A method for selecting and matching audit experts and a management system

Abstract

Multiple embodiments of this specification relate to information technology, and more particularly to a method and management system for selecting and matching review experts. The method includes the steps of: establishing an abstract extraction model for each expert and associating it with the type of review project; reading the materials of the project to be reviewed, and using the abstract extraction model corresponding to the type of review project to extract the abstract of the materials of the project to be reviewed, which is denoted as the project abstract; establishing a simulation review model for each expert; reading the project information of the project to be reviewed, and generating a confused abstract and confused information according to the project abstract and project information of the project to be reviewed; inputting the confused abstract and confused information into the simulation review model of each expert to obtain a confused review text; comparing it with the review rules corresponding to the type of review project to obtain a review result; when the review result matches the predetermined review result, including the corresponding expert in the candidate library; and selecting a predetermined number of experts as the review experts for the project to be reviewed.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to information technology, and specifically to a method and management system for selecting and matching review experts. Background Art

[0002] In various scientific research, engineering, and technology development projects, independent evaluation of the project by introducing external or internal professionals can provide an objective basis for decision-making. The core of expert review lies in using the knowledge and experience of authoritative figures in specific fields to review all aspects of the project. These experts usually come from academia, industry, or other institutions with relevant professional knowledge and rich practical experience. They can examine the project from different perspectives, put forward review opinions, and help control the overall level of the project. When a project enters the review stage, the organizer will carefully select review experts according to the nature and technical requirements of the project. The selection process not only considers the candidate's professional background but also evaluates their past work experience, research achievements, and the ability to handle similar complex problems. In addition, to ensure the fairness of the review, it is necessary to exclude personnel with direct interests in the project to avoid potential conflicts of interest. After selecting the review experts, they will receive detailed project materials, including but not limited to project proposals, technical documents, progress reports, etc. The experts will carefully study these materials and then conduct in-depth analysis of the scientific rationality, technological innovation, feasibility, etc. of the project, and give review opinions and explanations. Summary of the Invention

[0003] Multiple embodiments of this specification describe a method and management system for selecting and matching review experts.

[0004] In a first aspect, embodiments of this specification provide a method for selecting and matching review experts, including the steps of:

[0005] Establish a summary extraction model for each expert and associate it with the review project type;

[0006] Read the materials of the project to be reviewed, and use the summary extraction model corresponding to the review project type to extract the summary of the materials of the project to be reviewed, denoted as the project summary;

[0007] Establish a simulation review model for each expert, where the input of the simulation review model is the project summary and project information, and the output is a formatted text containing predetermined fields, denoted as the review text;

[0008] Read the project information of the project to be reviewed, and generate a confused summary and confused information based on the project summary and project information of the project to be reviewed, where the confused summary and confused information have a known predetermined review result;

[0009] Input the confusion summary and confusion information into the simulation review model of each expert to obtain the confusion review text;

[0010] Compare the review text and the confusion review text with the review rules of the corresponding review project types respectively to obtain the review results;

[0011] When the review results match the predetermined review results, include the corresponding experts in the candidate library;

[0012] Randomly select a predetermined number of experts from the candidate library as the review experts for the project to be reviewed.

[0013] In a second aspect, an embodiment of the present specification provides a review expert selection and matching management system, including:

[0014] An extraction module that establishes a summary extraction model for each expert and associates the review project types;

[0015] A summary module that reads the materials of the project to be reviewed and extracts the summary of the materials of the project to be reviewed using the summary extraction model of the corresponding review project type, denoted as the project summary;

[0016] A review module that establishes a simulation review model for each expert, where the input of the simulation review model is the project summary and project information, and the output is a formatted text containing predetermined fields, denoted as the review text;

[0017] A confusion module that reads the project information of the project to be reviewed and generates a confusion summary and confusion information based on the project summary and project information of the project to be reviewed, where the confusion summary and confusion information have known predetermined review results;

[0018] A confusion review module that inputs the confusion summary and confusion information into the simulation review model of each expert to obtain the confusion review text;

[0019] A comparison module that compares the review text and the confusion review text with the review rules of the corresponding review project types respectively to obtain the review results;

[0020] A candidate module that includes the corresponding experts in the candidate library when the review results match the predetermined review results;

[0021] A selection module that randomly selects a predetermined number of experts from the candidate library as the review experts for the project to be reviewed.

[0022] In a third aspect, an embodiment of the present specification provides an electronic device, including a processor and a memory;

[0023] The processor is connected to the memory;

[0024] The memory is used to store executable program code;

[0025] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method described in any of the above aspects.

[0026] In a fourth aspect, an embodiment of the present specification provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0027] In a fifth aspect, an embodiment of the present specification provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0028] The beneficial effects brought by the technical solutions provided in some embodiments of the present specification at least include:

[0029] In multiple embodiments of the present specification, the provided method and management system for selecting and matching review experts, by establishing an abstract extraction model for each expert, can not only facilitate the expert to obtain the abstracts of the projects suitable for or preferred by him / her for review, but also associate the review project types, achieving the effect of distinguishing the technical fields of experts and project types through the abstract extraction model. On the other hand, through the abstract extraction model, it is also possible to achieve the distinction of sub-technical fields, making the screening of the technical backgrounds of experts more refined and convenient, realizing a better match between the technical backgrounds of experts and the projects to be reviewed, and helping to improve the review effect. By establishing a simulated review model, it can not only be used to screen experts, but also predict the review results, which is helpful for the arrangement of review activities.

[0030] Other features and advantages of multiple embodiments of the present specification will be further revealed in the following specific embodiments and drawings. Description of the Drawings

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present specification, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0032] Figure 1 It is a schematic diagram of the application scenario of the embodiment of the present specification.

[0033] Figure 2 It is a schematic diagram of the selection and matching of review experts provided by the embodiment of the present specification.

[0034] Figure 3Schematic diagram of the application architecture of the review expert selection and matching method provided by the embodiments of this specification.

[0035] Figure 4 Schematic diagram of the interaction interface provided by the embodiments of this specification.

[0036] Figure 5 Schematic diagram of the process of the review expert selection and matching method provided by the embodiments of this specification.

[0037] Figure 6 Schematic diagram of the process of the abstract extraction model establishment method provided by the embodiments of this specification.

[0038] Figure 7 Schematic diagram of the process of the simulation review model establishment method provided by the embodiments of this specification.

[0039] Figure 8 Schematic diagram of the review expert selection and matching management system provided by the embodiments of this specification.

[0040] Figure 9 Schematic diagram of the electronic device provided by the embodiments of this specification. Detailed implementation manners

[0041] The technical solutions of the embodiments of this specification will be explained and described below with reference to the accompanying drawings of the embodiments of this specification. However, the following embodiments are only the preferred embodiments of this specification, not all of them. Based on the embodiments in the implementation manners, other embodiments obtained by those skilled in the art without creative efforts all fall within the protection scope of this specification.

[0042] Terms such as "first", "second", "third", etc. in the specification, claims and the above-mentioned drawings of this specification are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally further include steps or units not listed, or may optionally further include other steps or units inherent to these processes, methods, products or devices.

[0043] In the following description, terms such as "inner", "outer", "upper", "lower", "left", "right", etc. indicating orientation or positional relationship are only for the convenience of describing the embodiments and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of this specification.

[0044] The data involved in this application are all information and data authorized by users or fully authorized by all parties, and the collection of relevant data complies with the relevant laws, regulations and standards of relevant countries and regions.

[0045] Before introducing the technical solutions described in this specification, the application scenarios of the technical solutions and related technologies are introduced.

[0046] The expert review project activity is to evaluate the submitted research plans, results or applications. In the expert review project activity, it is necessary to determine the purpose, scope, and standards and principles to be followed for the review. For example, in the allocation of scientific research funds, the review purpose is to select the most promising research topics; while in the process of formulating technical standards, more attention is paid to the feasibility and safety of the solutions.

[0047] When forming the review team, it is crucial to select suitable review experts 11. The review experts 11 should be scholars or professionals with profound attainments and good reputations in specific fields. When selecting the review experts 11, multiple factors such as their professional backgrounds, research interests, and past performances often need to be considered, and conflicts of interest should be avoided as much as possible.

[0048] The materials 20 of the project usually include the detailed description of the project, expected goals, technical roadmap, budget arrangement, etc. The review experts 11 will score or rate each project according to the established standards, and usually need to pay attention to innovation, feasibility, influence, and benefits. The review experts 11 may also put forward constructive opinions and suggestions to help improve the project. Finally, a review conclusion will be formed based on the summary and analysis of the opinions of all review experts 11. This conclusion is not only the key basis for deciding whether the project can obtain support, but also provides a guiding direction for subsequent improvement and perfection.

[0049] The selection and matching of review experts 11 is an important link in modern scientific research management, project evaluation, and academic evaluation systems. In the early stage, the selection of review experts 11 mainly relied on the personal experience and network of the organizers or responsible persons, which was inefficient and easily affected by subjective factors. With the rapid development of information technology, the technology of selecting and matching review experts 11 has also experienced a transformation from manual selection to intelligent matching. Please refer to the appendix Figure 1 , the selection and matching of experts is to select and match several suitable review experts 11 from the expert database 10 for each project. At present, the selection of review experts 11 is still only matched according to the professional backgrounds of the experts and other situations, and there is still a lack of basis for judging the review effect, and the matching is not fine enough.

[0050] Therefore, this specification provides a method for selecting and matching review experts 11. Please refer to the appendix Figure 2, by extracting the project summary 21 from the materials 20 of the project to be reviewed and associating the project information 22. Generate the obfuscated summary 23 and obfuscated information 24 according to the project summary 21 and project information 22. Input the project summary 21, project information 22, obfuscated summary 23 and obfuscated information 24 into the simulation review model 30 constructed for each review expert 11, and realize the selection and matching of experts according to the output result of the simulation review model 30. Since the simulation review model 30 is established for each expert and can reflect the professional knowledge scope and review ability of the expert, it has a more accurate effect of knowledge matching and selection.

[0051] The review expert 11 selection and matching method provided in this specification is applied to a system architecture such as Figure 3 shown. Figure 3 This is an architecture schematic diagram of the system architecture in the embodiment of the present application. The system architecture includes a server 40 and a terminal device 50, and an interaction interface 51 is provided on the terminal device 50. Please refer to the appendix Figure 4 . Among them, the interaction interface 51 can run on the terminal device 50 in the form of a browser, or can run on the terminal device 50 in the form of an independent application (APP), etc. The specific display form of the client is not limited here. The server 40 involved in this application can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. The terminal device 50 can be a smart phone, a tablet computer, a notebook computer, a handheld computer, a personal computer, a smart speaker, a smart TV, a smart watch, a vehicle-mounted device, a wearable device, etc., but is not limited thereto. The terminal device 50 and the server 40 can be directly or indirectly connected through wired or wireless communication methods, and this application does not make any restrictions here. The number of the server 40 and the terminal device 50 is also not restricted.

[0052] Specifically, please refer to the appendix Figure 5 , the review expert 11 selection and matching method includes the steps:

[0053] Step S101) Establish a summary extraction model for each expert and associate the review project type.

[0054] In this embodiment, an abstract extraction model is established for each expert, and each abstract extraction model is associated with an audit project type. That is to say, when each expert audits an audit project type, an abstract extraction model will be established. Specifically, when expert A audits the natural science foundation application project, an abstract extraction model is established, and the audit project type associated with this abstract extraction model is the natural science foundation application project in the software field. When expert A audits the ITSS certification, an abstract extraction model is also established, and the audit project type associated with this abstract extraction model is the ITSS certification project. When expert B audits the natural science foundation application project, an abstract extraction model is established, and the audit project type associated with this abstract extraction model is the natural science foundation application project in the precision manufacturing field. It can be seen that the knowledge background of expert A is in the software field, and the knowledge background of expert B is in the mechanical manufacturing field. Moreover, when expert A reviews different projects, although the review is based on his knowledge in the software field, the extracted abstracts are different, so different abstract extraction models are established.

[0055] The establishment of the abstract extraction model is mainly achieved by collecting and using sample data. Specifically, the method for establishing the abstract extraction model for each expert and associating it with the audit project type includes the following steps:

[0056] Traverse each audit project type in turn and execute the following steps. Please refer to the appendix Figure 6 , including:

[0057] Step S201) Read the materials of multiple projects under the current audit project type, and generate multiple project abstracts 21 for each material 20.

[0058] Among the generated multiple project abstracts 21, some project abstracts 21 comprehensively reflect the key content of the material 20, some project abstracts 21 only reflect part of the key content of the material 20. Some project abstracts 21, although comprehensively reflecting the key content of the material 20, use inaccurate expressions such as words and sentences, and even some errors. These project abstracts 21 are all presented to the expert.

[0059] Step S202) Receive the marking of the reflectivity scores of some of these project abstracts 21. Some of these project abstracts 21 have manually marked reflectivity scores. The source of this reflectivity score is the staff, and the source of the staff is the experts already in the database, or the experts or scholars who have been determined to be in the corresponding field.

[0060] In step S203), when an expert is put into the database, the materials 20 of the project are shown to the expert, and then multiple project abstracts 21 of the project are shown to the expert. When an expert is put into the database, an abstract extraction model is established for this expert, and multiple abstract extraction models will be established. The types of review projects associated with each abstract extraction model are different. At the same time, the type of review project associated with the expert's abstract extraction model also determines the field in which the expert can conduct reviews, that is, the field of the expert is determined.

[0061] Step S204) Receive the reflectivity score of the project abstract 21 provided by the expert. The expert put into the database gives a reflectivity score to the provided project abstract 21. At this time, the project abstracts 21 shown to the expert are not divided by technical field or review project type.

[0062] In fact, according to whether the review project type itself has a technical field, there are the following situations: When the review project type itself faces multiple technical fields, such as the natural science research fund application project, which involves software field, chemical engineering field, mechanical manufacturing field, and also involves mathematical research, social research, etc., and has a relatively broad technical field. The ITSS certification project is a project exclusive to the software field.

[0063] At this time, the materials 20 and project abstracts 21 of the natural science research fund application projects in multiple fields will be shown to expert A. For the materials 20 and multiple project abstracts 21 of the projects in the software field, expert A can mark the reflectivity score more accurately. For those in non-software fields, such as the materials 20 and project abstracts 21 of the natural science research fund application project in the chemical engineering field, expert A will not be able to accurately mark the reflectivity score of the project abstract 21 because he / she doesn't understand.

[0064] Step S205) Obtain the proximity of the reflectivity score of the project abstract 21 marked with the reflectivity score by the expert.

[0065] In this way, in step S205), when expert A marks, the reflectivity scores of the materials 20 and project abstracts 21 of the projects belonging to the chemical engineering field will be inaccurate. At this time, when compared with those project abstracts 21 whose reflectivity scores have been manually marked in advance, it will be found that the proximity is low. It can be determined that expert A is not suitable for the natural science fund application projects in the chemical engineering field. That is, when the review project type is "natural science fund application project in the chemical engineering field", the abstract extraction model belonging to expert A will not be selected, and expert A will not be chosen either. Similarly, the reflectivity scores marked by expert A for the materials 20 and project abstracts 21 of the projects belonging to the software field will be relatively accurate. At this time, compared with those reflectivity scores manually marked in advance, the proximity is higher. It can be determined that when the review project type is "natural science fund application project in the software field", expert A may be selected.

[0066] Step S206): When the proximity is greater than a preset threshold, associate the expert with the current type of review project. That is, when the proximity corresponding to Expert A is greater than the preset threshold, Expert A will be associated with the "Natural Science Foundation Application Project in the Software Field".

[0067] Step S207): Establish sample data according to the reflectivity score of the expert for the project abstract 21 without a reflectivity score mark. The sample data includes the material 20, the project abstract 21, and the reflectivity score. The reflectivity score of Expert A for the project abstract 21 without a reflectivity score mark in the software field represents the abstract extraction result with the characteristics of Expert A. Based on this, sample data is established, so that an abstract extraction model suitable for Expert A can be trained based on the sample data. In subsequent steps, the abstract extraction model trained from these sample data will be associated with the "Natural Science Foundation Application Project in the Software Field" and will also be associated with Expert A. Similarly, the proximity between the reflectivity score of Expert A for the material 20 and its project abstract 21 belonging to the "ITSS Certification Project" and the pre-marked reflectivity score is also greater than the threshold. Thus, sample data in the "ITSS Certification Project" field can be obtained subsequently. The abstract extraction model trained from these sample data will be associated with the "ITSS Certification Project" and will also be associated with Expert A.

[0068] Step S208): Obtain an abstract extraction model according to the sample data, and associate it with the current type of review project and the expert.

[0069] On the other hand, in another embodiment, the method for obtaining an abstract extraction model according to the sample data includes:

[0070] Establish and use the sample data to train a machine learning model, and obtain an abstract extraction model according to the machine learning model.

[0071] On the other hand, in another embodiment, the method for obtaining an abstract extraction model according to the sample data includes:

[0072] Generate a natural language task according to the sample data, input the natural language task into a pre-connected large language model, and obtain an abstract extraction model according to the large language model.

[0073] On the other hand, in another embodiment, the method for obtaining an abstract extraction model according to the sample data includes:

[0074] Generate a natural language task according to the sample data, and input the natural language task into a pre-connected large language model;

[0075] A machine learning model is established and trained under the guidance of the large language model, and an abstract extraction model is obtained based on the trained machine learning model.

[0076] Alternatively, other techniques already disclosed in the art can be used to implement the method of obtaining the abstract extraction model according to the sample data.

[0077] Step S102) Read the materials 20 of the project to be audited, and use the abstract extraction model corresponding to the type of the audit project to extract the abstract of the materials 20 of the project to be audited, denoted as the project abstract 21.

[0078] According to the foregoing solution, an abstract extraction model for each expert is established, and each abstract extraction model is associated with an expert and the type of the audit project. For the materials 20 of the project to be audited, an abstract extraction model with a matching type can be selected according to the type of the audit project. In step S102), there are multiple abstract extraction models that match the type of the audit project, and each abstract extraction model is associated with an expert. At this time, the selection and matching of the experts have not been completed. Therefore, in step S102), an abstract extraction model is randomly selected, without considering whether the associated expert will be selected, nor considering the differences between the abstract extraction models associated with different experts. It is considered that most of the abstract extraction models of this type of audit project can generally extract the project abstract 21 of the materials 20.

[0079] Step S103) Establish a simulation audit model 30 for each expert. The input of the simulation audit model 30 is the project abstract 21 and the project information 22, and the output is a formatted text containing predetermined fields, denoted as the audit text 31.

[0080] The project information 22 includes the project name, project purpose, planned start time of the project, project cycle, project execution personnel, project budget, etc. There will be an overlapping part between the project information 22 and the project abstract 21, which does not affect the implementation of this embodiment. For example, the project name usually appears in the materials 20 and will thus be extracted by the abstract extraction model and put into the project abstract 21. The project cycle may not be extracted as a keyword by the abstract extraction model. The combination of the project abstract 21 and the project information 22 can provide more comprehensive content for the audit.

[0081] The reviewed text 31 is formatted text. Exemplarily, the formatted text is in the form of key-value pairs, such as project name: XX project, project budget: 1 million yuan, project executors: {A, B, C, D}, project planned start time: 2023, project success probability: 80%, professional matching degree of project executors: 68%, project expected benefit evaluation: high expected benefit, etc. Among them, the project name, project budget, project executors, and project planned start time can be directly obtained from the project summary 21 or project information 22. The project success probability, professional matching degree of project executors, and project expected benefit evaluation need to be obtained through an expert simulation model. Essentially, after the corresponding expert gives the result based on their knowledge and judgment ability, sample data is formed, and then the simulation review model 30 is obtained by training with the sample data.

[0082] Specifically, please refer to the appendix Figure 7 , and the method for establishing the simulation review model 30 for each expert includes:

[0083] Step S301): Show the expert the materials 20 of multiple projects under the review project type and the fields included in the reviewed text 31.

[0084] Step S302): Receive the values of the fields filled in by the expert to obtain the reviewed text 31.

[0085] Exemplarily, the expert fills in the project success probability, professional matching degree of project executors, project expected benefit evaluation, etc. for each project according to the materials 20. And samples are generated by subsequent steps.

[0086] Step S303): Use the corresponding abstract extraction model to extract the abstract of the materials 20 to obtain the project summary 21.

[0087] Step S304): Obtain the project information 22 of the project, and obtain samples according to the project summary 21, project information 22, and the reviewed text 31.

[0088] Step S305): Use the pre-connected large language model to learn the samples, and obtain the simulation review model 30 according to the large language model.

[0089] After obtaining the samples, use the pre-connected large language model to learn the samples, so that the large language model can learn the knowledge contained in the samples, and then the effect of simulation review can be achieved, that is, the simulation review model 30 can be obtained. On the one hand, the learned large language model can be directly used as the simulation review model 30. On the other hand, a machine learning model can be further specially trained as the simulation review model 30. Specifically, the method for obtaining the simulation review model 30 according to the large language model includes:

[0090] Build and use the project summary 21 and project information 22 to train an autoencoder model, and obtain a feature extraction model according to the autoencoder model;

[0091] Generate multiple project summaries 21 and project information 22;

[0092] Obtain the review text 31 of the project summary 21 and project information 22 according to the large language model;

[0093] Use the feature extraction model to extract the features of the project summary 21 and project information 22, and associate the features with the review text 31 as an extended sample;

[0094] Build and use the extended sample to train a machine learning model, and obtain a simulation review model 30 according to the trained machine learning model.

[0095] Step S104) Read the project information 22 of the project to be reviewed, and generate a confused summary 23 and confused information 24 according to the project summary 21 and project information 22 of the project to be reviewed. The confused summary 23 and confused information 24 have known predetermined review results. The project summary 21 and project information 22 can be partially modified to obtain the confused summary 23 and confused information 24. Exemplarily, replace the keywords related to the project background introduction in the project summary 21 with some other irrelevant keywords to generate the confused summary 23. At the same time, replace the project executors in the project information 22 with personnel in other professional fields, creating a situation where non-professionals conduct the project, that is, generate the confused information 24. It can be seen that if the simulation review model 30 is appropriate, that is, it can distinguish the project summary 21 and the confused summary 23, and can distinguish the project information 22 and the confused information 24, then the simulation review model 30 can output the correct review result.

[0096] On the contrary, if the technical background of a certain expert does not match the technology of the current project to be reviewed, the corresponding simulation review model 30 of this expert will not be able to correctly distinguish the project summary 21, project information 22, from the confused summary 23, confused information 24, and will output an incorrect review result. Thus, this expert can be excluded from the review of this project to be reviewed.

[0097] It should be noted that even if the expert is in the same field as the project to be reviewed, there are still differences in sub - fields, resulting in the fact that the expert's simulation review model 30 cannot correctly distinguish the project abstract 21, project information 22, from the confused abstract 23 and confused information 24, and will output incorrect review results. Exemplarily, even in the software field, there are sub - fields such as the embedded software field, the operating system software field, the application software field, the web application software field, etc. If the project to be reviewed is in the operating system software field, although it is also in the software field, expert D with an embedded development knowledge structure is not competent to review the project to be reviewed in the operating system software field. The abstract extraction model corresponding to expert D cannot correctly extract the project abstract 21 of the materials 20 in the operating system software field. Specifically, because in step S204), expert D cannot provide a correct reflection score.

[0098] Step S105) inputs the confused abstract 23 and confused information 24 into the simulation review model 30 of each expert to obtain the confused review text 32.

[0099] Exemplarily, the abstract extraction model associated with the "Natural Science Foundation Application Project in the Software Field" of expert C is selected to obtain the project abstract 21 of the materials 20 of the project to be reviewed. Then, after associating with the project information 22, it is input into the simulation review model 30 of each expert, that is, into the simulation review models 30 of experts A, B, C, D, etc.

[0100] Step S106) compares the review text 31 and the confused review text 32 with the review rules of the corresponding review project types respectively to obtain the review results. Exemplarily, the review rules include: Rule 1, the project success probability should be greater than or equal to 60%; Rule 2, the professional matching degree of the project executors should be greater than or equal to 75%; Rule 3, the project expected benefit evaluation should be high.

[0101] Step S107) When the review result is consistent with the predetermined review result, the corresponding expert is included in the candidate library 12.

[0102] When the expert's simulation review model 30 is suitable for the project to be reviewed, it will give a correct review result and can correctly distinguish the review results of the project abstract 21, project information 22, from the confused abstract 23 and confused information 24. The result can be obtained through steps S106) to S107). After all suitable experts are included in the candidate library 12, the next step is carried out.

[0103] Step S108) Randomly select a predetermined number of experts from the candidate library 12 as the review experts 11 for the project to be reviewed. Randomly selecting a predetermined number of experts from the candidate library 12 as the review experts 11 for the project to be reviewed can ensure fairness.

[0104] Specifically, the method of randomly selecting a predetermined number of experts from the candidate library 12 as the review experts 11 for the project to be reviewed includes:

[0105] Randomly select more experts than the predetermined number from the candidate library 12;

[0106] Use the abstract extraction models corresponding to the selected experts to extract project abstracts 21 from the materials 20 of the project to be reviewed respectively;

[0107] Use a clustering algorithm to cluster the project abstracts 21 to obtain a clustering center;

[0108] Sort all experts in ascending order of the distance from the clustering center;

[0109] Select experts as the review experts 11 for the project to be reviewed in sequence until the predetermined number is reached.

[0110] By the materials 20 of the project to be reviewed, the abstract is extracted by the abstract extraction model corresponding to the expert, and further selection is carried out according to the abstract, so that more suitable and appropriate experts can be further selected as the review experts 11 for the project to be reviewed.

[0111] On the other hand, this specification provides a review expert 11 selection and matching management system. Please refer to the appendix Figure 8 , including:

[0112] An extraction module 100 that establishes an abstract extraction model for each expert and associates the review project type;

[0113] An abstract module 200 that reads the materials 20 of the project to be reviewed and extracts the abstract of the materials 20 of the project to be reviewed using the abstract extraction model corresponding to the review project type, denoted as the project abstract 21;

[0114] A review module 300 that establishes a simulation review model 30 for each expert. The input of the simulation review model 30 is the project abstract 21 and the project information 22, and the output is a formatted text containing predetermined fields, denoted as the review text 31;

[0115] A confusion module 400 that reads the project information 22 of the project to be reviewed and generates a confusion abstract 23 and confusion information 24 based on the project abstract 21 and the project information 22 of the project to be reviewed. The confusion abstract 23 and confusion information 24 have known predetermined review results;

[0116] A confusion review module 500 that inputs the confusion abstract 23 and confusion information 24 into the simulation review model 30 of each expert to obtain a confusion review text 32;

[0117] The comparison module 600 compares the reviewed text 31 and the obfuscated reviewed text 32 with the review rules corresponding to the respective review item types to obtain a review result;

[0118] The candidate selection module 700 includes the corresponding experts in the candidate library 12 when the review result matches the predetermined review result;

[0119] The selection module 800 randomly selects a predetermined number of experts from the candidate library 12 as the review experts 11 for the item to be reviewed.

[0120] Please refer to Figure 9 the schematic structural diagram of an electronic device provided by the embodiment of the present specification shown.

[0121] As Figure 9 shown, the electronic device 1100 may include: at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. Among them, the communication bus 1102 can be used to realize the connection and communication of the above components. Among them, the user interface 1103 may include buttons, and the optional user interface may further include a standard wired interface and a wireless interface. Among them, the network interface 1104 may but is not limited to include a Bluetooth module, an NFC module, a Wi-Fi module, etc. Among them, the processor 1101 may include one or more processing cores. The processor 1101 connects various parts within the entire electronic device 1100 through various interfaces and lines, and by running or executing instructions, programs, code sets or instruction sets stored in the memory 1105, and calling data stored in the memory 1105, performs various functions of the routing device 1100 and processes data. Optionally, the processor 1101 may be implemented in at least one of the hardware forms of DSP, FPGA, and PLA. The processor 1101 may integrate one or a combination of several of CPU, GPU, and modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication.

[0122] It can be understood that the above modem may not be integrated into the processor 1101 and may be implemented separately by a chip.

[0123] Among them, the memory 1105 may include RAM or ROM. Optionally, the memory 1105 includes a non-transitory computer-readable medium. The memory 1105 can be used to store instructions, programs, codes, code sets, or instruction sets. The memory 1105 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store the data involved in the above-mentioned method embodiments. Optionally, the memory 1105 may also be at least one storage device located far from the aforementioned processor 1101. The memory 1105, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and application programs. The processor 1101 may be used to call the application programs stored in the memory 1105 and execute the methods in the above-mentioned multiple embodiments.

[0124] The embodiments of this specification also provide a computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are run on a computer or a processor, the computer or the processor is caused to execute multiple steps in the above-mentioned embodiments. If the respective component modules of the above-mentioned electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0125] The embodiments of this specification also provide a computer program product, including a computer program. When the computer program is executed by a processor, multiple steps in the above-mentioned embodiments are implemented.

[0126] Without conflict, the technical features in this embodiment and the implementation solutions can be combined arbitrarily.

[0127] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes a plurality of computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this specification are 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 through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server 40, or data center to another website, computer, server 40, or data center by wire (such as coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server 40 or data center that includes a plurality of available media integrated. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a Digital Versatile Disc (DVD)), or a semiconductor medium (for example, a Solid State Disk (SSD)), etc.

[0128] When implemented by hardware or firmware, the foregoing method flow is programmed into a hardware circuit to obtain a corresponding hardware circuit structure and implement the corresponding function. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit, and its logical function is determined by the user programming the device. A designer can program on their own to "integrate" a digital system on a single PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL), and there is not only one kind of HDL, but many kinds. Those skilled in the art should also be clear that only by slightly logically programming the method flow with the above several hardware description languages and programming it into an integrated circuit, it is easy to obtain a hardware circuit that implements the logical method flow.

[0129] The embodiments described above are only described in the preferred embodiment mode of this specification, and do not limit the scope of this specification. Without departing from the design spirit of this specification, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of this specification shall fall within the protection scope determined by the claims of this specification.

Claims

1. A method for selecting and matching audit experts, characterized in that: Includes steps: Build a summary extraction model for each expert and associate it with the audit project type; Read the information of the project to be reviewed, and use the summary extraction model corresponding to the review project type to extract the summary of the information of the project to be reviewed, and record it as the project summary; Establishing a simulation audit model for each expert, wherein the input of the simulation audit model is the project summary and project information, and the output is a formatted text containing predetermined fields, which is recorded as the audit text; Reading the project information of the project to be reviewed, generating an obfuscated summary and obfuscated information according to the project summary and the project information of the project to be reviewed, wherein the obfuscated summary and the obfuscated information have a known predetermined review result; Inputting the obfuscated summary and obfuscated information into the simulated audit model of each expert to obtain an obfuscated audit text; Compare the audit text and the obfuscated audit text with the audit rules of the corresponding audit item types to obtain the audit results; When the audit result is consistent with the predetermined audit result, the corresponding expert is included in the candidate pool; A predetermined number of experts are randomly selected from the candidate pool as review experts for the items to be reviewed.

2. A method for selecting and matching audit experts according to claim 1, characterized in that: The method for establishing a summary extraction model for each expert and associating the audit project type includes the following steps: Go through each audit item type in turn and perform the following steps: Read the data of multiple projects under the current audit project type and generate multiple project summaries for each data; Receive a mark of reflectivity rating for some of the project summaries; When an expert enters the database, the information of the project is displayed to the expert, and then multiple project summaries of the project are displayed to the expert; receiving a reflectivity score of the project summary provided by the expert; Obtaining the proximity of the experts' reflectivity scores to the project summaries with reflectivity score marks; When the proximity is greater than a preset threshold, the expert is associated with the current audit project type; According to the expert's responsiveness score for the project abstract without a responsiveness score mark, sample data is established, wherein the sample data includes the material, the project abstract and the responsiveness score; A summary extraction model is obtained based on the sample data, and the current audit project type and the expert are associated.

3. A method for selecting and matching audit experts according to claim 2, characterized in that: The method for obtaining a summary extraction model according to the sample data includes: Establishing and using the sample data to train a machine learning model, and obtaining a summary extraction model according to the machine learning model; or, The method for obtaining a summary extraction model according to the sample data includes: Generate a natural language task according to the sample data, input the natural language task into a pre-connected large language model, and obtain a summary extraction model according to the large language model; or, The method for obtaining a summary extraction model according to the sample data includes: Generate a natural language task according to the sample data, and input the natural language task into a pre-connected large language model; A machine learning model is established, and the large language model is used to guide training, and a summary extraction model is obtained according to the trained machine learning model.

4. A method for selecting and matching audit experts according to any one of claims 1 to 3, characterized in that: The method of establishing a simulated audit model for each expert includes: Displaying information of multiple projects under the audit project type and fields included in the audit text to the expert; Receive the value of the field filled in by the expert to obtain the audit text; Extract the material summary using the corresponding summary extraction model to obtain a project summary; Obtaining project information of the project, and obtaining a sample according to the project summary, project information and the review text; The sample is learned using a pre-accessed large language model, and a simulated audit model is obtained based on the large language model.

5. A method for selecting and matching audit experts according to claim 4, characterized in that: According to the large language model, the method for obtaining a simulated audit model includes: Establish and use the project summary and project information to train an autoencoder model, and obtain a feature extraction model based on the autoencoder model; Generate multiple project summaries and project information; Obtaining a review text of the project summary and project information according to the large language model; Extract features of the project summary and project information using the feature extraction model, and associate the features with the review text as an extended sample; Establish and use the extended sample to train a machine learning model, and obtain a simulated audit model based on the trained machine learning model.

6. A method for selecting and matching audit experts according to any one of claims 1 to 3, characterized in that: The method of randomly selecting a predetermined number of experts from the candidate pool as review experts for the project to be reviewed includes: Randomly selecting more than a predetermined number of experts from the pool to be selected; Extract the project summaries from the data of the projects to be reviewed using the corresponding summary extraction models of the selected experts; Clustering the project summaries using a clustering algorithm to obtain cluster centers; sorting all experts in ascending order of distance from the cluster center; Experts are selected in order as review experts for the items to be reviewed until the predetermined number is reached.

7. An audit expert selection and matching management system, characterized in that: include: The extraction module builds a summary extraction model for each expert and associates it with the audit project type; A summary module reads the information of the project to be reviewed, and uses a summary extraction model corresponding to the review project type to extract a summary of the information of the project to be reviewed, which is recorded as a project summary; The audit module establishes a simulated audit model for each expert, wherein the input of the simulated audit model is the project summary and project information, and the output is a formatted text containing predetermined fields, which is recorded as the audit text; an obfuscation module, which reads the project information of the project to be reviewed, and generates an obfuscated summary and obfuscated information according to the project summary and project information of the project to be reviewed, wherein the obfuscated summary and obfuscated information have a known predetermined review result; The obfuscation review module inputs the obfuscation summary and obfuscation information into the simulation review model of each expert to obtain an obfuscation review text; A comparison module compares the audit text and the obfuscated audit text with the audit rules of the corresponding audit item type to obtain the audit result; A selection module, when the audit result matches the predetermined audit result, the corresponding expert is included in the selection pool; The selection module randomly selects a predetermined number of experts from the candidate pool as review experts for the items to be reviewed.

8. An electronic device, characterized in that: including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1 to 6.

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

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images