Data processing method and device, nonvolatile storage medium and computer equipment

Abstract

The invention discloses a data processing method and device, a nonvolatile storage medium and computer equipment. The method comprises the steps of collecting multi-dimensional attribute data corresponding to a target object; extracting a plurality of initial features from the multi-dimensional attribute data; inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features; weights corresponding to a plurality of preset indexes are obtained, and the preset indexes are standards for measuring the state of the target object; based on the plurality of target features and the weights corresponding to the plurality of preset indexes, determining scores corresponding to the plurality of preset indexes; and based on the scores corresponding to the plurality of preset indexes, determining a classification result corresponding to the target object, the classification result including a high-level object, a standard object, a low-level object and an object with a short board. The technical problem that an analysis result is inaccurate due to the fact that an existing analysis method depends on a single data source is solved.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and in particular, to a data processing method, apparatus, non-volatile storage medium, and computer device. Background Art

[0002] Currently, in the field of evaluating the safety production capabilities of risk operation personnel, traditional data processing methods often have limitations, which seriously hinder the comprehensiveness and accuracy of the evaluation. Traditional data processing methods often focus on data in one or several aspects and rely more on single-dimensional data. This one-sided data collection method cannot comprehensively reflect the actual safety production capabilities of operation personnel in complex working environments and ignores the fact that safety production capabilities are jointly determined by multiple factors.

[0003] In addition, the phenomenon of data islands generally exists within the industry. Different types of data, including natural attribute data of personnel (such as age, job type, and specialty), associated attribute data (such as qualifications, unit, and workgroup information), and behavioral attribute data (such as work history and future work plans), are difficult to be effectively integrated and analyzed. The potential correlations between data are not fully explored, resulting in data processing results that can neither reflect the dynamic changes of time nor have the depth of comprehensive consideration of multi-dimensional data, thus affecting the accuracy and timeliness of data analysis.

[0004] In response to the above problems, no effective solution has been proposed yet. Summary of the Invention

[0005] Embodiments of the present invention provide a data processing method, apparatus, non-volatile storage medium, and computer device to at least solve the technical problem that the current analysis method depends on a single data source, resulting in inaccurate analysis results.

[0006] According to one aspect of the embodiments of the present invention, a data processing method is provided, including: collecting multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extracting a plurality of initial features from the multi-dimensional attribute data; inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to realize the fusion between different initial features; obtaining the weights corresponding to a plurality of preset metrics, where the preset metrics are criteria for measuring the state of the target object; determining the scores corresponding to the plurality of preset metrics based on the plurality of target features and the weights corresponding to the plurality of preset metrics; and determining a classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0007] Optionally, multiple initial features are input into a preset graph convolutional neural network to obtain multiple target features, including: constructing a graph structure based on the multiple initial features, where the graph structure includes multiple nodes and edges between the multiple nodes, the nodes represent the initial features, and the edges represent the association relationships between the initial features corresponding to the respective two nodes; based on the graph convolutional neural network, performing a convolution operation on the initial features corresponding to each of the multiple nodes and the initial features corresponding to adjacent nodes on the graph structure to obtain multiple target features corresponding to each of the multiple nodes.

[0008] Optionally, the calculation expression of the target feature is as follows:

[0009]

[0010] where, z i is the target feature, y i is the initial feature corresponding to node i, y j is the initial feature corresponding to node j, node j is an adjacent node of node i, N(i) is the set of adjacent nodes of node i, d i is the number of adjacent nodes of node i, d j is the number of adjacent nodes of node j, W3 is the weight matrix of the graph convolutional neural network, and b3 is a preset bias vector.

[0011] Optionally, obtaining the weights corresponding to multiple preset metrics includes: constructing judgment matrices corresponding to the multiple preset metrics, where each of the multiple preset metrics includes multiple criteria, and the elements in the judgment matrix represent the importance degree of one criterion relative to another criterion; respectively calculating the eigenvectors of the judgment matrices corresponding to the multiple preset metrics as the weights corresponding to the multiple preset metrics.

[0012] Optionally, determining the classification result corresponding to the target object based on the scores corresponding to the multiple preset metrics includes: performing a weighted sum of the scores corresponding to the multiple preset metrics based on the preset weights corresponding to the multiple preset metrics to obtain the target score of the target person; judging the magnitude relationship between the target score and a preset scoring threshold; in the case where the target score is less than the preset scoring threshold, determining that the classification result is a low-level object.

[0013] Optionally, determining the classification result corresponding to the target object based on the scores corresponding to the multiple preset metrics includes: respectively judging the magnitude relationships between the scores corresponding to the multiple preset metrics and the preset metric thresholds corresponding to the multiple preset metrics; determining the preset metrics corresponding to the scores less than the preset metric thresholds among the multiple preset metrics as the target preset metrics; based on the target preset metrics, determining that the classification result is an object with a shortcoming.

[0014] According to another aspect of the embodiments of the present invention, a data processing device is further provided, including: an acquisition module, configured to acquire multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavior dimension attributes; an extraction module, configured to extract a plurality of initial features from the multi-dimensional attribute data; a fusion module, configured to input the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to implement the fusion between different initial features; an acquisition module, configured to acquire the weights corresponding to a plurality of preset metrics respectively, where the preset metrics are criteria for measuring the state of the target object; a first determination module, configured to determine the scores corresponding to the plurality of preset metrics respectively based on the plurality of target features and the weights corresponding to the plurality of preset metrics respectively; a second determination module, configured to determine the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics respectively, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0015] According to still another aspect of the embodiments of the present invention, a non-volatile storage medium is further provided. The non-volatile storage medium includes a stored program, where, when the program runs, it controls the device where the non-volatile storage medium is located to execute any one of the above data processing methods.

[0016] According to still another aspect of the embodiments of the present invention, a computer device is further provided. The computer device includes a processor, and the processor is used to run a program, where, when the program runs, it executes any one of the above data processing methods.

[0017] According to still another aspect of the embodiments of the present invention, a computer program product is further provided, including a computer program, where when the computer program is executed by a processor, it implements any one of the above data processing methods.

[0018] In an embodiment of the present invention, a data processing method is adopted. By collecting multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extracting a plurality of initial features from the multi-dimensional attribute data; inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to realize the fusion between different initial features; obtaining the weights corresponding to a plurality of preset metrics, where the preset metrics are criteria for measuring the state of the target object; determining the scores corresponding to the plurality of preset metrics based on the plurality of target features and the weights corresponding to the plurality of preset metrics; and determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards. This achieves the purpose of comprehensively considering multi-dimensional attributes for data analysis, thereby realizing the technical effect of improving the accuracy and reliability of the classification result, and further solving the technical problem that the current analysis method relies on a single data source, resulting in inaccurate analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and the schematic embodiments of the present invention and their descriptions are used to explain the present invention, and do not constitute an improper limitation to the present invention. In the drawings:

[0020] Figure 1 shows a hardware structure block diagram of a computer terminal for implementing the data processing method;

[0021] Figure 2 is a flowchart of the data processing method provided by the embodiment of the present invention;

[0022] Figure 3 is a flowchart of a risk operation personnel safety production ability evaluation method based on multi-source data fusion provided by an optional embodiment of the present invention;

[0023] Figure 4 is a structure block diagram of the data processing device provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order different from those illustrated or described here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0026] According to an embodiment of the present invention, an embodiment of a data processing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order from here.

[0027] The method embodiment provided in the first embodiment of the present application can be executed on a mobile terminal, a computer terminal or a similar computing device. Figure 1 The hardware structure block diagram of a computer terminal for implementing the data processing method is shown. As Figure 1 shown, the computer terminal 10 may include one or more (shown as 102a, 102b,..., 102n in the figure) processors (the processor may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), and a memory 104 for storing data. In addition, it may further include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which can be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 10 may further include more or fewer components than Figure 1 shown, or have a different configuration from Figure 1 shown.

[0028] It should be noted that one or more of the above-mentioned processors and / or other data processing circuits can generally be referred to as "data processing circuits" herein. The data processing circuit can be embodied in software, hardware, firmware, or any combination thereof, in whole or in part. In addition, the data processing circuit can be a single independent processing module, or can be integrated, in whole or in part, into any one of the other components in the computer terminal 10. As involved in the embodiments of the present application, the data processing circuit is a kind of processor control (such as the selection of a variable resistance terminal path connected to an interface).

[0029] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage devices corresponding to the data processing method in the embodiments of the present invention. The processor runs the software programs and modules stored in the memory 104 to execute various functional applications and data processing, that is, to implement the data processing method of the above-mentioned application program. The memory 104 can include high-speed random access memory, and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory 104 can further include a memory remotely set relative to the processor, and these remote memories can be connected to the computer terminal 10 through a network. Examples of the above-mentioned network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network, and combinations thereof.

[0030] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables the user to interact with the user interface of the computer terminal 10.

[0031] Figure 2 is a schematic flowchart of the data processing method provided according to the embodiments of the present invention, as Figure 2 shown, the method includes the following steps:

[0032] Step S201, collect multi-dimensional attribute data corresponding to the target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes.

[0033] In this step, the multi-dimensional attribute data of the target object can be collected through various channels and classified according to the three dimensions of natural attributes, associated attributes, and behavioral attributes. For example, the target object can be a risk operation personnel. The natural dimension attributes refer to static information closely related to the personal identity, characteristics, and conditions of the target object, and can include age, job type, specialty, etc.; the associated dimension attributes focus on the position and role of the target object in the working environment and social relationships, covering qualifications, units, work groups, etc.; the behavioral dimension attributes reflect the dynamic performance and behavioral patterns of the target object in actual work, including work history files, future work plans, etc.

[0034] Step S202: Extract multiple initial features from the multi-dimensional attribute data.

[0035] In this step, for each type of multi-dimensional attribute data collected, an autoencoder can be applied for feature extraction. For example, for the natural dimension attribute data, the autoencoder learns the internal representation of the operator's age, job type, and specialty through training and transforms it into a feature vector closely related to the work safety ability; for the associated dimension attribute data, the autoencoder can learn the work safety environment and condition features implied in attributes such as qualifications, units, and work teams; while for the behavioral dimension attribute data, the autoencoder can extract the dynamic behavior pattern features from the work history files, future work plans, and health data.

[0036] For example, for each dimension of data X collected (X can represent a certain type of data among natural attributes, associated attributes, or behavioral attributes), an autoencoder E is used for feature extraction. The autoencoder consists of an encoder E and a decoder D.

[0037] The encoder E maps the input data X to a low-dimensional space to obtain a feature vector y, and its mathematical expression is:

[0038] y = E(X) = σ1(W1X + b1)

[0039] where W1 is the weight matrix of the encoder, b1 is the bias vector, and σ1 is the activation function (e.g., ReLU function: σ1(x) = max(0, x)). Through this process, high-dimensional and complex data is converted into low-dimensional and compact feature vectors, reducing data redundancy.

[0040] The role of the decoder is to reconstruct the original data from the feature vector, and its expression is:

[0041] X' = D(y) = σ2(W2y + b2)

[0042] where W2 is the weight matrix of the decoder, b2 is the bias vector, and σ2 is the activation function (such as the Sigmoid function: ). When training the autoencoder, the parameters of the encoder and decoder are optimized by minimizing the reconstruction error. The commonly used reconstruction error metric is the mean squared error (MSE), and the calculation formula is:

[0043]

[0044] where n is the number of data samples. By continuously adjusting the weight matrices W1, W2 and the bias vectors b1, b2, the reconstruction error is minimized, thereby obtaining the optimal feature vector representation.

[0045] Step S203: Input multiple initial features into a preset graph convolutional neural network to obtain multiple target features, where the graph convolutional neural network is used to achieve the fusion between different initial features.

[0046] In this step, the Graph Convolutional Networks (GCN) is a deep learning model specifically designed to process graph-structured data. Different types of initial features can be regarded as nodes in a graph, and the associations between nodes (such as the correlation between age and health data, the connection between job types and qualifications, the influence of work history on accident records, etc.) are represented by the edges in the graph. Such a graph structure can more intuitively reflect the interaction and internal connection between data of different attributes. By applying the graph convolutional neural network to the graph structure, the GCN can achieve the fusion between different initial features by transmitting information between nodes. The core idea of the GCN is to perform a convolution operation on each node in the graph. Different from the traditional Convolutional Neural Network (CNN) that performs convolution on images, the GCN can perform convolution on the graph structure, which means it can fully consider the connection relationship between nodes and node features, thereby learning more rich feature representations. The GCN will perform a convolution operation on the feature vector of each node, and at the same time consider the feature information of its neighboring nodes. Through iterative updates, the feature representation of each node not only contains its own initial features, but also fuses the features of other nodes related to it, so as to obtain more comprehensive target features.

[0047] Step S204: Obtain the weights corresponding to multiple preset metrics, where the preset metrics are the criteria for measuring the state of the target object.

[0048] In this step, the preset metrics are a series of evaluation criteria determined according to actual needs and theoretical research, covering multiple aspects of data analysis of the target object, including but not limited to operation skills, safety knowledge mastery, emergency handling ability, health status, work history performance, rationality of future work plans, etc. In the evaluation system, the weight represents the relative importance of each metric in the comprehensive evaluation, which can reflect the proportion and influence degree of different metrics, and is a numerical representation of quantifying the contribution of each metric to the overall evaluation result. The weight can be obtained by using the Analytic Hierarchy Process (AHP). AHP is a structured decision analysis method. It constructs a hierarchical structure of metrics and uses expert judgment to determine the relative importance between metrics, and then calculates the weight of each metric.

[0049] Step S205: Based on the multiple target features and the weights corresponding to the multiple preset metrics, determine the scores corresponding to the multiple preset metrics.

[0050] In this step, through a mathematical model, the fused target features can be associated with the preset index weights, and then the quantitative scores of each index can be calculated to comprehensively reflect the analysis results of the target object. Specifically, the norms corresponding to multiple target features can be calculated first to obtain multiple norms, and the multiple norms can be combined into a new feature vector. At the same time, in order to better meet the requirements of actual production, the new feature vector can also be augmented based on expert experience to obtain an extended feature vector. Generally, the weights corresponding to multiple preset indexes can be represented in the form of a matrix. Therefore, multiplying the extended feature vector by the weights corresponding to multiple preset indexes respectively can obtain the scores corresponding to multiple preset indexes respectively.

[0051] Step S206: Determine the classification result corresponding to the target object based on the scores corresponding to multiple preset indexes, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0052] In this step, the system can classify the target object (i.e., the operator) into different level categories based on the scores of multiple preset indexes, including high-level objects, standard objects, low-level objects, and objects with short boards, so as to reveal the ability distribution of each person in the field of work safety. The system can score the comprehensive feature vector of each operator according to the preset evaluation indexes. These preset indexes are designed to comprehensively cover all aspects of work safety, from basic operation skills to complex safety knowledge application, and then to the response ability in case of emergencies, ensuring that the evaluation can reach every corner of work safety ability.

[0053] Through the above steps, the purpose of comprehensively considering multi-dimensional attributes for data analysis is achieved, thus realizing the technical effect of improving the accuracy and reliability of the classification result, and further solving the technical problem that the current analysis method depends on a single data source, resulting in inaccurate analysis results.

[0054] As an alternative embodiment, multiple initial features are input into a preset graph convolutional neural network to obtain multiple target features, including: based on multiple initial features, a graph structure is constructed, where the graph structure includes multiple nodes and edges between multiple nodes, the nodes represent the initial features, and the edges represent the association relationship between the initial features corresponding to the two corresponding nodes; based on the graph convolutional neural network, convolutional operations are respectively performed on the initial features corresponding to multiple nodes and the initial features corresponding to adjacent nodes on the graph structure to obtain multiple target features corresponding to multiple nodes respectively.

[0055] Optionally, after being processed by an autoencoder, initial features of different dimensions are obtained. Suppose there are m initial features y1, y2, …, y m, construct a graph structure \(G=(V, E)\), where the nodes \(V\) represent the respective initial features, and the edges \(E\) represent the association relationships between the initial features. The graph convolutional neural network fuses the initial features by performing convolutional operations on the graph structure. In this way, the association relationships between the data are fully considered, the structural information of the data is mined, the fusion of the initial features is achieved, and the fused target feature set \(Z = \{z_1, z_2, \cdots, z\) m \}\).

[0056] As an alternative embodiment, the calculation expression of the target feature is as follows:

[0057]

[0058] where \(z\) i is the target feature, \(y\) i is the initial feature corresponding to node \(i\), \(y\) j is the initial feature corresponding to node \(j\), node \(j\) is an adjacent node of node \(i\), \(N(i)\) is the set of adjacent nodes of node \(i\), \(d\) i is the number of adjacent nodes of node \(i\), \(d\) j is the number of adjacent nodes of node \(j\), \(W_3\) is the weight matrix of the graph convolutional neural network, and \(b_3\) is a preset bias vector.

[0059] Optionally, for each node in the graph, the GCN will perform a weighted sum of the initial feature of the node and the features of all its neighbor nodes according to certain weights and normalization coefficients, then perform a linear transformation through the weight matrix and the bias vector, and finally perform a non-linear mapping through the activation function to obtain the target feature of the node. Through this operation of the graph convolutional neural network, the limitations of traditional data fusion methods can be overcome, such as the inability to effectively process high-dimensional and heterogeneous data, and the correlation between data. The GCN can learn the local structural information of each node and consider the global structure of the entire graph, thereby obtaining a more accurate and comprehensive node feature representation.

[0060] As an alternative embodiment, obtaining the weights corresponding to multiple preset metrics includes: constructing judgment matrices corresponding to multiple preset metrics, where each of the multiple preset metrics includes multiple criteria, and the elements in the judgment matrix represent the importance degree of one criterion relative to another criterion; respectively calculating the eigenvectors of the judgment matrices corresponding to multiple preset metrics as the weights corresponding to multiple preset metrics.

[0061] Optionally, the analytic hierarchy process (AHP) can be used to determine the weights of each preset metric. First, construct the judgment matrix \(A=(a\) ij ), where \(a\) ij represents the importance degree of criterion \(i\) relative to criterion \(j\), \(a\) ijThe values are usually determined by experts based on experience or pairwise comparison and satisfy a ii = 1. Then calculate the maximum eigenvalue λ of the judgment matrix A max and the corresponding eigenvector W. The maximum eigenvalue can be calculated using methods such as the power method. After obtaining the maximum eigenvalue λ max , solve the linear equation system (A - λ max I)W = 0 (I is the identity matrix) to obtain the eigenvector W. Finally, normalize the eigenvector W to obtain the weight vector w = (w1, w2,..., w n ), where

[0062] As an alternative embodiment, based on the scores corresponding to multiple preset indicators, determine the classification result corresponding to the target object, including: based on the preset weights corresponding to multiple preset indicators, perform weighted summation on the scores corresponding to multiple preset indicators to obtain the target score of the target person; judge the magnitude relationship between the target score and the preset scoring threshold; in the case where the target score is less than the preset scoring threshold, determine that the classification result is a low-level object.

[0063] Optionally, each preset indicator has a preset weight, and these weights can be set based on expert experience. Once the scores of each preset indicator are determined, the preset weights can be used to perform weighted summation on these scores. To distinguish objects with different ability levels, one or more preset thresholds need to be set. These thresholds can be the average value of safety production capabilities obtained through statistical analysis of historical data, or the minimum requirements stipulated by industry standards or enterprise policies. For example, a low-level threshold may be set at 60 points, meaning that any comprehensive score below 60 points will be considered low-level. When the target score is less than the preset low-level threshold, the classification result is determined to be a "low-level object". This not only means that the current ability of the object is insufficient to meet the job requirements, but also may indicate that urgent measures need to be taken, such as strengthening training, supervision and guidance, or job transfer, to improve its safety production level and avoid potential accidents.

[0064] As an alternative embodiment, based on the scores corresponding to multiple preset indicators, determine the classification result corresponding to the target object, including: respectively judge the magnitude relationship between the scores corresponding to multiple preset indicators and the preset indicator thresholds corresponding to multiple preset indicators; determine the preset indicators whose corresponding scores are less than the preset indicator thresholds among multiple preset indicators as the target preset indicators; based on the target preset indicators, determine that the classification result is an object with a shortcoming.

[0065] Optionally, for each preset metric, there is a preset threshold, which represents the basic compliance level for that metric. It is set based on historical data, expert experience, or industry norms and is used to determine whether the target object meets the requirements for a specific metric. If the score of the target object for any one or more preset metrics is lower than the corresponding preset metric threshold, the system classifies the target object as an "object with weaknesses". This indicates that although it may perform well in some aspects, there is at least one or more weaknesses in terms of capabilities, and training, guidance, or targeted intervention measures are needed to improve these lacking capabilities.

[0066] As an alternative embodiment, a system and method for evaluating and improving the work safety capabilities of risk workers based on multi-source data fusion are also proposed. Figure 3 It is a schematic flowchart of a method for evaluating the work safety capabilities of risk workers based on multi-source data fusion provided according to an alternative embodiment of the present invention, as Figure 3 shown. First, data in three dimensions of the natural attributes of personnel (age, job type, specialty, etc.), associated attributes (qualifications, unit, work group, etc.), and behavioral attributes (work history files, future work plans, etc.) can be used to extract features from multi-source heterogeneous data using the autoencoder technique in deep learning. The autoencoder can learn the intrinsic feature representation of the data, convert high-dimensional and complex data into low-dimensional and compact feature vectors, and effectively reduce data redundancy. A data fusion method based on the graph convolutional neural network (GCN) is used to fuse different types of feature vectors together. GCN can consider the correlation relationships between data and fully exploit the structural information of the data during the fusion process to improve the fusion effect, and finally obtain a multi-dimensional work safety capability score for the worker. A work safety capability evaluation model is constructed based on the analytic hierarchy process (AHP) according to different work safety requirements, forming the work safety capability scores and weaknesses of the worker for different specific operations.

[0067] The multi-dimensional data fusion model constructed in this alternative embodiment covers data in three dimensions of the natural attributes, associated attributes, and behavioral attributes of personnel. This multi-dimensional data collection method breaks the limitation of traditional evaluations relying on single or a small number of dimensions of data. The natural attribute data provides the basic characteristics of the personnel for the evaluation, the associated attributes reflect the working environment and organizational relationships of the personnel, and the behavioral attributes reflect the dynamic situation of the work. When processing the data, the autoencoder extracts features from each dimension of the data, removes redundant information, and retains the core features; the graph convolutional neural network further fuses the feature vectors of different dimensions and fully exploits the associated structure between the data. Through this model, the work safety capabilities of the workers can be comprehensively and accurately characterized, the accuracy and reliability of the evaluation can be improved, and more valuable data support is provided for work safety management.

[0068] AHP can scientifically and reasonably determine the weights of evaluation indicators according to different safety production needs, such as the importance ranking of operating skills, safety knowledge mastery, emergency response capabilities and other indicators. The neural network, with its powerful nonlinear mapping ability, conducts in-depth analysis and prediction of the fused data. The combination of the two can not only accurately derive the safety production ability scores of operators in different specific operations, but also conduct in-depth analysis to identify the shortcomings of their abilities. Based on this, it is possible to formulate personalized capacity improvement strategies for operators, effectively improve their safety production capabilities, promote the transformation of safety production management from the traditional experience model to the scientific and precise model, and significantly improve the efficiency and quality of safety production management.

[0069] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0070] Through the description of the above implementation methods, those skilled in the art can clearly understand that the data processing method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0071] According to an embodiment of the present invention, a data processing device for implementing the above data processing method is also provided. Figure 4 is a structural block diagram of a data processing device provided according to an embodiment of the present invention, such as Figure 4 As shown, the device includes: a collection module 41, an extraction module 42, a fusion module 43, an acquisition module 44, a first determination module 45 and a second determination module 46. The device is described below.

[0072] The collection module 41 is used to collect multi-dimensional attribute data corresponding to the target object, wherein the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes and behavioral dimension attributes.

[0073] The extraction module 42 is connected to the acquisition module 41 and is used to extract a plurality of initial features from the multi-dimensional attribute data.

[0074] The fusion module 43, connected to the extraction module 42, is configured to input multiple initial features into a preset graph convolutional neural network to obtain multiple target features, where the graph convolutional neural network is used to achieve the fusion between different initial features.

[0075] The acquisition module 44, connected to the fusion module 43, is configured to acquire the weights corresponding to multiple preset metrics, where the preset metrics are criteria for measuring the state of the target object.

[0076] The first determination module 45, connected to the acquisition module 44, is configured to determine the scores corresponding to multiple preset metrics based on the multiple target features and the weights corresponding to the multiple preset metrics.

[0077] The second determination module 46, connected to the first determination module 45, is configured to determine the classification result corresponding to the target object based on the scores corresponding to the multiple preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0078] It should be noted here that the above-mentioned acquisition module 41, extraction module 42, fusion module 43, acquisition module 44, first determination module 45, and second determination module 46 correspond to steps S201 to S206 in the embodiment. The instances and application scenarios implemented by the multiple modules and the corresponding steps are the same, but are not limited to the content disclosed in the above embodiment. It should be noted that the above modules, as part of the device, can run in the computer terminal 10 provided in the embodiment.

[0079] An embodiment of the present invention can provide a computer device. Optionally, in this embodiment, the above computer device can be at least one network device among multiple network devices in a computer network. The computer device includes a memory and a processor.

[0080] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the data processing method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, to implement the above data processing method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory may further include a memory remotely set relative to the processor, and these remote memories can be connected to the computer terminal through a network. Examples of the above network include, but are not limited to, the Internet, enterprise intranets, local area networks, mobile communication networks, and their combinations.

[0081] The processor can call the information and application programs stored in the memory through the transmission device to execute the following steps: collect multi-dimensional attribute data corresponding to the target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extract a plurality of initial features from the multi-dimensional attribute data; input the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to realize the fusion between different initial features; obtain the weights corresponding to each of a plurality of preset metrics, where the preset metrics are the criteria for measuring the state of the target object; determine the scores corresponding to each of the plurality of preset metrics based on the plurality of target features and the weights corresponding to each of the plurality of preset metrics; determine the classification result corresponding to the target object based on the scores corresponding to each of the plurality of preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0082] Optionally, the above-mentioned processor can also execute the program code of the following steps: input the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, including: based on the plurality of initial features, construct a graph structure, where the graph structure includes a plurality of nodes and edges between the plurality of nodes, the nodes represent the initial features, and the edges represent the association relationship between the initial features corresponding to the two respective nodes; based on the graph convolutional neural network, perform convolutional operations on the initial features corresponding to each of the plurality of nodes and the initial features corresponding to adjacent nodes on the graph structure to obtain a plurality of target features corresponding to each of the plurality of nodes.

[0083] Optionally, the above-mentioned processor can also execute the program code of the following steps: the calculation expression of the target feature is as follows:

[0084]

[0085] where, z i is the target feature, y i is the initial feature corresponding to node i, y j is the initial feature corresponding to node j, node j is an adjacent node of node i, N(i) is the set of adjacent nodes of node i, d i is the number of adjacent nodes of node i, d j is the number of adjacent nodes of node j, W3 is the weight matrix of the graph convolutional neural network, and b3 is a preset bias vector.

[0086] Optionally, the above-mentioned processor can also execute the program code of the following steps: Obtain the weights corresponding to multiple preset metrics, including: constructing judgment matrices corresponding to multiple preset metrics, where each of the multiple preset metrics includes multiple criteria, and the elements in the judgment matrix represent the importance degree of one criterion to another criterion; respectively calculate the eigenvectors corresponding to the judgment matrices of multiple preset metrics as the weights corresponding to multiple preset metrics.

[0087] Optionally, the above-mentioned processor can also execute the program code of the following steps: Determine the classification result corresponding to the target object based on the scores corresponding to multiple preset metrics, including: based on the preset weights corresponding to multiple preset metrics, perform weighted summation on the scores corresponding to multiple preset metrics to obtain the target score of the target person; judge the magnitude relationship between the target score and the preset scoring threshold; in the case where the target score is less than the preset scoring threshold, determine that the classification result is a low-level object.

[0088] Optionally, the above-mentioned processor can also execute the program code of the following steps: Determine the classification result corresponding to the target object based on the scores corresponding to multiple preset metrics, including: respectively judge the magnitude relationship between the scores corresponding to multiple preset metrics and the preset metric thresholds corresponding to multiple preset metrics; determine the preset metrics whose corresponding scores are less than the preset metric thresholds among the multiple preset metrics as the target preset metrics; based on the target preset metrics, determine that the classification result is an object with a shortcoming.

[0089] By adopting the embodiment of the present invention, a data processing method is provided. By collecting multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extracting multiple initial features from the multi-dimensional attribute data; inputting the multiple initial features into a preset graph convolutional neural network to obtain multiple target features, where the graph convolutional neural network is used to realize the fusion between different initial features; obtaining the weights corresponding to multiple preset metrics, where the preset metrics are criteria for measuring the state of the target object; determining the scores corresponding to multiple preset metrics based on the multiple target features and the weights corresponding to multiple preset metrics; determining the classification result corresponding to the target object based on the scores corresponding to multiple preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with shortcomings, which achieves the purpose of comprehensively considering multi-dimensional attributes for data analysis, thereby realizing the technical effect of improving the accuracy and reliability of the classification result, and further solving the technical problem that the current analysis method depends on a single data source and leads to inaccurate analysis results.

[0090] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, and the storage medium can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disc, etc.

[0091] An embodiment of the present invention also provides a non-volatile storage medium. Optionally, in this embodiment, the above non-volatile storage medium can be used to store the program code executed by the data processing method provided in the above embodiment.

[0092] Optionally, in this embodiment, the above non-volatile storage medium can be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.

[0093] Optionally, in this embodiment, the non-volatile storage medium is set to store the program code for executing the following steps: collecting multi-dimensional attribute data corresponding to the target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extracting a plurality of initial features from the multi-dimensional attribute data; inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to realize the fusion between different initial features; obtaining the weights corresponding to a plurality of preset metrics respectively, where the preset metrics are the criteria for measuring the state of the target object; determining the scores corresponding to the plurality of preset metrics respectively based on the plurality of target features and the weights corresponding to the plurality of preset metrics respectively; and determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics respectively, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0094] Optionally, in this embodiment, the non-volatile storage medium is set to store the program code for executing the following steps: inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, including: constructing a graph structure based on the plurality of initial features, where the graph structure includes a plurality of nodes and edges between the plurality of nodes, the nodes represent the initial features, and the edges represent the association relationship between the initial features corresponding to the two corresponding nodes; and performing convolution operations on the initial features corresponding to the plurality of nodes and the initial features corresponding to the adjacent nodes on the graph structure respectively based on the graph convolutional neural network to obtain a plurality of target features corresponding to the plurality of nodes respectively.

[0095] Optionally, in this embodiment, the non-volatile storage medium is set to store the program code for executing the following steps: The calculation expression of the target feature is as follows:

[0096]

[0097] Among them, z i is the target feature, y i is the initial feature corresponding to node i, y j is the initial feature corresponding to node j. Node j is an adjacent node of node i, N(i) is the set of adjacent nodes of node i, d i is the number of adjacent nodes of node i, d j is the number of adjacent nodes of node j, W3 is the weight matrix of the graph convolutional neural network, and b3 is a preset bias vector.

[0098] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining the weights corresponding to multiple preset metrics, including: constructing a judgment matrix corresponding to each of the multiple preset metrics, where each of the multiple preset metrics includes multiple criteria, and the elements in the judgment matrix represent the importance degree of one criterion relative to another criterion; respectively calculating the eigenvectors corresponding to the judgment matrices of the multiple preset metrics as the weights corresponding to the multiple preset metrics.

[0099] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the classification result corresponding to the target object based on the scores corresponding to the multiple preset metrics, including: performing weighted summation on the scores corresponding to the multiple preset metrics based on the preset weights corresponding to the multiple preset metrics to obtain the target score of the target person; judging the magnitude relationship between the target score and a preset scoring threshold; and in the case where the target score is less than the preset scoring threshold, determining that the classification result is a low-level object.

[0100] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the classification result corresponding to the target object based on the scores corresponding to the multiple preset metrics, including: respectively judging the magnitude relationship between the scores corresponding to the multiple preset metrics and the preset metric thresholds corresponding to the multiple preset metrics; determining the preset metrics whose corresponding scores are less than the preset metric thresholds among the multiple preset metrics as the target preset metrics; and determining that the classification result is an object with a shortcoming based on the target preset metrics.

[0101] An embodiment of the present invention further provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can implement: collecting multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; extracting a plurality of initial features from the multi-dimensional attribute data; inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to achieve the fusion between different initial features; obtaining the weights corresponding to a plurality of preset metrics, where the preset metrics are criteria for measuring the state of the target object; determining the scores corresponding to the plurality of preset metrics based on the plurality of target features and the weights corresponding to the plurality of preset metrics; and determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

[0102] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0103] In the above embodiments of the present invention, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0104] In several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the units or modules can be in an electrical or other form.

[0105] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0106] In addition, in each embodiment of the present invention, the functional units can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0107] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on such understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0108] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A data processing method, characterized in that Including: Collecting multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; Extracting a plurality of initial features from the multi-dimensional attribute data; Inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to achieve the fusion between different initial features; Obtaining the weights corresponding to a plurality of preset metrics, where the preset metrics are criteria for measuring the state of the target object; Determining the scores corresponding to the plurality of preset metrics based on the plurality of target features and the weights corresponding to the plurality of preset metrics; Determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics, where the classification result includes high-level objects, standard objects, low-level objects, and objects with short boards.

2. The method according to claim 1, wherein The step of inputting the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features includes: Constructing a graph structure based on the plurality of initial features, where the graph structure includes a plurality of nodes and edges between the plurality of nodes, the nodes represent the initial features, and the edges represent the association relationships between the initial features corresponding to the two respective nodes; Performing convolutional operations on the initial features corresponding to the plurality of nodes and the initial features corresponding to adjacent nodes on the graph structure respectively based on the graph convolutional neural network to obtain the plurality of target features corresponding to the plurality of nodes respectively.

3. The method according to claim 2, wherein The calculation expression of the target features is as follows: Among them, z i is the target feature, y i is the initial feature corresponding to node i, y j is the initial feature corresponding to node j, where node j is an adjacent node of node i, and N(i) is the set of adjacent nodes of node i, d i is the number of adjacent nodes of node i, d j is the number of adjacent nodes of node j, W3 is the weight matrix of the graph convolutional neural network, and b3 is a preset bias vector.

4. The method according to claim 1, wherein The step of obtaining the weights corresponding to the plurality of preset metrics includes: Constructing a judgment matrix corresponding to the plurality of preset metrics, where each of the plurality of preset metrics includes a plurality of criteria, and the elements in the judgment matrix represent the importance degree of one criterion to another criterion; Calculating the eigenvectors corresponding to the judgment matrices of the plurality of preset metrics respectively as the weights corresponding to the plurality of preset metrics.

5. The method according to claim 1, wherein The step of determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics includes: Performing weighted summation on the scores corresponding to the plurality of preset metrics based on the preset weights corresponding to the plurality of preset metrics to obtain the target score of the target person; Judging the magnitude relationship between the target score and a preset scoring threshold; When the target score is less than the preset scoring threshold, determining that the classification result is a low-level object.

6. The method according to claim 1, wherein The step of determining the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics includes: Respectively judging the magnitude relationships between the scores corresponding to the plurality of preset metrics and the preset metric thresholds corresponding to the plurality of preset metrics; Determining the preset metrics corresponding to the scores less than the preset metric thresholds among the plurality of preset metrics as target preset metrics; Based on the target preset metrics, determining that the classification result is an object with short boards.

7. A data processing device, characterized in that, Including: A collection module, configured to collect multi-dimensional attribute data corresponding to a target object, where the multi-dimensional attribute data includes natural dimension attributes, associated dimension attributes, and behavioral dimension attributes; An extraction module, configured to extract a plurality of initial features from the multi-dimensional attribute data; A fusion module, configured to input the plurality of initial features into a preset graph convolutional neural network to obtain a plurality of target features, where the graph convolutional neural network is used to implement the fusion between different initial features; An acquisition module, configured to acquire the weights corresponding to a plurality of preset metrics, where the preset metrics are criteria for measuring the state of the target object; A first determination module, configured to determine the scores corresponding to the plurality of preset metrics respectively based on the plurality of target features and the weights corresponding to the plurality of preset metrics; A second determination module, configured to determine the classification result corresponding to the target object based on the scores corresponding to the plurality of preset metrics respectively, where the classification result includes a high-level object, a standard object, a low-level object, and an object with a shortcoming.

8. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, where when the program runs, it controls the device where the non-volatile storage medium is located to execute the data processing method according to any one of claims 1 to 6.

9. A computer device, characterized in that, Comprising: A memory and a processor, The memory stores a computer program; The processor is configured to execute the computer program stored in the memory, and when the computer program runs, it causes the processor to execute the data processing method according to any one of claims 1 to 6.

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

Images