Implementation Method and System for Displaying Recommended Content on the Dashboard of a Science and Technology Management System

Abstract

The present invention relates to the technical field of data recommendation, and discloses an implementation method and system for displaying recommended content on a science and technology management system dashboard, including: dividing data for each dashboard in the science and technology management system to obtain a second subcategory corresponding to the secondary directory of each dashboard; setting the performance behavior information of each employee position, where the performance behavior information includes the work content executed by the employee and the execution time period and execution area corresponding to the work content; constructing a mapping table; extracting the feature vectors of the mapping table to obtain a feature vector set corresponding to the mapping table; inputting the feature vector set into a deep learning neural network for training to obtain a content recommendation main model; and sending the content recommendation main model to each terminal based on an edge computing method. The present invention improves the work efficiency of employees.

Description

Technical Field

[0001] The present invention relates to the technical field of data recommendation, and specifically relates to a method and system for implementing the display of recommended content on a dashboard of a science and technology management system. Background Art

[0002] In order to improve the efficiency and quality of daily science and technology management work, many enterprises have designed science and technology management systems. Through automated and intelligent means, the science and technology management systems improve the efficiency and quality of daily science and technology management work, realize the full-process management of scientific research projects, including declaration, review, progress tracking, fund management, etc., thereby improving the overall level of science and technology innovation management; through standardized process management, ensure the smooth progress of science and technology innovation projects.

[0003] When sorting out some declaration materials, enterprise employees need to log in to the enterprise's science and technology management system to view the data of each dashboard. The content displayed on the home page of the dashboard is the same. If the content displayed on the home page of the dashboard is not the data that the enterprise employee wants, the enterprise employee still needs to perform corresponding searches in the directory column of the dashboard. For newly recruited employees, if the content displayed on the home page of the dashboard is not what the new employee wants and they need to search in the directory column, they may not be able to find the desired data because they are not familiar with the directory column. Summary of the Invention

[0004] The present invention provides a method, system and medium for implementing the display of recommended content on a dashboard of a science and technology management system, aiming at the situation where employees of each position obtain some data through the science and technology management system at different time periods, and the science and technology management system can recommend and display on the home page of each dashboard the data that employees of this position need to see during the time period when they log in to the science and technology management system.

[0005] The present invention is implemented through the following technical solutions:

[0006] A method for implementing the display of recommended content on a dashboard of a science and technology management system includes:

[0007] Dividing the data of each dashboard in the science and technology management system to obtain a second subcategory corresponding to the second-level directory of each dashboard;

[0008] Setting the performance behavior information of each employee position, where the performance behavior information includes the work content performed by the employee, the execution time period and execution area corresponding to this work content, and the work content includes finding the required data from the second-level directory;

[0009] Constructing a mapping relationship between the position, work content and the corresponding execution time period, execution area and second subcategory, thereby forming a mapping table regarding the position, work content and execution time period, execution area and second subcategory;

[0010] Extract the feature vectors of the mapping table to obtain the set of feature vectors corresponding to the mapping table;

[0011] Input the set of feature vectors into a deep learning neural network for training to obtain a content recommendation main model that recommends the second sub-classification path and the corresponding link to the second-level directory name on the home page of each dashboard according to the login time of the technology management system and the employee position;

[0012] Based on the edge computing method, send the content recommendation main model to each terminal, and execute the content recommendation main model on the logged-in terminal after an employee logs in to the technology management system through any one of the terminals, so as to realize the display of recommended content on the dashboard home page.

[0013] In this technical solution, since a group company consists of multiple positions, and the responsibilities of each position are different, the data to be obtained in the technology management system is different due to different work. New employees often don't know where to obtain the required data on the dashboard of the technology management system because they are not familiar with the system. Therefore, the present invention designs a method for realizing the display of recommended content on the dashboard. The content recommendation main model is trained through the employee's position, work content, the execution time period corresponding to the work content, the execution area, and the path of the second sub-classification number. Through this content recommendation main model, when an employee of a certain position logs in to the technology management system and opens a certain dashboard, the path of the second sub-classification number of the data that the employee may use and the second-level directory name corresponding to the second sub-classification are displayed in the display area of the recommended content on the dashboard, and the employee can directly click. If the link displayed in the display area of the recommended content is not the data link that the employee wants, the employee can choose not to click the link in the display area of the recommended content.

[0014] As an optimization, the specific process of dividing the data of each dashboard in the technology management system to obtain the second sub-classification corresponding to the second-level directory of each dashboard is as follows:

[0015] Perform the first division on each dashboard in the technology management system to obtain the main classification corresponding to each dashboard type;

[0016] Respectively divide each first-level directory included in each dashboard to obtain the first sub-classification corresponding to each first-level directory;

[0017] Divide each second-level directory included in each first-level directory to obtain the second sub-classification corresponding to each second-level directory.

[0018] As an optimization, the mapping relationship is expressed as: , where Z represents the position, N represents the number corresponding to the work content, , respectively represent the start date and end date corresponding to the described work content, represents the path of the second sub-classification number. A represents the main classification number corresponding to the kanban, B represents the first sub-classification number corresponding to the first-level directory, C represents the second sub-classification number corresponding to the second-level directory, and D represents the number of the execution area.

[0019] As an optimization, based on the directory of the kanban, the first sub-classification number and / or the second sub-classification number are encoded as serial numbers increasing from small to large from front to back.

[0020] As an optimization, based on the directory of the kanban, the first sub-classification number and / or the second sub-classification number are encoded as serial numbers decreasing from large to small from front to back.

[0021] As an optimization, the specific process of inputting the set of feature vectors into the deep neural network for training is as follows:

[0022] Take the feature vectors of the position, execution time period, and execution area as input features and input them into the deep learning neural network. At the same time, take the feature vector of the path of the second sub-classification number as the output feature to train the deep learning neural network.

[0023] As an optimization, the specific method for training the deep learning neural network is to take the maximum sum of the similarities between the feature vector of the actual path of the second sub-classification number in the set of feature vectors and the predicted feature vector of the path of the second sub-classification number output by the deep learning neural network as the objective function. Consider a set of combined values of the parameters of the deep learning neural network as the position of a black-winged kite individual, and train the parameters of the deep learning neural network by combining the improved black-winged kite algorithm with the set of feature vectors:

[0024] The specific process of training the parameters of the deep learning neural network by combining the improved black-winged kite algorithm with the set of feature vectors is as follows:

[0025] S1. Initialize the algorithm parameters of the improved black-winged kite algorithm. The algorithm parameters include the number of the black-winged kite population, the maximum number of iterations, and the optimization range;

[0026] S2. Randomly initialize the black-winged kite population;

[0027] S3. Input the set of feature vectors into the deep learning neural networks corresponding to the black-winged kite population respectively, and calculate the fitness value generated by each position of the black-winged kite individual; among them, the fitness value is the same as the objective function;

[0028] S4. Reorder the black-winged kite individuals according to the fitness value;

[0029] S5. Perform attack behavior and improved migration behavior on the re - sorted black - winged kite individuals, thereby updating the positions of the black - winged kite individuals;

[0030] S6. Determine whether the maximum number of iterations is reached. If so, end and output the current black - winged kite population and the parameters of the deep - learning neural network corresponding to the black - winged kite individual with the largest fitness value in the black - winged kite population. Otherwise, return to S3.

[0031] As an optimization, the specific formula for randomly initializing the black - winged kite population is:

[0032] ;

[0033] Where, represents the position of the \(i\) - th black - winged kite individual, \(i\) is an integer between 1 and pop, and pop is the number of black - winged kite individuals in the black - winged kite population, and are the lower and upper bounds of the \(j\) - th dimension of the \(i\) - th black - winged kite individual respectively, that is, the optimization range, and rand is a value randomly selected from [0, 1]. Among them, the dimension of the black - winged kite individual is the same as the number of parameters for training the deep - learning neural network;

[0034] The mathematical model of the attack behavior of the black - winged kite individual is:

[0035]

[0036] ;

[0037] Where, represents the position of the \(i\) - th black - winged kite individual in the \(j\) - th dimension and the \((t + 1)\) - th iteration step; represents the position of the \(i\) - th black - winged kite individual in the \(j\) - th dimension and the \(t\) - th iteration step; \(r\) is a random number between 0 and 1, \(p\) is a constant of 0.9; \(T\) is the maximum number of iterations, and \(t\) is the number of iterations completed so far;

[0038] The mathematical model of the improved migration behavior of the black - winged kite individual is:

[0039] ;

[0040] ;

[0041] Where, represents the position of the \(i\) - th black - winged kite individual in the \(j\) - th dimension and the \(t\) - th iteration step; represents the The position of only the black-winged kite individual in the peacekeeping t iteration step; Indicates the leading scorer of the d -th dimension of the black-winged kite individual in the t -th iteration so far, that is, the optimal black-winged kite individual with the largest fitness value, which represents the current position of the i -th dimension of any black-winged kite individual in the t -th iteration; Indicates the fitness value of the random position of the d -th dimension of any black-winged kite individual in the t -th iteration;

[0042] The specific expression of the Cauchy mutation is:

[0043] ;

[0044] where x represents a random variable.

[0045] After the deep learning neural network model determines the optimal weights and thresholds, other parameters of the deep learning neural network model are fine-tuned through training samples, such as the learning rate, batch size, weight decay, learning rate decay, etc.

[0046] As an optimization, when an employee logs in to the technology management system, the execution rules of the content recommendation main model include:

[0047] Obtain the employee's login account, login area, and login date, and match the employee's position according to the login account;

[0048] Match the execution area according to the login area;

[0049] Judge whether there is a matching execution time period according to the login date. If so, extract the execution time period matching the login date, the execution area matching the login area, and the feature vectors of the employee's position as input features and input them into the content recommendation main model. If not, extract the next execution time period closest to the login date, and use the next execution time period and the feature vectors of the employee's position as input features and input them into the content recommendation main model.

[0050] As an optimization, the implementation method further includes:

[0051] Map each piece of work content in the execution area to the corresponding relevant file publishing website, and use web crawler technology to obtain the data published on the relevant file publishing website, so as to update the execution time period of the work content;

[0052] Update the mapping table according to the new execution time period, so as to update the feature vector set, and update the content recommendation main model through the updated feature vector set.

[0053] The present invention also discloses an implementation system for displaying recommended content on a science and technology management system dashboard, which is used to execute the foregoing implementation method for displaying recommended content on a science and technology management system dashboard, including:

[0054] A division module, which is used to divide the data of each dashboard in the science and technology management system to obtain a second sub-classification corresponding to the secondary directory of each dashboard;

[0055] A setting module, which is used to set the performance behavior information of each employee position, and the performance behavior information includes the work content executed by the employee and the corresponding execution time period and execution area of the work content, and the work content includes finding the required data from the secondary directory;

[0056] A mapping table construction module, which is used to construct the mapping relationship between the position, work content and the corresponding execution time period, execution area and second sub-classification, so as to form a mapping table about the position, work content and execution time period, execution area and second sub-classification;

[0057] A feature extraction module, which is used to extract the feature vectors of the mapping table to obtain the corresponding feature vector set of the mapping table;

[0058] A training module, which is used to input the feature vector set into a deep learning neural network for training, so as to obtain a content recommendation main model for recommending the second sub-classification link displayed on the home page of each dashboard according to the login time of the science and technology management system and the employee position;

[0059] A sending module, which is used to send the content recommendation main model to each terminal based on edge computing, and execute the content recommendation main model on the logged-in terminal after the employee logs in to the science and technology management system through any one of the terminals, so as to realize the display of recommended content on the home page of the dashboard.

[0060] The present invention also discloses a storage medium storing a computer program, and when the computer program is executed by a processor, it implements the foregoing implementation method for displaying recommended content on a science and technology management system dashboard.

[0061] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0062] 1. The present invention designs an implementation method for displaying recommended content on a kanban. By using the employee's position, job content, corresponding execution time period for the job content, execution area, and the path of the second sub-classification number, a main content recommendation model is trained. Through this main content recommendation model, when an employee of a certain position logs in to the technology management system and opens a certain kanban, the path of the second sub-classification number of the data that the employee may use and the name of the second-level directory corresponding to the second sub-classification are displayed in the display area of the recommended content on the kanban. If the recommended second-level directory name is what the employee wants, the employee can directly click on it, which facilitates the employee to directly obtain the desired data and improves the work efficiency of the employee.

[0063] 2. The main content recommendation model is sent to each terminal through edge computing. The employee directly performs content recommendation through the main content recommendation model on the terminal, reducing the computing data of the main content recommendation model and improving the computing efficiency of the main content recommendation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of this application, and do not limit the embodiments of the present invention. In the drawings:

[0065] Figure 1 is a flowchart of an implementation method for displaying recommended content on a kanban of a technology management system according to the present invention;

[0066] Figure 2 is a schematic diagram of the interface composition of a kanban in a technology management system. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and do not limit the present invention.

[0068] Embodiment 1 of the present invention provides an implementation method for displaying recommended content on a kanban of a technology management system, as Figure 1 shown, including:

[0069] S1. Divide the data of each kanban in the technology management system to obtain the second sub-classification corresponding to the second-level directory of each kanban.

[0070] In the present invention, the technology management system, as the support for the company's daily technology management work, includes but is not limited to the following functions, as shown in Table 1:

[0071] Table 1

[0072]

[0073] On the home page of the technology management system, there are multiple kanban module links. Clicking on one of the kanban module links will take you to the corresponding kanban, where you can search for data.

[0074] In the visualization interface of a kanban home page, it includes the directories corresponding to all the links of that kanban and the display area for recommended content. The directories are displayed in a hierarchical form. For example, Figure 2 As shown, it is a schematic diagram of the high-tech enterprise module kanban. Figure 2 In it, the left area is the display area for directories, and the right area is the display area for recommended content.

[0075] Since some employees enter the technology management system to obtain some data for filling out the declaration forms of some project applications, and to obtain this data, the usual method is to click on the directory bar in the left area and then enter the corresponding directory link. However, for new employees, they do not know which directory links in the technology management system the data they need is located in. Therefore, it is very necessary to display the directory links of the materials that the new employee wants to view predicted in the display area of the recommended content.

[0076] And the data that needs to be found can usually be found in the secondary directories in the technology management system. Therefore, in the present invention, it is only necessary to obtain the classification of the secondary directories of each kanban. If some kanbans only have primary directories, the secondary directories under that primary directory level can be set to 0. Of course, the secondary directories can also be set to be empty.

[0077] In some embodiments, the specific process of dividing the data of each kanban in the technology management system to obtain the second sub-classification corresponding to the secondary directory of each kanban is as follows:

[0078] Perform an initial division on each kanban in the technology management system to obtain the main classification corresponding to each kanban type.

[0079] Respectively divide the primary directories included in each said kanban to obtain the first sub-classification corresponding to each obtained primary directory.

[0080] Divide the secondary directories included in each said primary directory to obtain the second sub-classification corresponding to each said secondary directory.

[0081] S2. Set the performance behavior information of each employee's position. The performance behavior information includes the work content performed by the employee, the corresponding execution time period and execution area of this work content. The work content includes searching for the required data from the secondary directories.

[0082] For example, in the application for high-tech enterprise projects, if an employee's position is a project application specialist, the work content is to fill out the high-tech enterprise application form for the region where the employee's enterprise is located. Since the policy plans in each region are different, resulting in differences in the application time, the execution period of this work content is the application time of the high-tech enterprise application policy, and the execution area is the city where the enterprise is located.

[0083] S3. Construct the mapping relationship between the position, work content and the corresponding execution period, execution area and second sub-category, so as to form a mapping table of the position, work content, execution period, execution area and second sub-category.

[0084] The mapping table can be updated according to specific circumstances. The update process is an existing process and will not be elaborated here.

[0085] In some embodiments, the mapping relationship is expressed as: , where Z represents the position, N represents the number corresponding to the work content, 、 respectively represent the start date and end date corresponding to the work content, represents the path of the second sub-category number, A represents the main category number corresponding to the kanban, B represents the first sub-category number corresponding to the first-level directory, and C represents the second sub-category number corresponding to the second-level directory.

[0086] The numbers of the position, work content, start date and end date, main category, first sub-category, and second sub-category can be compiled according to the company's own internal rules.

[0087] In some embodiments, the first sub-category number and / or the second sub-category number are encoded in ascending order from front to back.

[0088] In some embodiments, based on the directory of the kanban, the first sub-category number and / or the second sub-category number are encoded in descending order from front to back.

[0089] If the first sub-category number and the second sub-category number are encoded in ascending order, then if there are the first sub-category number and the second sub-category number, the smallest encoding serial number is 1. When there is no second sub-category, the second sub-category serial number can be set to 0, or it can also be set to empty.

[0090] If the first sub-category number and the second sub-category number are encoded in descending order, then a maximum value can be set, and this maximum value should be greater than the number of any second sub-category in all kanbans.

[0091] S4. Extract the feature vectors of the mapping table to obtain the feature vector set corresponding to the mapping table.

[0092] In some embodiments, a position may involve multiple job contents. Therefore, for a single position, there may be multiple mapping relationships, that is, for a single position, there may be multiple feature vectors. All the feature vectors corresponding to all positions are combined together to form a feature vector set.

[0093] S5. Input the feature vector set into a deep learning neural network for training, so as to obtain a content recommendation main model that can recommend the second sub-classification path and the link to the corresponding second-level directory name on the home page of each dashboard according to the login time of the technology management system and the employee's position.

[0094] In some embodiments, the specific process of inputting the feature vector set into a deep neural network for training is as follows:

[0095] Use the feature vectors of the position, execution time period, and execution area as input features and input them into the deep learning neural network. At the same time, use the feature vector of the path of the second sub-classification number as the output feature to train the deep learning neural network.

[0096] During specific training, the feature vectors of the position, execution time period, and execution area are used as input features and input into the deep learning neural network. Then, the deep learning neural network outputs the predicted path of the second sub-classification number. The predicted second sub-classification number and the actual path of the second sub-classification number in the mapping table are used by a loss function to determine whether the deep learning neural network meets the requirements.

[0097] Here, the path of the second sub-classification number includes the main classification number corresponding to the dashboard where the second sub-classification is located and the first sub-classification number corresponding to the first-level directory, that is, the content described above .

[0098] When the content recommendation main model outputs the path of the second sub-classification number, the corresponding second-level directory name of the second sub-classification can also be obtained. Finally, what is displayed in the display area of the recommended content is the path of the second sub-classification number and the second-level directory name.

[0099] For example, in Figure 2 the display area of the recommended content, the displayed content can be: 1 / 1 / 2 RD Table Query.

[0100] In some embodiments, the specific method for training the deep learning neural network is to maximize the sum of similarities between the feature vectors of the paths with the actual second sub-classification numbers in the set of feature vectors and the feature vectors of the paths with the predicted second sub-classification numbers output by the deep learning neural network as the objective function. Regarding a set of combined values of the various parameters of the deep learning neural network as the position where a black-winged kite individual is located, the various parameters of the deep learning neural network are trained by combining the improved black-winged kite algorithm with the set of feature vectors:

[0101] The specific process of training the various parameters of the deep learning neural network by combining the improved black-winged kite algorithm with the set of feature vectors is as follows:

[0102] S1. Initialize the algorithm parameters of the improved black-winged kite algorithm. The algorithm parameters include the number of the black-winged kite population, the maximum number of iterations, and the optimization range;

[0103] S2. Randomly initialize the black-winged kite population;

[0104] S3. Input the set of feature vectors into the deep learning neural networks corresponding to the black-winged kite population respectively, and calculate the fitness value generated by the position of each black-winged kite individual; among them, the fitness value is the same as the objective function;

[0105] S4. Reorder the black-winged kite individuals according to the fitness value;

[0106] S5. Perform attack behaviors and improved migration behaviors on the reordered black-winged kite individuals, thereby updating the positions of the black-winged kite individuals;

[0107] S6. Determine whether the maximum number of iterations is reached. If so, end and output the current black-winged kite population, and the various parameters of the deep learning neural network corresponding to the black-winged kite individual with the maximum fitness value in the black-winged kite population; otherwise, return to S3.

[0108] In some embodiments, the specific formula for randomly initializing the black-winged kite population is:

[0109] ;

[0110] where, represents the position of the i-th black-winged kite individual, is an integer between 1 and pop, and pop is the number of black-winged kite individuals in the black-winged kite population, and They are the lower and upper bounds of the black-winged kite individual in the j-th dimension, i.e., the optimization range. rand is a value randomly selected between [0, 1]. Among them, the dimension of the black-winged kite individual is the same as the number of parameters for training the deep learning neural network;

[0111] The mathematical model of the attack behavior of the black-winged kite individual is:

[0112] ;

[0113] ;

[0114] Among them, represents the position of the i-th black-winged kite individual in the j-th dimension and at the (t + 1)-th iteration step; represents the position of the i-th black-winged kite individual in the j-th dimension and at the t-th iteration step; r is a random number between 0 and 1, and p is a constant of 0.9; T is the maximum number of iterations, and t is the number of iterations completed so far;

[0115] The mathematical model of the improved migration behavior of the black-winged kite individual is:

[0116] ;

[0117] ;

[0118] Among them, represents the position of the -th black-winged kite individual in the -th dimension and at the -th iteration step; represents the position of the -th black-winged kite individual in the -th dimension and at the t -th iteration step; represents the leading scorer of the -th iteration in the -th dimension of the black-winged kite individual, that is, the optimal black-winged kite individual with the largest fitness value, represents the current position of any black-winged kite individual in the -th iteration in the -th dimension; represents the fitness value of any black-winged kite individual at a random position in the -th dimension in the -th iteration; represents Cauchy mutation; represents the ranking of the i-th black-winged kite individual in the j-th dimension and at the t-th iteration step;

[0119] The specific expression of Cauchy mutation is:

[0120] ;

[0121] Among them, x represents a random variable.

[0122] It should be noted that the parameters of the deep learning neural network trained by the improved black-winged kite algorithm include weights and thresholds. After the deep learning neural network determines the optimal weights and thresholds, other parameters of the deep learning neural network model are fine-tuned through training samples, such as learning rate, batch size, weight decay, learning rate decay, etc.

[0123] Adding the ranking of black-winged kite individuals to the migration behavior makes full use of the current position ranking information, which is beneficial for the position of black-winged kite individuals to jump out of the local optimum, enabling black-winged kite individuals to perform global search and optimization within a given area range, and improving the optimization ability of the black-winged kite algorithm.

[0124] In some embodiments, when an employee logs in to the science and technology management system, the execution rules of the content recommendation main model include:

[0125] A1. Obtain the login account, login area, and login date of the employee, and match the position of the employee according to the login account;

[0126] A2. Determine whether there is a matching execution time period according to the login date. If so, extract the execution time period matching the login date and the feature vector of the employee's position as input features and input them into the content recommendation main model. If not, extract the next execution time period closest to the login date, and use the next execution time period and the feature vector of the employee's position as input features and input them into the content recommendation main model.

[0127] For example, when an employee logs in to the science and technology management system, the content recommendation main model will predict the login purpose of the employee based on the login account, login area, and login date, that is, judge the policy requirements closest to the location of the company where the employee is located according to the login area, and check whether there is a project application form that needs to be filled in when viewing the time of logging in to the system. If so, present all the paths of the second sub-classification numbers corresponding to the data for filling in the project application form and the names of the second-level directories corresponding to the second sub-classification in Figure 2 the display area of the recommended content. The employee can directly click on the paths of the second sub-classification numbers and the names of the corresponding second-level directories in the display area of the recommended content to find the relevant data for filling in the project application form.

[0128] It should be noted that the link of the path of the second sub-classification number and the corresponding second-level directory name is one link. That is, the main body displayed in the display area of the recommended content is the path of the second sub-classification number + the name of the second-level directory.

[0129] If the time of logging in to the system is not within the time period of any project declaration, it is determined that the employee fills in the project declaration form in advance, and the path of the second sub-classification number corresponding to the data required by the project declaration policy that has been announced and is about to start, which is closest to the login time, is displayed in the display area of the recommended content.

[0130] For example, the time of logging in to the system is February 1st of a certain year. However, at this time, there is no project to be declared. But it has been announced that Project A will be declared from February 15th to April 30th, and the declaration time of Project A is the project policy declaration time closest to the login time (February 1st). Then the content recommendation main model will automatically consider that the employee is filling in the declaration form of Project A in advance. Therefore, in the display area of the recommended content on the dashboard related to Project A, the path of the second sub-classification number of the data required for the declaration form of Project A will be displayed.

[0131] S6. Based on the edge computing method, the content recommendation main model is sent to each terminal, and after the employee logs in to the science and technology management system through any one of the terminals, the content recommendation main model on the logged-in terminal is executed, so as to realize the display of the recommended content on the home page of the dashboard.

[0132] The content recommendation main model trained in the cloud is sent to each terminal through the edge computing method. In this way, the employee uses the content recommendation main model of the terminal to display the recommended content, avoiding the situation that multiple employees use the content recommendation main model at the same time, which may cause the operation data of the model to be too large and reduce the operation speed.

[0133] In some embodiments, the implementation method further includes:

[0134] S7. Map and connect each work content with the corresponding relevant file release website, and use web crawler technology to obtain the data released by the relevant file release website, so as to update the execution time period of the work content;

[0135] Update the mapping table according to the new execution time period, so as to update the feature vector set, and update the content recommendation main model through the updated feature vector set.

[0136] The content recommendation main model can be automatically updated, and the updated new content recommendation main model is sent to each terminal again, so that the content recommendation main models of each terminal are synchronized.

[0137] If there are new employee positions in the group company, the content recommendation main model can also be trained in the cloud and then distributed to each terminal.

[0138] Embodiment 2 discloses an implementation system for displaying recommended content on a science and technology management system dashboard, which is used to execute the implementation method for displaying recommended content on a science and technology management system dashboard described in Embodiment 1, including:

[0139] A partitioning module, which is used to partition the data of each dashboard in the science and technology management system to obtain a second sub-classification corresponding to the secondary directory of each dashboard;

[0140] A setting module, which is used to set the performance behavior information of each employee position. The performance behavior information includes the work content executed by the employee and the execution time period and execution area corresponding to the work content. The work content includes searching for required data from the secondary directory;

[0141] A mapping table construction module, which is used to construct the mapping relationship between the position, work content and the corresponding execution time period, execution area and second sub-classification, so as to form a mapping table about the position, work content and execution time period, execution area and second sub-classification;

[0142] A feature extraction module, which is used to extract the feature vectors of the mapping table to obtain a set of feature vectors corresponding to the mapping table;

[0143] A training module, which is used to input the set of feature vectors into a deep learning neural network for training, so as to obtain a content recommendation main model for displaying recommended second sub-classification links on the home page of each dashboard according to the login time of the science and technology management system and the employee position;

[0144] A sending module, which is used to send the content recommendation main model to each terminal based on edge computing, and execute the content recommendation main model on the logged-in terminal after the employee logs in to the science and technology management system through any one of the terminals, so as to realize the display of recommended content on the home page of the dashboard.

[0145] Embodiment 3 discloses a storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the implementation method for displaying recommended content on a science and technology management system dashboard described in Embodiment 1.

[0146] The specific embodiments described above further elaborate on the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for displaying recommended content on a technology management system dashboard, characterized in that: include: Divide the data of each kanban board in the technology management system to obtain the second sub-category corresponding to the secondary directory of each kanban board; Setting the job performance information of each employee position, wherein the job performance information includes the work content performed by the employee and the execution time period and execution area corresponding to the work content, wherein the work content includes searching for required data from the secondary directory; Constructing a mapping relationship between the position, work content and the corresponding execution time period, execution area and second sub-classification, thereby forming a mapping table of the position, work content and the execution time period, execution area and second sub-classification; Extracting the feature vectors of the mapping table to obtain a feature vector set corresponding to the mapping table; The feature vector set is input into a deep learning neural network for training, thereby obtaining a content recommendation main model by displaying the recommended second sub-category path and the corresponding secondary directory name link on the homepage of each dashboard according to the login time and employee position of the technology management system; Based on the edge computing method, the content recommendation main model is sent to each terminal, and after the employee logs in to the technology management system through any of the terminals, the content recommendation main model on the login terminal is executed, thereby realizing the display of recommended content on the homepage of the dashboard; When an employee logs into the technology management system, the execution rules of the content recommendation main model include: Obtaining the employee's login account, login region, and login date, and matching the employee's position according to the login account; Matching an execution area according to the login area; Determine whether there is a matching execution time period based on the login date. If so, extract the execution time period matching the login date, the execution area matching the login area, and the feature vector of the employee's position as input features and input them into the main content recommendation model. If not, extract the next execution time period closest to the login date, and use the feature vector of the next execution time period and the employee's position as input features and input them into the main content recommendation model.

2. According to claim 1, a method for implementing the display of recommended content on a technology management system dashboard is characterized in that: The specific process of dividing the data of each kanban board in the technology management system to obtain the second sub-category corresponding to the secondary directory of each kanban board is as follows: Perform the first division on each kanban board in the technology management system to obtain the main classification corresponding to each kanban board type; Dividing the first-level directories contained in each of the kanban boards respectively to obtain first subcategories corresponding to each first-level directory; The second-level directories contained in each of the first-level directories are divided to obtain second subcategories corresponding to each of the second-level directories.

3. The method for implementing the display of recommended content on a technology management system dashboard according to claim 1, characterized in that: The mapping relationship is expressed as: ZNT start ~T end -DA / B / C, where Z represents the position, N represents the number corresponding to the job content, and T start , T end They respectively represent the start date and end date corresponding to the work content, A / B / C represent the path of the second sub-classification number, A represents the main classification number corresponding to the dashboard, B represents the first sub-classification number corresponding to the first-level directory, C represents the second sub-classification number corresponding to the second-level directory, and D represents the number of the execution area.

4. The method for implementing the display of recommended content on a technology management system dashboard according to claim 3, characterized in that: Based on the catalog of the billboard, the first sub-category number and / or the second sub-category number are encoded as serial numbers from small to large from the front to the back, or based on the catalog of the billboard, the first sub-category number and / or the second sub-category number are encoded as serial numbers from large to small from the front to the back.

5. The method for implementing the display of recommended content on a technology management system dashboard according to claim 1, characterized in that: The specific process of inputting the feature vector set into the deep neural network for training is: The feature vectors of the position, execution time period and execution area are input into the deep learning neural network as input features, and the feature vector of the path of the second sub-classification number is used as an output feature to train the deep learning neural network.

6. A method for implementing the display of recommended content on a technology management system dashboard according to claim 5, characterized in that: The specific method for training the deep learning neural network is to maximize the sum of similarities between the feature vectors of the actual second sub-classification number path in the feature vector set and the feature vectors of the predicted second sub-classification number path output by the deep learning neural network as the objective function, regard one set of combined values ​​of the parameters of the deep learning neural network as the location of a black kite individual, and train the parameters of the deep learning neural network by combining the feature vector set with the improved black kite algorithm: The specific process of training the parameters of the deep learning neural network by combining the improved black kite algorithm with the feature vector set is as follows: S1, initializing algorithm parameters of the improved black-winged kite algorithm, wherein the algorithm parameters include the number of black-winged kite populations, the maximum number of iterations, and the optimization range; S2, randomly initializing the black-winged kite population; S3, respectively inputting the feature vector set into the deep learning neural network corresponding to the black-winged kite population, and calculating the fitness value generated at the location of each black-winged kite individual; wherein the fitness value is the same as the objective function; S4, reordering the black-winged kite individuals according to the fitness values; S5, performing attack behaviors and improved migration behaviors on the reordered black-winged kite individuals, thereby updating the positions of the black-winged kite individuals; S6. Determine whether the maximum number of iterations has been reached. If so, terminate the process and output the current black-winged kite population and the parameters of the deep learning neural network corresponding to the black-winged kite individual with the largest fitness value in the black-winged kite population. Otherwise, return to S3.

7. A method for implementing the display of recommended content on a technology management system dashboard according to claim 6, characterized in that: The specific formula for randomly initializing the black-winged kite population is: X i =BK lb +rand(UK ub -BK lb ); Among them, X i Indicates the position of the i-th black-winged kite individual, i is an integer between 1 and pop, pop is the number of black-winged kites in the black-winged kite population, BK lb and BK ub are the lower bound and upper bound of the j-th dimension black-winged kite individual, i.e., the optimization range, and rand is a randomly selected value between [0,1], wherein the dimension of the black-winged kite individual is the same as the number of parameters for training the deep learning neural network; The mathematical model of the attack behavior of the black-winged kite individual is: in, represents the position of the i-th black-winged kite individual in the j-th dimension and the t+1-th iteration step; represents the position of the i-th black-winged kite individual in the j-th dimension and the t-th iteration step; r is a random number between 0 and 1, p is a constant of 0.9; T is the maximum number of iterations, and t is the number of iterations completed so far; The mathematical model of the improved migration behavior of the black-winged kite individual is: m = 2sin(r + π / 2); in, represents the position of the i-th black-winged kite individual in the j-th dimension and the t+1-th iteration step; represents the position of the i-th black-winged kite individual in the j-th dimension and the t-th iteration step; F represents the leading scorer of the j-th dimension black-winged kite individual in the t-th iteration so far, that is, the optimal black-winged kite individual with the largest fitness value, i represents the j-th dimension current position of any black-winged kite individual in the t-th iteration; F ri represents the fitness value of any black-winged kite individual at a random position in the jth dimension in the tth iteration; C(0,1) represents the Cauchy mutation; represents the ranking of the i-th black-winged kite individual in the j-th dimension and the t-th iteration step; The specific expression of Cauchy mutation is: Here, x represents a random variable.

8. The method for implementing the display of recommended content on a technology management system dashboard according to claim 1, characterized in that: The implementation method also includes: Mapping and connecting each of the work contents in the execution area with the corresponding related file publishing website, and obtaining the publishing data of the related file publishing website through crawler technology, so as to update the execution time period of the work contents; The mapping table is updated according to the new execution time period, thereby updating the feature vector set, and the content recommendation main model is updated through the updated feature vector set.

9. A system for implementing the display of recommended content on a technology management system dashboard, used to execute the method for implementing the display of recommended content on a technology management system dashboard as claimed in any one of claims 1 to 8, characterized in that: include: A partitioning module is used to partition the data of each kanban board in the technology management system to obtain the second sub-category corresponding to the secondary directory of each kanban board; A setting module, used to set the job performance information of each employee position, wherein the job performance information includes the work content performed by the employee and the execution time period and execution area corresponding to the work content, wherein the work content includes searching for required data from the secondary directory; A mapping table construction module, used to construct a mapping relationship between the position, work content and the corresponding execution time period, execution area and second sub-classification, so as to form a mapping table about the position, work content and the execution time period, execution area and second sub-classification; A feature extraction module, used to extract feature vectors of the mapping table to obtain a set of feature vectors corresponding to the mapping table; A training module, used for inputting the feature vector set into a deep learning neural network for training, so as to obtain a content recommendation main model by displaying the recommended second sub-category link on the homepage of each dashboard according to the login time and employee position of the technology management system; The sending module is used to send the content recommendation main model to each terminal based on edge computing, and execute the content recommendation main model on the login terminal after the employee logs in to the technology management system through any of the terminals, so as to realize the recommended content display on the homepage of the dashboard.

Images