Workflow management method and device based on width learning, electronic equipment and storage medium

Abstract

The application provides a workflow management method and device based on width learning, electronic equipment and storage medium. The method comprises the following steps: acquiring process data corresponding to a workflow and generating a test process set and a training feature set according to the process data; generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set; comparing a test process in the test process set with a training process in the training process set and generating a deviation rate set according to the test process and the training process; determining whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set; if not, training the width learning model according to the deviation rate set and regenerating the training process set according to the width learning model and the training feature set; and if yes, generating a process branch of the workflow according to the training process set and a work system code. The process branch meeting the user's expectation is automatically generated through width learning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of process management, in particular to a workflow management method and device based on width learning, electronic equipment and storage medium. BACKGROUND

[0002] The existing server configuration verification system, the workflow management technical solution generally completes the management of the workflow according to the set steps, and other workflow management systems generally schedule the workflow task priority in the same system. When encountering cross-system workflow, especially when the existing workflow does not meet the demand and a new process branch needs to be established, the original system code logic needs to be manually modified or the system needs to be redeveloped to meet the demand, which greatly wastes time, especially when cross-system workflow management is involved, and system debugging is also required. The existing server configuration verification system, the workflow management covers pre-sale demand, research and development, testing, production, after-sales and other nodes, whether in a single system or in multiple system interactions, basically according to the set personnel node to reverse the process. SUMMARY

[0003] Therefore, it is necessary to provide a workflow management method and device based on width learning, which can automatically generate a process branch meeting the user's expectation.

[0004] In a first aspect, a workflow management method based on width learning is provided, characterized in that the method comprises:

[0005] Obtaining process data corresponding to a workflow and generating a test process set and a training feature set according to the process data;

[0006] Generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set;

[0007] Comparing a test process in the test process set with a training process in the training process set and generating a deviation rate set according to the test process and the training process;

[0008] Determining whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set;

[0009] If not, training the width learning model according to the deviation rate set and regenerating a training process set according to the width learning model and the training feature set;

[0010] If yes, generating a process branch of the workflow according to the training process set and a work system code.

[0011] In one of the embodiments, the acquiring the process data corresponding to the workflow and generating a test process set and a training feature set according to the process data comprises:

[0012] The process data comprises process node information and process type information;

[0013] The test process set is generated according to the process node information and the process type.

[0014] The static feature data is extracted from the process node information, and the training feature set is generated according to the static feature data and the process type.

[0015] In one of the embodiments, the generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set comprises:

[0016] A plurality of data sets are generated according to the training feature set;

[0017] A weight matrix is generated according to the width learning model and a bias parameter input by a user;

[0018] Initial process node data is generated according to the plurality of data sets and the weight matrix, and the training process set is generated by integrating the initial process node data and the data sets.

[0019] In one of the embodiments, the comparing a test process in the test process set with a training process in the training process set and generating a bias rate set according to the test process and the training process comprises:

[0020] An enhanced node is determined according to the training process and the test process, and inactivation data is generated according to the enhanced node and the training process set;

[0021] A node bias rate is calculated according to the inactivation data and the bias parameter, and the bias rate set is generated according to the node bias rate, wherein the node bias rate comprises a node quantity bias rate, a node sequence bias rate, a node personnel post bias rate and a node personnel role bias rate.

[0022] In one of the embodiments, the determining whether the process bias rate corresponding to the workflow is less than a bias rate threshold set by a user according to the bias rate set comprises:

[0023] The node bias rate is weighted according to the bias parameter, and the process bias rate corresponding to the process type is calculated according to the weighted node bias rate.

[0024] In one of the embodiments, the generating the process branch of the workflow according to the training process set and the working system code comprises:

[0025] The process branch comprises system backend code, table structure and front-end page code;

[0026] The system backend code and the table structure are generated according to the training process and the working system code;

[0027] The front-end page code is generated according to the process data type of the training process.

[0028] In one of the embodiments, the generating the process branch of the workflow according to the training process set and the working system code comprises:

[0029] The process branch is sent to the user;

[0030] In response to the user determining that the process branch meets the requirement, the node deviation rate corresponding to the process branch is all modified to 0 and the process branch is published;

[0031] In response to the user determining that the process branch does not meet the requirement, the node deviation rate corresponding to the process branch is all modified to 1 and the node deviation rate is returned to the width learning model.

[0032] In another aspect, a width learning-based workflow management device is provided, characterized in that the device comprises:

[0033] A first set generating module is configured to acquire process data corresponding to a workflow and generate a test process set and a training feature set according to the process data;

[0034] A second set generating module is configured to generate a width learning model according to the training feature set and generate a training process set according to the width learning model and the training feature set;

[0035] A third set generating module is configured to compare a test process in the test process set with a training process in the training process set and generate a deviation rate set according to the test process and the training process;

[0036] A determining module is configured to determine whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set;

[0037] A training module is configured to, if not, train the width learning model according to the process deviation rate and regenerate a training process set according to the width learning model and the training feature set;

[0038] The flow branch generation module is configured to generate a flow branch of the workflow according to the set of training flows and the work system code if yes.

[0039] In one of the embodiments, the first set generation module is configured to obtain flow data corresponding to the workflow and generate a set of test flows and a set of training features according to the flow data, including:

[0040] The flow data includes flow node information and flow type information.

[0041] The set of test flows is generated according to the flow node information and the flow type.

[0042] Static feature data is extracted from the flow node information, and the set of training features is generated according to the static feature data and the flow type.

[0043] In one of the embodiments, the second set generation module is configured to generate a width learning model according to the set of training features and generate a set of training flows according to the width learning model and the set of training features, including:

[0044] A plurality of data sets are generated according to the set of training features.

[0045] A weight matrix is generated according to the width learning model and a bias parameter input by a user.

[0046] Initial flow node data is generated according to the plurality of data sets and the weight matrix, and the set of training flows is generated by integrating the initial flow node data and the data sets.

[0047] In one of the embodiments, the third set generation module is configured to compare a test flow in the set of test flows with a training flow in the set of training flows and generate a set of bias rates according to the test flow and the training flow, including:

[0048] An enhanced node is determined according to the training flow and the test flow, and inactivation data is generated according to the enhanced node and the set of training flows.

[0049] A node bias rate is calculated according to the inactivation data and the bias parameter, and the set of bias rates is generated according to the node bias rate, wherein the node bias rate includes a node quantity bias rate, a node sequence bias rate, a node post bias rate, and a node role bias rate.

[0050] In one of the embodiments, the determination module is configured to determine whether a flow bias rate corresponding to the workflow is less than a bias rate threshold set by a user according to the set of bias rates, including:

[0051] The node deviation rate is weighted according to the deviation parameter, and a process deviation rate corresponding to the process type is calculated according to the weighted node deviation rate.

[0052] In one of the embodiments, the process branch generation module generates the process branch of the work flow according to the set of training processes and work system code, which includes:

[0053] The process branch includes system backend code, table structure and front-end page code;

[0054] The system backend code and the table structure are generated according to the training process and the work system code;

[0055] The front-end page code is generated according to the process data type of the training process.

[0056] In one of the embodiments, the process branch generation module generates the process branch of the work flow according to the set of training processes and work system code, which includes:

[0057] The process branch is sent to the user;

[0058] In response to the user determining that the process branch meets the requirements, the node deviation rate corresponding to the process branch is modified to 0, and the process branch is published;

[0059] In response to the user determining that the process branch does not meet the requirements, the node deviation rate corresponding to the process branch is modified to 1, and the node deviation rate is returned to the width learning model.

[0060] In another aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the computer program:

[0061] Obtaining process data corresponding to a work flow and generating a set of test processes and a set of training features according to the process data;

[0062] Generating a width learning model according to the set of training features, and generating a set of training processes according to the width learning model and the set of training features;

[0063] Comparing a test process in the set of test processes with a training process in the set of training processes, and generating a set of deviation rates according to the test process and the training process;

[0064] Determining whether a process deviation rate corresponding to the work flow is less than a deviation rate threshold set by a user according to the set of deviation rates;

[0065] If not, the width learning model is trained according to the deviation rate set and a training procedure set is regenerated according to the width learning model and the training feature set;

[0066] If yes, a procedure branch of the workflow is generated according to the training procedure set and a work system code.

[0067] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0068] The procedure data corresponding to the workflow is acquired and a test procedure set and a training feature set are generated according to the procedure data, which includes:

[0069] The procedure data includes procedure node information and procedure type information;

[0070] The test procedure set is generated according to the procedure node information and the procedure type;

[0071] The static feature data is extracted from the procedure node information and the training feature set is generated according to the static feature data and the procedure type.

[0072] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0073] The width learning model is generated according to the training feature set and a training procedure set is generated according to the width learning model and the training feature set, which includes:

[0074] A plurality of data sets are generated according to the training feature set;

[0075] A weight matrix is generated according to the width learning model and a deviation parameter input by a user;

[0076] Initial procedure node data is generated according to the plurality of data sets and the weight matrix and the training procedure set is generated by integrating the initial procedure node data and the data sets.

[0077] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0078] The test procedure in the test procedure set and the training procedure in the training procedure set are compared and a deviation rate set is generated according to the test procedure and the training procedure, which includes:

[0079] An enhanced node is determined according to the training procedure and the test procedure and inactivation data is generated according to the enhanced node and the training procedure set;

[0080] According to the inactivation data and the bias parameter, a node bias rate is calculated, and the node bias rate is used to generate the bias rate set, wherein the node bias rate includes a node quantity bias rate, a node sequence bias rate, a node post bias rate, and a node role bias rate.

[0081] In one of the embodiments, the processor, when executing the computer program, implements the following steps:

[0082] The determination of whether the process bias rate corresponding to the workflow is less than a bias rate threshold set by a user according to the bias rate set includes:

[0083] The node bias rate is weighted according to the bias parameter, and the process bias rate corresponding to the process type is calculated according to the weighted node bias rate.

[0084] In one of the embodiments, the processor, when executing the computer program, implements the following steps:

[0085] The generation of the process branch of the workflow according to the training process set and the working system code includes:

[0086] The process branch includes system backend code, table structure, and front-end page code.

[0087] The system backend code and the table structure are generated according to the training process and the working system code.

[0088] The front-end page code is generated according to the process data type of the training process.

[0089] In one of the embodiments, the processor, when executing the computer program, implements the following steps:

[0090] The generation of the process branch of the workflow according to the training process set and the working system code includes:

[0091] The process branch is sent to the user.

[0092] In response to the determination of the user that the process branch meets the requirement, the node bias rates corresponding to the process branch are all modified to 0, and the process branch is published.

[0093] In response to the determination of the user that the process branch does not meet the requirement, the node bias rates corresponding to the process branch are all modified to 1, and the node bias rates are returned to the width learning model.

[0094] In another aspect, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium, and the computer program, when executed by a processor, implements the following steps:

[0095] acquiring process data corresponding to the workflow and generating a test process set and a training feature set according to the process data;

[0096] generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set;

[0097] comparing a test process in the test process set with a training process in the training process set and generating a deviation rate set according to the test process and the training process;

[0098] determining whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set;

[0099] if not, training the width learning model according to the deviation rate set and regenerating a training process set according to the width learning model and the training feature set;

[0100] if yes, generating a process branch of the workflow according to the training process set and a work system code.

[0101] In one of the embodiments, the computer program is executed by a processor to implement the following steps:

[0102] The acquiring process data corresponding to the workflow and generating a test process set and a training feature set according to the process data comprises:

[0103] The process data comprises process node information and process type information;

[0104] The test process set is generated according to the process node information and the process type;

[0105] Static feature data is extracted from the process node information and the training feature set is generated according to the static feature data and the process type.

[0106] In one of the embodiments, the computer program is executed by a processor to implement the following steps:

[0107] The generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set comprises:

[0108] A plurality of data sets are generated according to the training feature set;

[0109] A weight matrix is generated according to the width learning model and a deviation parameter input by a user;

[0110] generating initial process node data according to the plurality of data sets and the weight matrix and integrating the initial process node data and the data sets to generate the training process set.

[0111] In one embodiment, the computer program, when executed by the processor, implements the following steps:

[0112] The comparing the test process in the test process set and the training process in the training process set and generating a deviation rate set according to the test process and the training process comprises:

[0113] According to the training process and the test process, determine an enhanced node and generate inactivation data according to the enhanced node and the training process set;

[0114] According to the inactivation data and the deviation parameter, calculate a node deviation rate and generate the deviation rate set according to the node deviation rate, wherein the node deviation rate comprises a node number deviation rate, a node sequence deviation rate, a node personnel post deviation rate and a node personnel role deviation rate.

[0115] In one embodiment, the computer program, when executed by the processor, implements the following steps:

[0116] The determining whether the process deviation rate corresponding to the workflow is less than the deviation rate threshold set by the user according to the deviation rate set comprises:

[0117] According to the deviation parameter, the node deviation rate is weighted and the process type corresponding to the process deviation rate is calculated according to the weighted node deviation rate.

[0118] In one embodiment, the computer program, when executed by the processor, implements the following steps:

[0119] The generating the process branch of the workflow according to the training process set and the working system code comprises:

[0120] The process branch comprises system backend code, table structure and front-end page code;

[0121] According to the training process and the working system code, the system backend code and the table structure are generated;

[0122] According to the process data type of the training process, the front-end page code is generated.

[0123] In one embodiment, the computer program, when executed by the processor, implements the following steps:

[0124] The generating the process branch of the workflow according to the training process set and the working system code comprises:

[0125] sending the procedure branch to the user;

[0126] in response to the user determining that the procedure branch meets the requirement, modifying the node deviation rate corresponding to the procedure branch to 0 and publishing the procedure branch;

[0127] in response to the user determining that the procedure branch does not meet the requirement, modifying the node deviation rate corresponding to the procedure branch to 1 and returning the node deviation rate to the width learning model.

[0128] The above-mentioned workflow management method, device, electronic equipment and storage medium based on width learning, by obtaining the procedure data corresponding to the workflow and generating the test procedure set and the training feature set according to the procedure data; then generating the width learning model according to the training feature set and generating the training procedure set according to the width learning model and the training feature set; then comparing the test procedure in the test procedure set with the training procedure in the training procedure set and generating the deviation rate set according to the test procedure and the training procedure; then determining whether the procedure deviation rate corresponding to the workflow is less than the deviation rate threshold set by the user according to the deviation rate set; if not, training the width learning model according to the deviation rate set and regenerating the training procedure set according to the width learning model and the training feature set; if yes, generating the procedure branch of the workflow according to the training procedure set and the work system code. By integrating and planning the workflow data and automatically generating new procedure branches, the user's work pressure is reduced and the work efficiency of the background system is improved; the deviation parameter is introduced and the width learning method is used to finally generate the procedure branch of the workflow meeting the user's expectation. BRIEF DESCRIPTION OF DRAWINGS

[0129] Figure 1 It is a flowchart of the workflow management method based on width learning;

[0130] Figure 2 It is a step schematic diagram of the workflow management method based on width learning;

[0131] Figure 3 It is a flowchart of the workflow management method based on width learning;

[0132] Figure 4 It is an example diagram of the test procedure set in the workflow management method based on width learning;

[0133] Figure 5 It is an example diagram of the training feature set in the workflow management method based on width learning;

[0134] Figure 6 It is an example diagram of the training procedure set in the workflow management method based on width learning;

[0135] Figure 7 A structural schematic diagram of a workflow management system based on width learning;

[0136] Figure 8 An internal structure diagram of a computer device in an embodiment of the application. DETAILED DESCRIPTION

[0137] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0138] The system topology diagram of the workflow management method based on width learning provided by the present application is shown in Figure 1 The width learning is a mapping feature of a random vector function link neural network, an efficient incremental learning mode of network lateral expansion based on a single hidden layer neural network through a neural enhancement node and directly connecting the mapping feature and the enhancement node to an output end. The width learning is suitable for a system with few data features but high real-time prediction requirements. In order to realize automation, intelligentization and integration of work process management, the workflow can be regarded as a set composed of multiple basic tasks, and some task nodes have a sequence and a data transmission dependency relationship. A test process set and a training feature set are generated from collected process data; multiple data sets are generated from the training feature set, a weight matrix is generated from a user input bias parameter as a mapping feature through a width learning model, and then initial process node data is generated from the weight matrix and the multiple data sets, and a training process set is generated from the initial process node data and the data sets; a comparison between the test process and the training process determines an enhancement node of each process type and generates a bias rate set from the enhancement node and the training process set; whether each process bias rate is less than a user set bias rate threshold is calculated from the bias rate set; if not, the width learning model is trained according to the node bias rate corresponding to the process bias rate and the training process corresponding to the process bias rate is regenerated through the width learning model; if yes, a process branch of the workflow is generated from the training process and a work system code, and there are as many corresponding training processes as there are process bias rates less than the bias rate threshold, and there are as many process branches as there are corresponding training processes.

[0139] In one embodiment, as shown in Figure 2 The present application provides a workflow management method based on width learning, characterized in that the method comprises:

[0140] S201, acquiring process data corresponding to a workflow and generating a test process set and a training feature set from the process data;

[0141] S202, generating a width learning model according to the training feature set and generating a training procedure set according to the width learning model and the training feature set;

[0142] S203, comparing a test procedure in the test procedure set with a training procedure in the training procedure set and generating a deviation rate set according to the test procedure and the training procedure;

[0143] S204, determining whether a procedure deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set;

[0144] S205, if not, training the width learning model according to the deviation rate set and regenerating a training procedure set according to the width learning model and the training feature set;

[0145] S206, if yes, generating a procedure branch of the workflow according to the training procedure set and a work system code.

[0146] Specifically, a basic width learning model is generated according to a training feature set, then a deviation parameter is introduced, that is, an adjustable deviation parameter is set in the width learning model, then training procedures corresponding to each procedure type are generated according to the width learning model and the training feature set, then node deviation rates are obtained by comparing the training procedures with test procedures, each procedure type corresponding procedure deviation rate is calculated after weighting the node deviation rates, whether the procedure deviation rate is less than a deviation rate threshold is compared, if yes, the width learning model is retrained according to the corresponding node deviation rate and new training procedures are generated, if no, the corresponding procedure branch is generated according to the training procedure. Figure 3As shown, first a certain amount of process data is collected and the collected data is classified and integrated into a preset database. For example, existing process data includes: node post personnel, process type data (component FW verification process, HDD verification process, overall compatibility verification process, etc.), data types in the process such as: text, String, boolean, etc. Then, static feature extraction is performed on each of the process data to construct a workflow feature library, which is divided into a training set and a test set. According to the data of the training set, a basic model of width learning is constructed, and the test set and the width learning network training set after training are combined for comparison. Through comparison, a certain amount of difference data is obtained, and a deviation rate y (recorded for each process type) is obtained. Then, the deviation rate y is collected as a new feature vector set, and incremental learning is continued on the width learning basic model to optimize the basic model to obtain the final width learning model, until the deviation rate y is less than a certain threshold value such as 0.001%. Then, according to the width learning model generated by the data learning module, a preliminary process branch is generated: according to the node post personnel, process type data and data type in the process, combined with the existing system code, the system backend code and table structure are automatically generated, and the front-end page code is generated according to the data type in the process. After the generation of the new process branch, the system administrator is reminded by email to check, if the new process branch meets the requirements, the deviation rate of the process type is corrected to 0, if it does not meet the requirements, the deviation rate is modified to 100%, and the data learning instruction is initiated. Learning is performed again.

[0147] In one embodiment, the process data corresponding to the workflow is obtained, and the test process set and the training feature set are generated according to the process data, which includes:

[0148] The process data includes process node information and process type information;

[0149] The test process set is generated according to the process node information and the process type;

[0150] The static feature data is extracted from the process node information, and the training feature set is generated according to the static feature data and the process type.

[0151] Specifically, as Figure 4As shown, the present application can be used in business processes, including component FW verification processes, HDD verification processes, overall compatibility verification processes, and the like process types; and can also be used in company internal affairs processes, including leave application processes, office supplies application processes, vehicle pass application processes, and the like process types. The process node information corresponding to the component FW verification process includes SE nodes, TE nodes, TL nodes, and VM nodes, and the data types of the node data in these nodes include text, String, boolean, and the like. The node data and the process types are integrated through the relationships between the nodes in the collected process data to generate a test process, and finally a test process set is generated. For example Figure 5 As shown, the training feature set is generated after the static feature extraction of the process data, and the SIV nodes, HWE nodes, and PIV nodes are generated after the feature extraction of the SIV / HWE / PIV nodes. The training feature set contains the process node information and the process type information.

[0152] In one embodiment, generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set includes:

[0153] Generating a plurality of data sets according to the training feature set;

[0154] Generating a weight matrix according to the width learning model and a bias parameter input by a user;

[0155] Generating initial process node data according to the plurality of data sets and the weight matrix, and integrating the initial process node data and the data sets to generate the training process set.

[0156] Specifically, a data set [X, Y] is generated according to the training feature set, such as [component FW verification process, SE node], [HDD verification process, TE node], and the like. Then, the concept of bias rate is introduced into the existing width learning model, that is, the user sets an adjustable bias parameter d j in the existing width learning model. Different process types are assigned different bias parameters, and the bias parameter of each data set is determined by the process type in the data set. Then, a weight matrix is generated through the bias parameters of all data sets. However, the number of elements in the weight matrix generated each time is different because the number of data sets generated each time is uncertain. Then, the initial process node data FF = {W1, W2, …, WN} is generated according to the data set and the weight matrix μ. N As shown, the initial process node data is integrated with the data set [X, Y] to obtain the training process set H: H = [H1, H2, H3, H4, …, HN]. Figure 6 m ​].

[0157] In one embodiment, the comparing the test procedure in the test procedure set and the training procedure in the training procedure set and generating a set of bias rates according to the test procedure and the training procedure comprises:

[0158] determining an enhanced node according to the training procedure and the test procedure and generating inactivation data according to the enhanced node and the training procedure set;

[0159] calculating a node bias rate according to the inactivation data and the bias parameter and generating the set of bias rates according to the node bias rate, wherein the node bias rate comprises a node number bias rate, a node sequence bias rate, a node post bias rate and a node role bias rate.

[0160] Specifically, comparing the training procedure set H and the test set, for each procedure type, an enhanced node Z = [Z1, Z2, …, Zn] corresponding to the procedure type is generated; the training procedure set and the enhanced node group are combined to obtain inactivation data S: S = H

[0161] [Z n |H m ], and then according to the equation Y = [H m ,Z n ]d j , a set of bias rates Y = [Y1, Y2, Y3, Y4, …] is obtained, wherein each procedure type has four kinds of node bias rates, including a node number bias rate, a node sequence bias rate, a node post bias rate and a node role bias rate. If there are three training procedures in the training procedure set, there should be 12 node bias rates in the finally generated set of bias rates.

[0162] In one embodiment, the determining whether the procedure bias rate corresponding to the workflow is less than the bias rate threshold set by the user according to the set of bias rates comprises:

[0163] weighting the node bias rate according to the bias parameter and calculating the procedure bias rate corresponding to the procedure type according to the weighted node bias rate.

[0164] Specifically, first, the four kinds of node deviation rates are weighted by the deviation rate parameters corresponding to each process type, and the average of the four kinds of node deviation rates after weighting is the process deviation rate corresponding to the process type. If the process deviation rate is less than the deviation rate threshold set by the user, such as 0.001%, it means that the generated training process meets the requirements, otherwise it does not meet the requirements. For example, the process deviation rate of the component FW verification process does not meet the requirements, so a new component FW verification process needs to be generated. The HDD verification process and the overall compatibility verification process meet the requirements, so a new process branch is generated according to the training process of the HDD verification process and the training process of the overall compatibility verification process. The specific steps of generating a new training process set containing the component FW verification process include: taking the node deviation rate y of the process type as a new feature vector set, and performing incremental learning on the width learning model through the feature vector set to optimize the width learning model; then generate a new data set corresponding to the process type according to the training feature set, and then generate a new component FW verification process through the trained width learning model and the new data set, continue to determine whether the new process deviation rate is less than the deviation rate threshold, until it is less than 0.001%.

[0165] In one embodiment, the generating the process branch of the workflow according to the training process set and the working system code comprises:

[0166] The process branch comprises system backend code, table structure and front-end page code;

[0167] The system backend code and the table structure are generated according to the training process and the working system code;

[0168] The front-end page code is generated according to the process data type of the training process.

[0169] Specifically, a new training process is generated according to the width learning model: the system backend code and the table structure are automatically generated according to the node data and the process type data in the training process and in combination with the existing system code, the front-end page code is generated according to the data type corresponding to the process data, and the system backend code, the table structure and the front-end page code are generated.

[0170] In one embodiment, the generating the process branch of the workflow according to the training process set and the working system code comprises:

[0171] The process branch is sent to the user;

[0172] In response to the user determining that the process branch meets the requirements, the node deviation rates corresponding to the process branch are all modified to 0 and the process branch is published;

[0173] In response to the user determining that the process branch does not meet the requirement, the node deviation rate corresponding to the process branch is modified to 1, and the node deviation rate is returned to the width learning model.

[0174] Specifically, after the new process branch is generated, the system administrator can be reminded to check by email or the like. If the new process branch meets the requirement, the node deviation rate of the process type is modified to 0, and the process branch is published online. If the requirement is not met, the node deviation rate is modified to 1 and returned to the width learning model. The width learning model is retrained through these node deviation rates, and then a new training process is generated according to the width learning model and the training feature set.

[0175] The scheme of the present application has the following beneficial effects:

[0176] 1) By integrating and planning the workflow data and automatically generating a new process branch, the user's work pressure is reduced, and the work efficiency of the background system is improved.

[0177] 2) The bias parameter is introduced and the width learning method is used to finally generate the process branch of the workflow that meets the user's expectation.

[0178] It should be understood that, although Figure 2 the steps in the flowchart of the process are displayed in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated otherwise in this document, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, Figure 2 At least part of the steps in the flowchart can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0179] In one embodiment, as shown in Figure 7 Another width learning-based workflow management device, characterized in that the device comprises:

[0180] The first set generation module 701 is configured to obtain process data corresponding to the workflow and generate a test process set and a training feature set according to the process data.

[0181] The second set generation module 702 is configured to generate a width learning model according to the training feature set and generate a training process set according to the width learning model and the training feature set.

[0182] The third set generation module 703 is configured to compare the test processes in the test process set with the training processes in the training process set and generate a deviation rate set according to the test processes and the training processes.

[0183] The determination module 704 is configured to determine whether the process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set.

[0184] The training module 705 is configured to, if not, train the width learning model according to the process deviation rate and regenerate a training process set according to the width learning model and the training feature set.

[0185] The process branch generation module 706 is configured to, if yes, generate a process branch of the workflow according to the training process set and the work system code.

[0186] In one of the embodiments, the first set generation module acquires process data corresponding to a workflow and generates a test process set and a training feature set according to the process data, which includes:

[0187] The process data includes process node information and process type information.

[0188] The test process set is generated according to the process node information and the process type.

[0189] Static feature data is extracted from the process node information, and the training feature set is generated according to the static feature data and the process type.

[0190] In one of the embodiments, the second set generation module generates a width learning model according to the training feature set and generates a training process set according to the width learning model and the training feature set, which includes:

[0191] A plurality of data sets are generated according to the training feature set.

[0192] A weight matrix is generated according to the width learning model and a deviation parameter input by a user.

[0193] Initial process node data is generated according to the plurality of data sets and the weight matrix, and the training process set is generated by integrating the initial process node data and the data sets.

[0194] In one of the embodiments, the third set generation module compares the test processes in the test process set with the training processes in the training process set and generates a deviation rate set according to the test processes and the training processes, which includes:

[0195] determine an enhanced node according to the training procedure and the testing procedure and generate deactivation data according to the enhanced node and the training procedure set;

[0196] calculate a node bias rate according to the deactivation data and the bias parameter and generate the bias rate set according to the node bias rate, wherein the node bias rate comprises a node quantity bias rate, a node sequence bias rate, a node post bias rate and a node role bias rate.

[0197] In one of the embodiments, the determining module determines whether the procedure bias rate corresponding to the work flow is less than a bias rate threshold set by the user according to the bias rate set, comprising:

[0198] weight the node bias rate according to the bias parameter and calculate the procedure bias rate corresponding to the procedure type according to the weighted node bias rate.

[0199] In one of the embodiments, the procedure branch generating module generates the procedure branch of the work flow according to the training procedure set and the work system code, comprising:

[0200] the procedure branch comprises system backend code, table structure and front-end page code;

[0201] generate the system backend code and the table structure according to the training procedure and the work system code;

[0202] generate the front-end page code according to the procedure data type of the training procedure.

[0203] In one of the embodiments, the procedure branch generating module generates the procedure branch of the work flow according to the training procedure set and the work system code, comprising:

[0204] send the procedure branch to the user;

[0205] in response to the user determining that the procedure branch meets the requirement, modify all the node bias rates corresponding to the procedure branch to 0 and publish the procedure branch;

[0206] in response to the user determining that the procedure branch does not meet the requirement, modify all the node bias rates corresponding to the procedure branch to 1 and return the node bias rate to the width learning model.

[0207] The specific limitations of the workflow management apparatus based on the width learning can refer to the limitations of the workflow management method based on the width learning in the above, which will not be described here. Each module in the above workflow management apparatus based on the width learning can be realized by software, hardware and combinations thereof, in whole or in part. The above each module can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to the above each module.

[0208] In an embodiment, a computer device, which can be a terminal, can have an internal structure diagram as shown in Figure 8 The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement the alarm information processing method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0209] Those skilled in the art can understand that Figure 8 The structure shown in the above

[0210] In an embodiment, an electronic device is provided, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the computer program:

[0211] Obtaining process data corresponding to the workflow and generating a test process set and a training feature set according to the process data;

[0212] Generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set;

[0213] comparing the test procedure in the test procedure set and the training procedure in the training procedure set and generating a set of bias rates according to the test procedure and the training procedure;

[0214] determining whether the procedure bias rate corresponding to the workflow is less than a bias rate threshold set by a user according to the set of bias rates;

[0215] if not, training the width learning model according to the set of bias rates and regenerating a set of training procedures according to the width learning model and the set of training features;

[0216] if yes, generating a procedure branch of the workflow according to the set of training procedures and the working system code.

[0217] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0218] The obtaining of the procedure data corresponding to the workflow and the generating of the set of test procedures and the set of training features according to the procedure data comprises:

[0219] The procedure data comprises procedure node information and procedure type information;

[0220] The set of test procedures is generated according to the procedure node information and the procedure type;

[0221] The set of training features is generated according to the static feature data and the procedure type extracted from the procedure node information.

[0222] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0223] The generating of the width learning model according to the set of training features and the set of training procedures according to the width learning model and the set of training features comprises:

[0224] A plurality of data sets are generated according to the set of training features;

[0225] A weight matrix is generated according to the width learning model and a bias parameter input by a user;

[0226] Initial procedure node data is generated according to the plurality of data sets and the weight matrix, and the set of training procedures is generated by integrating the initial procedure node data and the data sets.

[0227] In one of the embodiments, the processor implements the following steps when executing the computer program:

[0228] The comparing the test flow in the test flow set and the training flow in the training flow set and generating a set of bias rates according to the test flow and the training flow comprises:

[0229] Determining an enhanced node according to the training flow and the test flow and generating inactivation data according to the enhanced node and the training flow set;

[0230] Calculating a node bias rate according to the inactivation data and the bias parameter and generating the set of bias rates according to the node bias rate, wherein the node bias rate comprises a node number bias rate, a node sequence bias rate, a node post bias rate and a node role bias rate.

[0231] In one embodiment, the processor implements the following steps when executing the computer program:

[0232] The determining whether the process bias rate corresponding to the workflow is less than a bias rate threshold set by the user according to the set of bias rates comprises:

[0233] The process type corresponding to the process bias rate is calculated according to the weighted node bias rate.

[0234] In one embodiment, the processor implements the following steps when executing the computer program:

[0235] The generating the process branch of the workflow according to the training flow set and the working system code comprises:

[0236] The process branch comprises system backend code, table structure and front-end page code;

[0237] The system backend code and the table structure are generated according to the training flow and the working system code;

[0238] The front-end page code is generated according to the process data type of the training flow.

[0239] In one embodiment, the processor implements the following steps when executing the computer program:

[0240] The generating the process branch of the workflow according to the training flow set and the working system code comprises:

[0241] The process branch is sent to the user;

[0242] In response to the user determining that the process branch meets the requirements, the node bias rates corresponding to the process branch are all modified to 0 and the process branch is published;

[0243] In response to the user determining that the process branch does not meet the requirement, modifying the node deviation rates corresponding to the process branch to all be 1 and returning the node deviation rates to the width learning model.

[0244] In one embodiment, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the following steps:

[0245] obtaining process data corresponding to a workflow and generating a test process set and a training feature set according to the process data;

[0246] generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set;

[0247] comparing a test process in the test process set with a training process in the training process set and generating a deviation rate set according to the test process and the training process;

[0248] determining whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set;

[0249] if not, training the width learning model according to the deviation rate set and regenerating a training process set according to the width learning model and the training feature set;

[0250] if yes, generating a process branch of the workflow according to the training process set and a work system code.

[0251] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0252] the obtaining process data corresponding to a workflow and generating a test process set and a training feature set according to the process data comprises:

[0253] the process data comprises process node information and process type information;

[0254] generating the test process set according to the process node information and the process type;

[0255] extracting static feature data from the process node information and generating the training feature set according to the static feature data and the process type.

[0256] In one embodiment, the computer program, when executed by a processor, implements the following steps:

[0257] The generating a width learning model according to the training feature set and generating a training flow set according to the width learning model and the training feature set comprises:

[0258] Generating a plurality of data sets according to the training feature set;

[0259] Generating a weight matrix according to the width learning model and a bias parameter input by a user;

[0260] Generating initial flow node data according to the plurality of data sets and the weight matrix and integrating the initial flow node data and the data sets to generate the training flow set.

[0261] In one of the embodiments, the computer program, when executed by the processor, implements the following steps:

[0262] The comparing a test flow in the test flow set with a training flow in the training flow set and generating a bias rate set according to the test flow and the training flow comprises:

[0263] Determining an enhanced node according to the training flow and the test flow and generating deactivation data according to the enhanced node and the training flow set;

[0264] Calculating a node bias rate according to the deactivation data and the bias parameter and generating the bias rate set according to the node bias rate, wherein the node bias rate comprises a node number bias rate, a node sequence bias rate, a node personnel post bias rate and a node personnel role bias rate.

[0265] In one of the embodiments, the computer program, when executed by the processor, implements the following steps:

[0266] The determining whether a flow bias rate corresponding to the work flow is less than a bias rate threshold set by a user according to the bias rate set comprises:

[0267] Weighting the node bias rate according to the bias parameter and calculating the flow bias rate corresponding to the flow type according to the weighted node bias rate.

[0268] In one of the embodiments, the computer program, when executed by the processor, implements the following steps:

[0269] The generating a flow branch of the work flow according to the training flow set and a work system code comprises:

[0270] The flow branch comprises a system backend code, a table structure and a front-end page code;

[0271] The generating the system backend code and the table structure according to the training flow and the work system code;

[0272] generating the front-end page code according to the process data type of the training process.

[0273] In one of the embodiments, the computer program, when executed by the processor, implements the following steps:

[0274] The generating the process branch of the workflow according to the set of training processes and the working system code further comprises:

[0275] sending the process branch to the user;

[0276] In response to the user determining that the process branch meets the requirement, modifying all the node deviation rates corresponding to the process branch to 0 and publishing the process branch;

[0277] In response to the user determining that the process branch does not meet the requirement, modifying all the node deviation rates corresponding to the process branch to 1 and returning the node deviation rates to the width learning model.

[0278] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, the computer program can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium in each embodiment provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM) and the like.

[0279] Each technical feature of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of each technical feature in the above embodiments are not described, however, as long as the combination of technical features does not exist, it should be considered as the scope of the present disclosure.

[0280] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a more specific and detailed manner, but should not be construed as limiting the scope of the patent. It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method of workflow management based on width learning, characterized by, The method is applied to a server configuration verification system, and the method comprises the following steps: Obtaining process data corresponding to a workflow, wherein the process data comprises process node information and process type information, the process type information at least comprises one of a component FW verification process, an HDD verification process, an overall compatibility verification process, a leave application process, an office supply application process or a vehicle pass application process, and the process node information at least comprises one of an SE node, a TE node, a TL node or a VM node corresponding to the component FW verification process, and the data type of node data corresponding to a process node at least comprises one of text, String or boolean; Generating a test process set according to the process node information and the process type; Extracting static feature data from the process node information and generating a training feature set according to the static feature data and the process type; Generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set; Determining an enhanced node of each process type by comparing the test process and the training process and generating inactivation data according to the enhanced node and the training process set; Calculating a node deviation rate corresponding to each process type according to the inactivation data and a bias parameter and generating a bias rate set according to the node deviation rate, wherein the node deviation rate comprises a node quantity deviation rate, a node sequence deviation rate, a node personnel post deviation rate and a node personnel role deviation rate; Determining whether a process deviation rate corresponding to the workflow is less than a bias rate threshold set by a user according to the bias rate set; If not, training the width learning model according to the bias rate set and regenerating a training process set according to the width learning model and the training feature set; If yes, generating system backend code and table structure according to node personnel, process type data and data type in the process and work system code, and generating front-end page code according to the process data type of the training process, wherein the process branch comprises the system backend code, the table structure and the front-end page code.

2. The method of claim 1, wherein, The step of generating a width learning model according to the training feature set and generating a training process set according to the width learning model and the training feature set comprises the following steps: Generating a plurality of data sets according to the training feature set; Generating a weight matrix according to the width learning model and a bias parameter input by a user; Generating initial process node data according to the plurality of data sets and the weight matrix and integrating the initial process node data and the data sets to generate the training process set.

3. The method of claim 2, wherein, The step of determining whether a process deviation rate corresponding to the workflow is less than a bias rate threshold set by a user according to the bias rate set comprises the following steps: Weighting the node deviation rate according to the bias parameter and calculating the process deviation rate corresponding to the process type according to the weighted node deviation rate.

4. The method of claim 1, wherein, The step of generating a process branch of the workflow according to the training process set and work system code comprises the following steps: Sending the process branch to the user; in response to the user determining that the process branch meets the requirement, modifying the node deviation rates corresponding to the process branch to all be 0 and publishing the process branch; in response to the user determining that the process branch does not meet the requirement, modifying the node deviation rates corresponding to the process branch to all be 1 and returning the node deviation rates to the width learning model.

5. An apparatus for implementing the method of any one of claims 1-4, wherein, The apparatus comprises: a first set generation module configured to obtain process data corresponding to a workflow and generate a test process set and a training feature set according to the process data; a second set generation module configured to generate a width learning model according to the training feature set and generate a training process set according to the width learning model and the training feature set; a third set generation module configured to compare a test process in the test process set with a training process in the training process set and generate a deviation rate set according to the test process and the training process; a determination module configured to determine whether a process deviation rate corresponding to the workflow is less than a deviation rate threshold set by a user according to the deviation rate set; a training module configured to, if not, train the width learning model according to the process deviation rate and regenerate a training process set according to the width learning model and the training feature set; a process branch generation module configured to, if yes, generate a process branch of the workflow according to the training process set and a work system code.

6. An electronic device, comprising: comprise: one or more processors; and a memory associated with the one or more processors, the memory configured to store program instructions that, when executed by the one or more processors, perform the method of any one of claims 1-4.

7. A computer storage medium, characterized in that a computer program stored thereon, wherein the program, when executed by a processor, implements the method of any one of claims 1-4.

Images