding one-to-one to at least one fourth descriptive information, each third information including a summary and / or keywords of the fourth descriptive information. The first device obtains the initial feature information of the second information and the initial feature information of each third information, and may then generate third similarity information; wherein the third similarity information includes the similarity between the initial feature information of the second information and the initial feature information of each third information, and the similarity between the initial feature information of the second information and the initial feature information of each third information can be used as the similarity between the second information and each fourth descriptive information, that is, as the similarity between the first descriptive information and each fourth descriptive information.

[0205] In an embodiment of the present application, since the first description information may include text information entered by the user, and the text information entered by the user may carry invalid information, the summary and / or keywords of the first description information are first obtained, that is, the invalid information in the first description information is filtered out, and then the similarity between the summary and / or keywords of the first description information and the fourth description information is generated, which is conducive to obtaining more accurate similarity information, and thus conducive to obtaining a more matching first case.

[0206] In another implementation, after obtaining the first description information in text form (or the updated first description information), the first device may also convert the aforementioned text form into a graphical form, with the aforementioned at least one target question serving as a point of the first description information in graphical form (or the updated first description information), and the correct answer to each target question included in the first description information (or the updated first description information) serving as an edge of the first description information in graphical form (or the updated first description information). The first device may obtain initial feature information of the first description information in graphical form and initial feature information of each fourth description information, and then generate fourth similarity information, the fourth similarity information including the similarity between the initial feature information of the first description information in graphical form and the initial feature information of each fourth description information, the similarity between the initial feature information of the first description information in graphical form and the initial feature information of each fourth description information being used as the similarity between the first description information and each fourth description information, and so on.

[0207] It should be noted that the first device can also use other methods to obtain the similarity between the first description information (or the updated first description information) and the fourth description information included in each case in at least one case. The above examples of various specific implementation methods of the aforementioned steps are only for the convenience of understanding this solution and are not used to limit this solution.

[0208] Exemplarily, after determining the similarity between the first description information (or the updated first description information) and the fourth description information included in each case in at least one case, the first device can determine K fourth description information whose similarity with the first description information (or the updated first description information) meets the preset conditions from at least one fourth description information corresponding to the at least one case, and then obtain a first case to which each of the K fourth description information belongs, that is, it is possible to screen out K first cases from at least one case, and the similarity between the first description information and the fourth description information included in each of the K first cases meets the preset conditions, and K is an integer greater than or equal to 1.

[0209] The similarity between each of the K fourth description information and the first description information (or the updated first description information) is greater than or equal to a similarity threshold, and / or the K fourth description information include at least one of the fourth description information that is most similar to the first description information (or the updated first description information).

[0210] For example, the similarity threshold may be 80%, 85%, 90% or other values, etc., which are not limited in the embodiments of the present application.

[0211] For a more intuitive understanding of this solution, please refer to Figure 10, which is a flow chart of determining at least one first case from at least one case provided in an embodiment of the present application. The icons in Figure 10 represent a large-scale machine learning model, which can be used to generate a summary and keywords for each fourth description information in the case library, and then generate initial feature information for the summary and keywords of each fourth description information through the large-scale machine learning model.

[0212] It is also possible to generate a summary and keywords of the first description information through a machine learning model in the form of a large model, and then generate initial feature information of the summary and keywords of the first description information through a machine learning model in the form of a large model.

[0213] Based on the initial feature information of the summary and keywords of the first description information, as well as the initial feature information of the summary and keywords of each fourth description information, K fourth description information that are most similar to the first description information are determined. Since each of the K fourth description information belongs to a case in the case library, the K first cases that best match the first description information are obtained. It should be understood that the example in Figure 10 is only for the convenience of understanding this solution and is not used to limit this solution.

[0214] 404. Obtain first information according to the first description information and at least one first case.

[0215] In the embodiment of the present application, step 404 is an optional step. After the first device determines K first cases from at least one case through step 403, the first description information (or the updated first description information) and the K first cases can be combined to obtain first information, so as to process the first information through the first machine learning model in a subsequent step. If the first device does not obtain any first cases from at least one case through step 403, step 404 may not be performed. Alternatively, if step 403 is not performed, step 404 may not be performed.

[0216] In an embodiment of the present application, at least one first case that best matches the first descriptive information is obtained from at least one pre-stored case, and the similarity between the first descriptive information and the descriptive information included in each first case meets a preset condition, and then each first case and the first descriptive information are used as inputs to the first machine learning model, that is, each first case is used as a reference case for the first machine learning model, which is beneficial to assisting the first machine learning model in generating a more accurate first mathematical model, and is also beneficial to improving the efficiency of the first machine learning model in the process of generating the first mathematical model.

[0217] 405. Input the first information into a first machine learning model to obtain a first mathematical model corresponding to the first operations research optimization problem, wherein the first information is obtained based on the first descriptive information, the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem.

[0218] In an embodiment of the present application, steps 403 and 404 are both optional steps. In one case, if steps 403 and 404 are executed and K first cases are determined by step 403, the first information may include K first cases and first description information. In one implementation, the first machine learning model is a large model. Optionally, the first information includes first description information (or updated first description information) and prompt information (hereinafter referred to as "second prompt word" for the convenience of distinction), and the first device can combine the K first cases into the second prompt word in the form of <problem description, mathematical model> to obtain the first information.

[0219] Exemplarily, the K first cases include: {problem description 1, mathematical model 1}, {problem description 2, mathematical model 2},…, {problem description K, mathematical model K}, and the first information can be: "Prompt word (instruction)": "Please give a mathematical model under the new operations optimization problem description based on the provided <problem description, mathematical model> example. Example: {problem description 1, mathematical model 1}, {problem description 2, mathematical model 2},…, {problem description K, mathematical model K}", "Input (input)": "First description information". It should be understood that the examples here are only for the convenience of understanding this solution and are not used to limit this solution.

[0220] The phrase "the first device inputs the first information into the first machine learning model" can be understood as the first device sending the first information to a device deployed with the first machine learning model, which then inputs the first information into the first machine learning model in the form of a large model. The phrase "the first device obtains a first mathematical model corresponding to the first operations optimization problem" can be understood as the first device receiving the first mathematical model corresponding to the first operations optimization problem.

[0221] To understand this solution more intuitively, please refer to Figure 11, which is a flow chart of obtaining a first mathematical model based on first information provided in an embodiment of the present application. As shown in Figure 11, after obtaining K first cases, a second prompt word can be obtained based on the K first cases, and the second prompt word and the first descriptive information both belong to the first information; the second prompt word and the first descriptive information can be input into a first machine learning model in the form of a large model to obtain a first mathematical model corresponding to the first operations optimization problem output by the first machine learning model in the form of a large model. It should be understood that the example in Figure 11 is only for the convenience of understanding this solution and is not used to limit this solution.

[0222] In another case, if steps 403 and 404 are not performed, step 405 can be performed directly after step 402 is performed. Alternatively, if steps 403 and 404 are performed and no first case is determined through step 403, the first information and the first descriptive information can be the same information, and "the first device inputs the first information into the first machine learning model" can be understood as the first device sending the first information (which can also be understood as the first descriptive information) to the device on which the first machine learning model is deployed, and then the device capable of deploying the first machine learning model inputs the first information (which can also be understood as the first descriptive information) into the first machine learning model in the form of a large model.

[0223] After obtaining the first mathematical model corresponding to the first operations optimization problem, the first device can output the first mathematical model to the user through a display screen; the aforementioned output method includes but is not limited to: displaying the first mathematical model through a display screen, outputting a file containing the first mathematical model through a display screen, or other output methods, etc.

[0224] For a more intuitive understanding of this solution, please refer to Figure 12, which is a schematic diagram of outputting the first mathematical model through a display screen provided in an embodiment of the present application. In Figure 12, the first mathematical model is displayed through a display screen as an example. As shown in Figure 12, not only the objective function and constraints included in the first mathematical model are shown, but also the variables involved in the first mathematical model are shown. Optionally, the first case matched based on the first description information can also be displayed through the display screen, or the user can be informed through the display screen that no case is matched. It should be understood that the example in Figure 12 is only for the convenience of understanding this solution and is not used to limit this solution.

[0225] In addition, in order to understand this solution more intuitively, please refer to Figure 13, which is another schematic diagram of the method for obtaining a mathematical model provided in an embodiment of the present application. As shown in Figure 13, stage one: obtaining the first field to which the mathematical model to be established belongs. Stage two: at least one target problem corresponding to the first field adopts a directed graph data structure, and based on the at least one target problem in the form of the aforementioned directed graph, the answer to each first target problem is obtained in a guided manner, that is, the first descriptive information is obtained in a guided manner. Stage three: based on the first descriptive information, K first cases are determined from at least one case. Stage four: based on the K first cases, the second prompt word is determined, that is, the K first cases are combined into the second prompt word. Then, the second prompt word and the first descriptive information can be input into the first machine learning model in the form of a large model to obtain the first mathematical model generated by the first machine learning model. It should be understood that the first mathematical model includes an objective function and constraints. The example in Figure 13 is only for the convenience of understanding this solution and is not used to limit this solution.

[0226] Optionally, after the first device obtains the first mathematical model, the second device may also solve the first mathematical model to obtain a solution result of the first mathematical model, and the solution result of the first mathematical model is used to solve the first operations optimization problem.

[0227] Exemplarily, the second device and the first device may be the same device; or, the second device and the first device may be different devices. After obtaining the first mathematical model, the first device may send the first mathematical model to the second device.

[0228] For example, when the first domain to which the first operations optimization problem to be solved belongs is the domain of site selection, the solution of the first mathematical model can indicate which one or several locations in the set of candidate locations are the best locations. For another example, when the first domain to which the first operations optimization problem to be solved belongs is the domain of scheduling, the solution of the first mathematical model is used to inform how to arrange tasks under the constraints of limited resources to maximize work efficiency. For another example, when the first domain to which the first operations optimization problem to be solved belongs is the domain of order fulfillment, the solution of the first mathematical model is used to inform how to arrange the processing and delivery of orders under the premise of given orders, so as to optimally use resources to meet customer needs, etc. It should be noted that the examples here are only for the convenience of understanding the relationship between the "solution result of the first mathematical model" and the "first operations optimization problem" and are not used to limit this solution.

[0229] In order to have a more intuitive understanding of the beneficial effects brought about by the method provided by the present application, the beneficial effects brought about by the present application are explained below in combination with experimental data, and the experimental data are shown in Table 1 below.

[0230] Table 1

[0231] Referring to the above experimental data, it can be seen that the method provided in this application is used to obtain a mathematical model corresponding to the operations optimization problem, which can obtain a mathematical model that is more suitable for the operations optimization problem, that is, improve the accuracy of the obtained mathematical model and increase the speed of the process of obtaining the mathematical model.

[0232] On the basis of the embodiments corresponding to Figures 1 to 13, in order to better implement the above-mentioned scheme of the embodiment of the present application, the following also provides related equipment for implementing the above-mentioned scheme. Please refer to Figure 14 in detail. Figure 14 is a structural diagram of a device for obtaining a mathematical model provided in an embodiment of the present application. The device for obtaining a mathematical model 1400 includes: an output module 1401 for outputting at least one first question; an acquisition module 1402 for obtaining at least one answer based on at least one first question, and the at least one answer is used to obtain first descriptive information, wherein the first descriptive information is descriptive information for describing the first operational optimization problem, and the first problem is a problem used to obtain the descriptive information of the first operational optimization problem; a determination module 1403 for determining the first information based on the first descriptive information; an input module 1404 for inputting the first information into a machine learning model to obtain a first mathematical model, the first mathematical model including an objective function and constraints, and the first mathematical model is used to solve the first operational optimization problem.

[0233] Optionally, the at least one first question is determined based on a pre-stored first question set.

[0234] Optionally, the first question set is expressed in a data form of a directed graph, and the data form of the directed graph indicates an appearance order between at least two different questions in the first question set.

[0235] Optionally, the acquisition module 1402 is also used to obtain second descriptive information, wherein the first question set includes at least one first question and a second question corresponding to the second descriptive information, the second descriptive information is obtained through user input, the second question is the question used to obtain the descriptive information of the first operations optimization problem, and the second descriptive information includes the answer to the second question; the determination module 1403 is specifically used to determine the first information based on the first descriptive information and the second descriptive information, and the first information includes the first descriptive information and the second descriptive information.

[0236] Optionally, the mathematical model acquisition device 1400 is applied to the first device, and at least one case is stored in the first device. Any one of the at least one case includes descriptive information corresponding to the operations optimization problem and a mathematical model. The determination module 1403 is specifically used to: determine at least one first case from the at least one case based on the similarity between the first descriptive information and each descriptive information included in the at least one case, wherein the similarity between the first descriptive information and the descriptive information included in the first case meets a preset condition; and obtain first information based on the first descriptive information and the at least one first case.

[0237] Optionally, the acquisition module 1402 is also used to obtain second information, the second information including a summary of the first descriptive information and at least one of the keywords of the first descriptive information; the determination module 1403 is also used to determine the similarity between the second information and each descriptive information included in at least one case, wherein the similarity between the second information and each descriptive information included in at least one case is used as the similarity between the first descriptive information and each descriptive information included in at least one case.

[0238] Optionally, the determination module 1403 is further used to determine the field to which the mathematical model to be established belongs; the determination module 1403 is further used to determine a first set of questions from at least one pre-stored set of questions based on the field to which the mathematical model to be established belongs, and the at least one pre-stored set of questions corresponds one-to-one to at least one field; the determination module 1403 is further used to determine at least one first question based on the first set of questions.

[0239] Optionally, the field to which the mathematical model to be established belongs includes any one of the following: site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field or production planning problem field.

[0240] It should be noted that the information interaction, execution process, etc. between the modules / units in the mathematical model acquisition device 1400 are based on the same concept as the various method embodiments corresponding to Figures 1 to 13 in this application. For specific contents, please refer to the description in the method embodiments shown above in this application, and will not be repeated here.

[0241] Please continue to refer to Figure 15, which is another structural schematic diagram of the mathematical model acquisition device provided in an embodiment of the present application. The mathematical model acquisition device 1500 is applied to a first device, and at least one case is stored in the first device. Any one of the at least one case includes descriptive information and a mathematical model corresponding to the operations optimization problem. The mathematical model acquisition device 1500 includes: an acquisition module 1501, which is used to obtain first descriptive information, where the first descriptive information is descriptive information used to describe the first operations optimization problem; a determination module 1502, which is used to determine at least one second mathematical model from at least one mathematical model included in the at least one case based on the similarity between the first descriptive information and each descriptive information included in the at least one case, where the second mathematical model belongs to the first case in the at least one case, and the similarity between the first descriptive information and the descriptive information included in the first case meets a preset condition; a processing module 1503, which is used to obtain first information based on the first descriptive information and the at least one second mathematical model; and an input module 1504, which is used to input the first information into a machine learning model to obtain a first mathematical model, wherein the first mathematical model is output through a display screen, the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations optimization problem.

[0242] Optionally, the acquisition module 1501 is specifically used to obtain first descriptive information corresponding to at least one first question, the first descriptive information includes the answer to each first question in the at least one first question, and the answer to each first question in the at least one first question is used to obtain the first descriptive information, wherein the first question is the question used to obtain the descriptive information of the first operations optimization problem.

[0243] It should be noted that the information interaction, execution process, etc. between the modules / units in the mathematical model acquisition device 1500 are based on the same concept as the various method embodiments corresponding to Figures 1 to 13 in this application. For specific contents, please refer to the description in the method embodiments shown above in this application, and will not be repeated here.

[0244] Next, a device provided in an embodiment of the present application will be described. Please refer to Figure 16, which is a schematic structural diagram of a device provided in an embodiment of the present application. Specifically, the device 1600 includes: a receiver 1601, a transmitter 1602, a processor 1603, and a memory 1604 (wherein the number of processors 1603 in the device 1600 may be one or more, and Figure 16 uses one processor as an example). The processor 1603 may include an application processor 16031 and a communication processor 16032. In some embodiments of the present application, the receiver 1601, the transmitter 1602, the processor 1603, and the memory 1604 may be connected via a bus or other means.

[0245] Memory 1604 may include read-only memory and random access memory, and provides instructions and data to processor 1603. A portion of memory 1604 may also include non-volatile random access memory (NVRAM). Memory 1604 stores processor and operation instructions, executable modules, or data structures, or subsets or extended sets thereof. The operation instructions may include various operation instructions for implementing various operations.

[0246] Processor 1603 controls the operation of the device. In specific applications, the various components of the device are coupled together via a bus system. In addition to a data bus, the bus system may also include a power bus, a control bus, and a status signal bus. However, for clarity, all bus systems are referred to as a bus system in the figure.

[0247] The methods disclosed in the above embodiments of the present application can be applied to or implemented by processor 1603. Processor 1603 can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in processor 1603 or software instructions. The above processor 1603 can be a general-purpose processor, a digital signal processor (DSP), a microprocessor, or a microcontroller, and can further include an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The processor 1603 can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in conjunction with the embodiments of the present application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in memory 1604, and processor 1603 reads the information in memory 1604 and, in conjunction with its hardware, completes the steps of the above method.

[0248] Receiver 1601 can be used to receive input digital or character information and generate signal input related to device settings and function control. Transmitter 1602 can be used to output digital or character information through the first interface. Transmitter 1602 can also be used to send instructions to the disk pack through the first interface to modify data in the disk pack. Transmitter 1602 can also include a display device such as a display screen.

[0249] In the embodiment of the present application, the processor 1603 is used to execute the method for obtaining the mathematical model executed by the first device in the embodiments corresponding to Figures 1 to 13. It should be noted that the specific manner in which the application processor 16031 in the processor 1603 executes the aforementioned steps is based on the same concept as the various method embodiments corresponding to Figures 1 to 13 in the present application, and the technical effects brought about are the same as the various method embodiments corresponding to Figures 1 to 13 in the present application. For specific details, please refer to the description of the method embodiments shown above in the present application, and will not be repeated here.

[0250] A computer-readable storage medium is also provided in an embodiment of the present application. The computer-readable storage medium stores a program for signal processing. When the program is run on a computer, the computer executes the steps performed by the first device in the method described in the embodiments shown in Figures 1 to 13 above.

[0251] An embodiment of the present application also provides a computer program product, which includes a program. When the program is run on a computer, it enables the computer to execute the steps executed by the first device in the method described in the embodiments shown in Figures 1 to 13 above.

[0252] The first device provided in the embodiment of the present application can be specifically a chip, and the chip includes: a processing unit and a communication unit. The processing unit can be, for example, a processor, and the communication unit can be, for example, an input / output interface, a pin, or a circuit. The processing unit can execute the computer-executable instructions stored in the storage unit to enable the chip to execute the method for obtaining the mathematical model described in the embodiments shown in Figures 1 to 13 above. Optionally, the storage unit is a storage unit within the chip, such as a register, a cache, etc. The storage unit can also be a storage unit located outside the chip within the wireless access device, such as a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM), etc.

[0253] The processor mentioned in any of the above places can be a general-purpose central processing unit, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the program of the above-mentioned first aspect method.

[0254] It should also be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided in this application, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.

[0255] Through the description of the above embodiments, it is clear to those skilled in the art that the present application can be implemented by means of software plus necessary general-purpose hardware, and of course it can also be implemented by means of dedicated hardware including application-specific integrated circuits, dedicated CPUs, dedicated memories, dedicated components, etc. In general, all functions performed by computer programs can be easily implemented with corresponding hardware, and the specific hardware structures used to implement the same function can also be various, such as analog circuits, digital circuits, or dedicated circuits, etc. However, for the present application, software program implementation is a better implementation method in most cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer's floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes a number of instructions to enable a computer device (which can be a personal computer, a first device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0256] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0257] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a first device or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode to another website, a computer, a first device or a data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a first device, a data center, etc. that includes one or more available media integrations. The available medium can be a magnetic medium, (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state drive (SSD)).

Claims

1. A method for obtaining a mathematical model, characterized in that: The method comprises: Outputting at least one first question, obtaining at least one answer based on the at least one first question, wherein the at least one answer is used to obtain first description information, wherein the first description information is description information used to describe a first operations research optimization problem, and the first question is a question used to obtain the description information of the first operations research optimization problem; Determine first information according to the first description information; The first information is input into a machine learning model to obtain a first mathematical model, wherein the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem.

2. The method according to claim 1, characterized in that The at least one first question is determined based on a pre-stored first question set.

3. The method according to claim 2, characterized in that The first question set is expressed in a data form of a directed graph, and the data form of the directed graph indicates an appearance order between at least two different questions in the first question set.

4. The method according to claim 2 or 3, characterized in that: The method further comprises: Obtaining second description information, wherein the first question set includes the at least one first question and a second question corresponding to the second description information, the second description information is obtained through user input, the second question is a question used to obtain the description information of the first operations research optimization problem, and the second description information includes an answer to the second question; The determining the first information according to the first description information includes: The first information is determined according to the first description information and the second description information, where the first information includes the first description information and the second description information.

5. The method according to any one of claims 1 to 3, characterized in that: The method is applied to a first device, wherein at least one case is stored in the first device, and any one of the at least one case includes description information and a mathematical model corresponding to an operations research optimization problem, and determining first information according to the first description information includes: Determine at least one first case from the at least one case according to the similarity between the first description information and each description information included in the at least one case, wherein the similarity between the first description information and the description information included in the first case satisfies a preset condition; The first information is obtained according to the first description information and the at least one first case.

6. The method according to claim 5, characterized in that The method further comprises: Acquire second information, where the second information includes at least one of a summary of the first description information and a keyword of the first description information; Determine the similarity between the second information and each descriptive information included in the at least one case, wherein the similarity between the second information and each descriptive information included in the at least one case is taken as the similarity between the first descriptive information and each descriptive information included in the at least one case.

7. The method according to claim 2 or 3, characterized in that: Before outputting at least one first question, the method further includes: Determine the field to which the mathematical model to be established belongs; According to the field to which the mathematical model to be established belongs, determining the first set of questions from at least one set of questions stored in advance, wherein the at least one set of questions stored in advance corresponds to at least one field in a one-to-one manner; The at least one first question is determined based on the first question set.

8. The method according to claim 7, characterized in that The field to which the mathematical model to be established belongs includes any of the following: site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field or production planning problem field.

9. A method for obtaining a mathematical model, characterized in that: The method is applied to a first device, wherein at least one case is stored in the first device, and any one of the at least one case includes description information and a mathematical model corresponding to an operations research optimization problem. The method includes: Acquire first description information, where the first description information is description information used to describe a first operations optimization problem; Determining at least one second mathematical model from at least one mathematical model included in the at least one case according to the similarity between the first description information and each description information included in the at least one case, the second mathematical model belonging to a first case in the at least one case, and the similarity between the first description information and the description information included in the first case meets a preset condition; Obtaining first information according to the first description information and the at least one second mathematical model; The first information is input into a machine learning model to obtain a first mathematical model, wherein the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem.

10. The method according to claim 9, characterized in that The obtaining of the first description information includes: Obtain first description information corresponding to at least one first question, the first description information including an answer to each of the at least one first question, the answer to each of the at least one first question being used to obtain first description information, wherein the first question is a question used to obtain description information of the first operations optimization problem.

11. A device for acquiring a mathematical model, characterized in that: The device comprises: An output module, configured to output at least one first question; an acquisition module, configured to acquire at least one answer based on the at least one first question, wherein the at least one answer is used to obtain first description information, wherein the first description information is description information used to describe a first operations research optimization problem, and the first question is a question used to acquire the description information of the first operations research optimization problem; A determination module, configured to determine first information according to the first description information; An input module is used to input the first information into a machine learning model to obtain a first mathematical model, wherein the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem.

12. The device according to claim 11, characterized in that The at least one first question is determined based on a pre-stored first question set.

13. The device according to claim 12, characterized in that The first question set is expressed in a data form of a directed graph, and the data form of the directed graph indicates an appearance order between at least two different questions in the first question set.

14. The device according to claim 12 or 13, characterized in that The acquisition module is further used to acquire second description information, wherein the first problem set includes the at least one first problem and the second problem corresponding to the second description information, the second description information is obtained through user input, the second problem is a problem used to acquire the description information of the first operations optimization problem, and the second description information includes an answer to the second problem; The determination module is specifically configured to determine the first information according to the first description information and the second description information, where the first information includes the first description information and the second description information.

15. The device according to any one of claims 11 to 13, characterized in that The apparatus is applied to a first device, wherein at least one case is stored in the first device, and any one of the at least one case includes description information and a mathematical model corresponding to an operations optimization problem, and the determination module is specifically used to: Determine at least one first case from the at least one case according to the similarity between the first description information and each description information included in the at least one case, wherein the similarity between the first description information and the description information included in the first case satisfies a preset condition; The first information is obtained according to the first description information and the at least one first case.

16. The device according to claim 15, characterized in that The acquisition module is further configured to acquire second information, wherein the second information includes at least one of a summary of the first description information and a keyword of the first description information; The determination module is also used to determine the similarity between the second information and each descriptive information included in the at least one case, wherein the similarity between the second information and each descriptive information included in the at least one case is taken as the similarity between the first descriptive information and each descriptive information included in the at least one case.

17. The device according to claim 12 or 13, characterized in that The determination module is also used to determine the field to which the mathematical model to be established belongs; The determination module is further used to determine the first set of questions from at least one pre-stored set of questions according to the field to which the mathematical model to be established belongs, wherein the at least one pre-stored set of questions corresponds to at least one field in a one-to-one manner; The determination module is further used to determine the at least one first question according to the first question set.

18. The device according to claim 17, characterized in that The field to which the mathematical model to be established belongs includes any of the following: site selection problem field, scheduling problem field, order fulfillment field, supply chain field, packing problem field, transportation field, resource allocation field, revenue management field or production planning problem field.

19. A device for acquiring a mathematical model, characterized in that: The mathematical model acquisition device is applied to a first device, wherein at least one case is stored in the first device, and any one of the at least one case includes description information and a mathematical model corresponding to an operations research optimization problem, and the device includes: An acquisition module, used to acquire first description information, where the first description information is description information used to describe a first operations optimization problem; a determination module, configured to determine at least one second mathematical model from at least one mathematical model included in the at least one case according to the similarity between the first description information and each description information included in the at least one case, the second mathematical model belonging to a first case in the at least one case, and the similarity between the first description information and the description information included in the first case meeting a preset condition; A processing module, configured to obtain first information according to the first description information and the at least one second mathematical model; An input module is used to input the first information into a machine learning model to obtain a first mathematical model, wherein the first mathematical model includes an objective function and constraints, and the first mathematical model is used to solve the first operations research optimization problem.

20. The device according to claim 19, characterized in that The acquisition module is specifically used to obtain first descriptive information corresponding to at least one first problem, the first descriptive information includes the answer to each first question in the at least one first problem, and the answer to each first question in the at least one first problem is used to obtain the first descriptive information, wherein the first problem is the problem used to obtain the descriptive information of the first operations optimization problem.

21. A device, characterized in that comprising a processor and a memory, the processor being coupled to the memory, The memory is used to store programs; The processor is configured to execute the program in the memory so that the device performs the method according to any one of claims 1 to 10.

22. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, and when the program is executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 10.

23. A computer program product, characterized in that The computer program product comprises a program, and when the program is run on a computer, the computer is caused to execute the method according to any one of claims 1 to 10 .

24. An operations optimization method, characterized in that: The method includes: solving a first mathematical model to obtain a solution result of the first mathematical model, wherein the first mathematical model includes an objective function and constraints, the first mathematical model is used to solve a first operations research optimization problem, and the first mathematical model is obtained based on the method described in any one of claims 1 to 10.

Images