A method, system, equipment and medium for organizing and summarizing high-tech enterprise RD projects

Abstract

The present invention discloses a method, system, equipment and medium for arranging and summarizing RD projects of high-tech enterprises of enterprises, and relates to the technical field of enterprise R&D project management. The method comprises: obtaining project data of each project to be arranged and summarized, establishing an initial model for judging whether the project data is an RD project, and training the initial model; using an RD project identification model to perform project judgment on each project data; searching and obtaining the knowledge product IP corresponding to the target data, and associating and storing the project name of the target data with the corresponding knowledge product IP; obtaining multiple PS fields of each target data, and obtaining the knowledge product IP corresponding to each PS field, and associating and storing each PS field with a valid knowledge product IP; and solving the problems existing in the existing enterprise RD project arrangement method, such as low efficiency, poor data accuracy, difficulty in information sharing and collaboration, difficulty in effective monitoring and management, and high cost.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise R&D project management, and more specifically, to a method, system, equipment and medium for organizing and summarizing enterprise high-tech R&D projects. Background Art

[0002] In the process of enterprises applying for high-tech enterprises, the compilation and summarization of research and development projects (RD projects) is a core link in the high-tech enterprise certification process, so the compilation and summarization of RD projects is crucial; however, most enterprises currently use manual methods to organize RD projects. This traditional model has many limitations, including low efficiency, poor data accuracy, difficulties in information sharing and collaboration, difficulty in effective monitoring and management, and high costs. These problems have seriously affected the smooth progress of the enterprise's high-tech enterprise certification work and the quality of project management.

[0003] Taking the actual scenario in the application work as an example, when screening RD projects that meet the conditions for high-tech enterprise recognition, it is difficult to quickly and accurately identify qualified projects from a large number of projects based on complex recognition standards through manual methods; when statistically analyzing the components of project costs, due to the limitations of manual accounting, it is difficult to achieve accurate analysis of cost details; when it comes to the intellectual property (IP) generated by the related projects and the product and service (PS) areas to which they belong, manual operations are prone to omissions or incorrect associations, resulting in low efficiency in the application work and the accuracy and completeness of the application materials are difficult to effectively guarantee; this series of problems seriously hinders the smooth progress of the company's high-tech enterprise recognition work, and also has a negative impact on the company's project management level and innovation and development capabilities. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, system, equipment and medium for organizing and summarizing RD projects of high-tech enterprises, which can accurately screen out qualified RD projects, realize the rapid association of projects with intellectual property rights and related fields, and improve the efficiency and accuracy of organizing and summarizing RD projects in high-tech enterprise applications. It has the advantages of high efficiency, accuracy, convenience, real-time monitoring and management, and low cost, and solves the problems existing in the existing enterprise RD project organization methods such as low efficiency, poor data accuracy, difficulty in information sharing and collaboration, difficulty in effective monitoring and management, and high cost.

[0005] The above technical objectives of the present invention are achieved through the following technical solutions:

[0006] This application provides a method for organizing and summarizing enterprise high-tech R&D projects, including the following specific steps:

[0007] Obtain project data for each project to be collated and summarized, and establish the eligibility criteria for RD projects;

[0008] Based on the conditions, an initial model is established to determine whether the project data is an RD project, and the initial model is trained. The initial model that meets the training end conditions is determined as the RD project identification model;

[0009] Use the RD project identification model to determine whether each project data meets the RD project requirements, and determine the project data that meets the requirements as the target data;

[0010] For each target data, find and obtain the knowledge product IP corresponding to the target data, and associate and store the project name of the target data with the corresponding knowledge product IP;

[0011] Obtain multiple PS fields of each target data, and obtain the knowledge product IP corresponding to each PS field, determine whether the corresponding knowledge product IP is valid, and associate and store each PS field with the valid knowledge product IP.

[0012] The beneficial effects of the present invention are as follows: First, in this solution, project data for each project is obtained and eligibility criteria for RD projects are established. Second, based on the eligibility criteria, an initial model is established to determine whether the project data is an RD project. The initial model is trained until the initial model meets the training end criteria, thereby obtaining an RD project identification model. The RD project identification model is then used to filter out data that meets RD projects from the project data and identify the RD project data as target data. Finally, for the filtered RD projects that meet the criteria, the intellectual property rights generated by them are associated in the RD project basic table, establishing a correspondence between the project and the intellectual property rights. At the same time, the RD projects that meet the criteria are associated with the fields to which the PSs belong, clarifying the technical fields to which the projects belong. Furthermore, information such as RD project-related data, processed data, and associated relationships can be stored for easy subsequent query and management. Furthermore, the RD project organization results can be displayed visually, including project lists, cost structures, IP associations, PS field distribution, etc., for user viewing and analysis.

[0013] In this plan, RD projects are sorted and summarized in this way, which improves the accuracy and efficiency of RD project screening, avoids omissions and errors in manual screening, reduces the human, material and financial resources required for manual sorting of RD projects, reduces the company's operating costs and management costs, improves work efficiency, and shortens the company's high-tech enterprise certification cycle; it also achieves a rapid connection between projects and intellectual property rights and their respective fields, providing comprehensive and accurate data support for high-tech enterprise applications; at the same time, it optimizes the company's R&D project management process and improves the company's overall R&D management level.

[0014] On the basis of the above technical solution, the present invention can also be improved as follows.

[0015] Furthermore, the above-mentioned conditions are met:

[0016] Project data also includes labor costs, material costs, and equipment costs.

[0017] The beneficial effect of adopting the above further scheme is: traverse all R&D projects, and for each project, check whether it incurs any expenses during the recognition period; if expenses are incurred, further check whether the cost structure includes the three elements of labor costs, material costs, and equipment costs. If so, the project will be judged to meet the requirements of the RD project.

[0018] Furthermore, the output of the above initial model is:

[0019] ;

[0020] in:

[0021] ;

[0022] Where, Indicates the existence of The probability value of the cost, , and correspond to labor costs, material costs and equipment costs respectively; Indicates an inactive The probability value of the cost, Indicates the probability value that one of the charges is not activated; represents the classification weight matrix, represents the feature vector, Represents the bias vector.

[0023] The beneficial effect of adopting the above further scheme is: the input text is recognized and judged through the established initial model, and the final output is the probability value of the existence of the three elements. When the probability value corresponding to each element exceeds the threshold, it is determined that the three elements exist in the text data.

[0024] Furthermore, the above training termination condition is: the value of the loss function of the initial model meets the preset condition; wherein, the loss function of the initial model is:

[0025] ;

[0026] Where, represents the value of the loss function, , and correspond to labor costs, material costs and equipment costs respectively, Indicates the existence of the initial model output The probability value of the cost, Indicates the existence of The true value of this cost.

[0027] The beneficial effect of adopting the above further solution is: judging whether the initial model is trained through the constructed loss function.

[0028] Furthermore, the above method also includes: introducing an attention mechanism into the initial model to reduce the impact of changes in word segmentation position information in the sentence on the semantics of the sentence.

[0029] The beneficial effects of adopting the above further scheme are: since the different order of words will lead to different meanings of the whole sentence, the introduced attention mechanism enables the model to more accurately capture the positional relationship between words when processing the input sequence, thereby avoiding the problem of semantic understanding errors caused by changes in text order; when processing languages ​​with complex grammatical and semantic structures such as Chinese, the calculation method based on position information embedding can enhance the model's understanding of text semantics; and by introducing position encoding information in the introduced attention mechanism, the model's dynamic focusing ability and semantic understanding ability when processing input sequences can be improved.

[0030] Furthermore, the processing results of the attention mechanism introduced above are specifically as follows:

[0031] ;

[0032] Where, Indicates the index sequence of the input item data. i The result of the attention mechanism processing of each element, is the attention weight, indicating the i The element pair j The degree of attention paid to each element The index sequence of the input item data j The vector representation of elements, The linear transformation matrix of the vector, Represents the position encoding vector to introduce the j The element relative to the i The position information of the elements, The length of the index sequence of the input item data.

[0033] Furthermore, the above i The element pair j The degree of attention of each element is:

[0034] ;

[0035] in:

[0036] ;

[0037] Where, Indicates thei The element pair j The degree of attention paid to each element is the attention score, indicating the i The element pair j The correlation of the elements, Indicates the i The element pair k The correlation of the elements, The length of the index sequence representing the input item data, Represents the linear transformation result of the query vector, which is used to match the key vector. The linear transformation result of the key vector plus the position encoding vector is used to perform dot product operation with the query vector. Dimension representing the scaling factor.

[0038] In a second aspect, the present application provides a system for organizing and summarizing RD projects of enterprises and high-tech enterprises, which is applied to a method for organizing and summarizing RD projects of enterprises and high-tech enterprises in any one of the first aspects, including:

[0039] The data acquisition module is used to obtain the project data of each project to be sorted and summarized, and to establish the eligibility conditions of the RD project;

[0040] The model training module is used to establish an initial model for determining whether the project data is an RD project based on the conditions met, and train the initial model, and determine the initial model that meets the training end conditions as the RD project identification model;

[0041] The project judgment module is used to use the RD project identification model to judge whether each project data meets the RD project requirements and determine the project data that meets the requirements as the target data;

[0042] The IP association module is used to search and obtain the intellectual product IP corresponding to each target data, and associate and store the project name of the target data with the corresponding intellectual product IP;

[0043] The PS association module is used to obtain multiple PS fields of each target data, and obtain the knowledge product IP corresponding to each PS field, determine whether the corresponding knowledge product IP is valid, and associate and store each PS field with the valid knowledge product IP.

[0044] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the methods in the first aspect when executing the computer program.

[0045] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0046] Compared with the prior art, the present invention has at least the following beneficial effects:

[0047] In this application, first, the project data of each project is obtained, and the eligibility conditions of the RD project are established; secondly, an initial model for determining whether the project data is an RD project is established based on the eligibility conditions, and the initial model is trained until the initial model meets the training end conditions, thereby obtaining an RD project identification model, and then using the RD project identification model to filter out data that meets the RD project from each project data, and determine the data of the RD project as the target data; finally, for the filtered RD projects, the intellectual product IP generated by them is associated in the RD project basic table to establish a correspondence between the project and the intellectual property rights. At the same time, the RD projects that meet the requirements will be associated with the fields to which PS belongs, and the technical field to which the projects belong will be clarified; of course, the RD project-related data, processed data, and association relationship information can also be stored for subsequent query and management. At the same time, the RD project sorting results can be displayed in a visual manner, including project lists, cost structures, IP association status, PS field distribution, etc., for user viewing and analysis. It can accurately screen out RD projects that meet the requirements, realize the rapid association of projects with intellectual property rights and their fields, and improve the efficiency and accuracy of RD project sorting and summarization in high-tech enterprise applications. It has the advantages of high efficiency, accuracy, convenience, real-time monitoring and management, and low cost, and solves the problems of low efficiency, poor data accuracy, difficulty in information sharing and collaboration, difficulty in effective monitoring and management, and high cost in the existing enterprise RD project sorting methods.

[0048] In this application, RD projects are organized and summarized in this way, which improves the accuracy and efficiency of RD project screening, avoids omissions and errors in manual screening, reduces the human, material and financial resources required for manual organization of RD projects, reduces the company's operating costs and management costs, improves work efficiency, and shortens the company's high-tech enterprise certification cycle; it also achieves a rapid connection between projects and intellectual property rights and their respective fields, providing comprehensive and accurate data support for high-tech enterprise applications; at the same time, it optimizes the company's R&D project management process and improves the company's overall R&D management level.

[0049] In this application, since the different order of words will lead to different meanings of the entire sentence, the introduced attention mechanism enables the model to more accurately capture the positional relationship between words when processing the input sequence, thereby avoiding the problem of semantic understanding errors caused by changes in text order; when processing languages ​​with complex grammatical and semantic structures such as Chinese, the calculation method based on position information embedding can enhance the model's ability to understand the semantics of the text; and by introducing position encoding information in the introduced attention mechanism, the model's dynamic focusing ability and semantic understanding ability when processing input sequences can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0051] Figure 1 A flowchart of a method for arranging and summarizing a method according to an embodiment of the present invention;

[0052] Figure 2 A connection diagram of the system for arranging and summarizing the embodiments of the present invention is provided;

[0053] Figure 3 Schematic diagram of the connection of electronic equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0055] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0056] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0057] In the description of the embodiments of the present invention, "a plurality of" means at least two.

[0058] Example 1: When screening RD projects that meet the high-tech enterprise recognition conditions, it is difficult to quickly and accurately identify qualified projects from a large number of projects based on complex recognition standards; when statistically analyzing the cost components of projects, due to the limitations of manual accounting, it is difficult to achieve accurate analysis of cost details. Therefore, this embodiment provides a method for sorting and summarizing high-tech enterprise RD projects, such as Figure 1 As shown, the following specific steps are included:

[0059] S1, obtain the project data of each project to be sorted and summarized, and establish the eligibility conditions of the RD project.

[0060] Specifically, R&D projects for the certification year are screened, identifying projects with expenses incurred during the certification period. Furthermore, projects with expenses comprised of labor, materials, and equipment costs are screened to determine if these projects meet the RD project criteria. The specific steps are as follows: First, determine the company's certification year plan. Based on this plan, select R&D projects from the R&D project database that incurred expenses in the past three years of the certification period. Further verification is performed to determine if the expense composition includes labor, materials, and equipment costs. If so, the project is determined to meet the RD project criteria and added to the list of projects meeting the RD project criteria.

[0061] Optionally, the above conditions are:

[0062] Project data also includes labor costs, material costs, and equipment costs.

[0063] S2: Based on the conditions, an initial model is established to determine whether the project data is an RD project, and the initial model is trained, and the initial model that meets the training end conditions is determined as the RD project recognition model.

[0064] Specifically, the initial model can include an input layer, an embedding layer, a BiL STM layer, a pooling layer, a fully connected layer, and an output layer; wherein the input layer receives a text sequence with a maximum length of L, and each word is converted into an index sequence , text digitization is achieved through vocabulary mapping; the embedding layer maps discrete word indexes to d-dimensional continuous vectors; the BiL STM layer uses bidirectional LSTM to capture contextual semantic features, the pooling layer extracts sequence global features, and the fully connected layer generates the existence probability of elements, which are finally output through the output layer.

[0065] Furthermore, in the above initial model, the bidirectional LSTM effectively captures the contextual features of the previous and next sentences, the maximum pooling layer retains significant features and suppresses noise interference; the multi-label classification structure realizes the joint recognition of the three elements, and then optimizes the synergy between feature extraction and classification through end-to-end training; among them, the model can be used with the PyTorch / TensorFlow framework, Kai uses the Adam optimizer, with an initial learning rate set to 3e-4, a batch size of 64, and a training cycle of 20-30.

[0066] The above training ends when the loss function of the initial model satisfies the preset conditions. The loss function of the initial model is:

[0067] ;

[0068] Where, represents the value of the loss function, , and correspond to labor costs, material costs and equipment costs respectively, Indicates the existence of the initial model output The probability value of the cost, Indicates the existence of The true value of this cost.

[0069] Optionally, the above method also includes: introducing an attention mechanism into the initial model to reduce the impact of changes in word segmentation position information in the sentence on the semantics of the sentence.

[0070] Among them, since the different order of words will lead to different meanings of the whole sentence, the introduced attention mechanism enables the model to more accurately capture the positional relationship between words when processing the input sequence, thereby avoiding the problem of semantic understanding errors caused by changes in text order; when processing languages ​​with complex grammatical and semantic structures such as Chinese, the calculation method based on position information embedding can enhance the model's understanding of text semantics; and by introducing position encoding information in the introduced attention mechanism, the model's dynamic focusing ability and semantic understanding ability when processing input sequences can be improved.

[0071] Optionally, the processing result of the attention mechanism introduced above is specifically as follows:

[0072] ;

[0073] Where, Indicates the index sequence of the input item data. i The result of the attention mechanism processing of each element, is the attention weight, indicating the i The element pair j The degree of attention paid to each element The index sequence of the input item data j The vector representation of elements, The linear transformation matrix of the vector, Represents the position encoding vector to introduce the j The element relative to the i The position information of the elements, The length of the index sequence of the input item data.

[0074] Furthermore, the above i The element pair j The degree of attention of each element is:

[0075] ;

[0076] in:

[0077] ;

[0078] Where, Indicates the i The element pair j The degree of attention paid to each element is the attention score, indicating the i The element pair j The correlation of the elements, Indicates the i The element pair k The correlation of the elements, The length of the index sequence representing the input item data, Represents the linear transformation result of the query vector, which is used to match the key vector. The linear transformation result of the key vector plus the position encoding vector is used to perform dot product operation with the query vector. Dimension representing the scaling factor.

[0079] S3, using the RD project identification model to determine whether each project data meets the RD project, and determining the project data that meets the conditions as the target data.

[0080] Optionally, the output of the above initial model is:

[0081] ;

[0082] in:

[0083] ;

[0084] Where, Indicates the existence of The probability value of the cost, , and correspond to labor costs, material costs and equipment costs respectively; Indicates an inactive The probability value of the cost, Indicates the probability value that one of the charges is not activated; represents the classification weight matrix, represents the feature vector, Represents the bias vector.

[0085] Among them, the input text is recognized and judged through the established initial model, and the final output is the probability value of the existence of three elements. When the probability values ​​corresponding to each element exceed the threshold, it is determined that the three elements exist in the text data; specifically, the threshold of the probability value can be set to 0.5, and if it exceeds 0.5, it is determined that the corresponding elements exist in the text.

[0086] S4: For each target data, search and obtain the knowledge product IP corresponding to the target data, and associate and store the project name of the target data with the corresponding knowledge product IP.

[0087] The IP rights for the annual recognition plan are screened. By accessing the IP database acquired during the recognition period, the selected IP results that meet the requirements of the RD project are evaluated and screened. The IP that meets the RD project requirements is then transferred to the RD project-related IP database. For R&D projects that meet the RD project requirements, the IP rights generated by the intellectual property are analyzed and a corresponding relationship between RDs and IP is established. For example, one RD can generate one or more IPs, or multiple RDs can jointly generate one IP.

[0088] S5, obtain multiple PS fields of each target data, and obtain the knowledge product IP corresponding to each PS field, determine whether the corresponding knowledge product IP is valid, and associate each PS field with the valid knowledge product IP for storage.

[0089] Among them, the selected RD projects are associated with the intellectual property IP they generate in the RD project basic table to establish a corresponding relationship between the project and the intellectual property. At the same time, the RD projects that meet the requirements are associated with the fields to which the PS belongs to clarify the technical fields to which the projects belong.

[0090] Specifically, RD project-related data, processed data, and associated relationships can also be stored for easy subsequent query and management. At the same time, the RD project organization results can be displayed in a visual manner, including project lists, cost structures, IP associations, PS field distribution, etc., for user viewing and analysis. It can accurately screen out eligible RD projects, quickly associate projects with intellectual property rights and their respective fields, and improve the efficiency and accuracy of RD project organization and summarization in high-tech enterprise applications. It has the advantages of high efficiency, accuracy, convenience, real-time monitoring and management, and low cost, and solves the problems of low efficiency, poor data accuracy, difficulty in information sharing and collaboration, difficulty in effective monitoring and management, and high cost in existing enterprise RD project organization methods.

[0091] In this embodiment, RD projects are sorted and summarized in this way, which improves the accuracy and efficiency of RD project screening, avoids omissions and errors in manual screening, reduces the human, material and financial resources required for manual sorting of RD projects, reduces the company's operating costs and management costs, improves work efficiency, and shortens the company's high-tech enterprise certification cycle; it also achieves a rapid association between projects and intellectual property rights and their respective fields, providing comprehensive and accurate data support for high-tech enterprise applications; at the same time, it optimizes the company's R&D project management process and improves the company's overall R&D management level.

[0092] Example 2: This embodiment of the application provides a system for summarizing and organizing high-tech enterprise RD projects, which is applied to a method for summarizing high-tech enterprise RD projects in Example 1, such as Figure 2 Shown, including:

[0093] The data acquisition module is used to obtain the project data of each project to be sorted and summarized, and to establish the eligibility conditions of the RD project;

[0094] The model training module is used to establish an initial model for determining whether the project data is an RD project based on the conditions met, and train the initial model, and determine the initial model that meets the training end conditions as the RD project identification model;

[0095] The project judgment module is used to use the RD project identification model to judge whether each project data meets the RD project requirements and determine the project data that meets the requirements as the target data;

[0096] The IP association module is used to search and obtain the intellectual product IP corresponding to each target data, and associate and store the project name of the target data with the corresponding intellectual product IP;

[0097] The PS association module is used to obtain multiple PS fields of each target data, and obtain the knowledge product IP corresponding to each PS field, determine whether the corresponding knowledge product IP is valid, and associate and store each PS field with the valid knowledge product IP.

[0098] Specifically, the operation process of the above-mentioned sorting and summarizing system can be:

[0099] System initialization: Install and configure the enterprise high-tech RD project organization and summary system, set system parameters and user permissions, connect to relevant internal enterprise data sources such as financial systems, project management systems, etc., to ensure that data can be obtained normally.

[0100] Data collection: According to preset time intervals or trigger conditions, RD project-related data is automatically collected from the company's internal data sources, including project name, project leader, project start and end time, labor costs, material costs, equipment costs and other cost details, as well as information such as the IP name, type, application time of the knowledge product, and product or service name, field, revenue and other information, and the collected data is stored.

[0101] Project screening: Read the collected RD project data from the stored data, and screen the projects according to the preset screening conditions, such as the expenses incurred during the recognition period and including the three elements of labor costs, material costs, and equipment costs; mark the projects that meet the conditions as projects that meet the RD project standards, and store the screening results.

[0102] Information association: For projects that meet the RD project standards, the corresponding intellectual property (IP) data is further read, and the project and the intellectual property generated by it are associated through preset matching rules (such as project name, technical field, time range, etc.); at the same time, the system determines the technical field to which each product or service (PS) belongs based on the classification rules of the PS, and associates the PS with the corresponding intellectual property; after the association is completed, the system stores the correspondence between the RD project, intellectual property and PS field to facilitate subsequent query and analysis.

[0103] Example 3: This embodiment of the present application provides an electronic device, such as Figure 3 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method of embodiment 1 is implemented.

[0104] Example 4: The present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute the method of Example 1.

[0105] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0106] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0107] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0109] Those skilled in the art will understand that all or part of the steps in implementing the above facts and methods can be completed by instructing relevant hardware through a program, and the program involved or the program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: the corresponding method steps are then brought out, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0110] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for summarizing and organizing high-tech enterprise RD projects, characterized by: The specific steps include: Obtain project data for each project to be collated and summarized, and establish eligibility criteria for RD projects, where RD projects are research and development projects; Based on the conditions, an initial model for determining whether the project data is an RD project is established, and the initial model is trained, and the initial model that meets the training end conditions is determined as the RD project identification model; Using the RD project identification model to determine whether each of the project data meets the RD project requirements, and determining the project data that meets the requirements as target data; For each target data, search and obtain the intellectual property rights corresponding to the target data, and associate and store the project name of the target data with the corresponding intellectual property rights, wherein the intellectual property rights are intellectual property rights generated by the associated project; Obtain multiple PS fields of each target data, and obtain the intellectual product IP corresponding to each PS field, determine whether the corresponding intellectual product IP is valid, and associate and store each PS field with the valid intellectual product IP, wherein the PS field is the product service field to which it belongs; The qualifying conditions are: The project data also includes labor costs, material costs and equipment costs; The output of the initial model is: in: Where, Indicates the existence of The probability value of the cost, , and correspond to labor costs, material costs and equipment costs respectively; Indicates an inactive The probability value of the cost, Indicates the probability value that one of the charges is not activated; represents the classification weight matrix, represents the feature vector, Represents the bias vector.

2. The method for arranging and summarizing high-tech enterprise RD projects according to claim 1 is characterized in that: The training end condition is: the value of the loss function of the initial model meets the preset condition; wherein the loss function of the initial model is: Where, represents the value of the loss function, , and correspond to labor costs, material costs and equipment costs respectively, Indicates the existence of the initial model output The probability value of the cost, Indicates the existence of The true value of this cost.

3. The method for arranging and summarizing high-tech enterprise RD projects according to claim 1 is characterized in that: The method also includes: introducing an attention mechanism into the initial model to reduce the impact of changes in word segmentation position information in a sentence on the semantics of the sentence.

4. The method for arranging and summarizing high-tech enterprise RD projects according to claim 3 is characterized in that: The processing results of the introduced attention mechanism are as follows: Where, Indicates the index sequence of the input item data. i The result of the attention mechanism processing of each element, is the attention weight, indicating the i The element pair j The degree of attention paid to each element, The index sequence of the input project data j The vector representation of elements, The linear transformation matrix of the vector, Represents the position encoding vector to introduce the j The element relative to the i The position information of the elements, The length of the index sequence of the input item data.

5. The method for arranging and summarizing high-tech enterprise RD projects according to claim 4 is characterized in that: The said i The element pair j The degree of attention of each element is: in: Where, Indicates the i The element pair j The degree of attention paid to each element, is the attention score, indicating the i The element pair j The correlation of the elements, Indicates the i The element pair k The correlation of the elements, The length of the index sequence representing the input item data, Represents the linear transformation result of the query vector, which is used to match the key vector. The linear transformation result of the key vector plus the position encoding vector is used to perform dot product operation with the query vector. Dimension representing the scaling factor.

6. A system for summarizing and organizing RD projects of high-tech enterprises, applied to a method for summarizing and organizing RD projects of high-tech enterprises according to any one of claims 1 to 5, characterized in that: include: The data acquisition module is used to obtain the project data of each project to be sorted and summarized, and to establish the eligibility conditions of the RD project; A model training module is used to establish an initial model for determining whether the project data is an RD project based on the conditions, and train the initial model, and determine the initial model that meets the training end conditions as the RD project identification model; A project judgment module, configured to use the RD project identification model to judge whether each of the project data meets the RD project requirements, and determine the project data that meets the requirements as target data; An IP association module is used to search and obtain the intellectual product IP corresponding to each target data, and associate and store the project name of the target data with the corresponding intellectual product IP; The PS association module is used to obtain multiple PS fields of each target data, and obtain the knowledge product IP corresponding to each PS field, determine whether the corresponding knowledge product IP is valid, and associate and store each PS field with the valid knowledge product IP.

7. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method according to any one of claims 1 to 5 is implemented when the processor executes the computer program.

8. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method according to any one of claims 1 to 5.

Images