Engineering project process and personnel matching method and system

Abstract

The invention discloses a process and personnel matching method and system for an engineering project, and the method comprises the steps: carrying out the content recognition of a historical engineering project file, and obtaining a first sample data set; performing feature extraction on the first sample data set to obtain process feature data of each engineering project participated by the target personnel and corresponding project result data; constructing a personnel project graph network according to the target personnel and process feature data, inputting the personnel project graph network into a first graph neural network model, and extracting implicit state information output by the first graph neural network model as empirical feature data; constructing personnel feature data based on the basic feature data and the empirical feature data; and in an actual processing process, inputting the obtained target process feature data and personnel feature data into the second graph neural network model for prediction, and generating a matching result corresponding to the target engineering project according to a prediction result. According to the method, the requirements of different types of projects can be flexibly met.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a method and system for matching process and personnel of an engineering project. Background Art

[0002] In today's complex and ever-changing engineering project environment, matching human resources with projects in a reasonable and efficient manner is one of the key challenges facing enterprises.

[0003] Traditional personnel and project matching methods often rely on manual experience and judgment. However, manual judgment lacks in-depth mining and systematic analysis of massive historical engineering project data. As the number of projects accumulated by enterprises increases, the fields, technologies and personnel involved become more and more complex. It is difficult to fully and accurately grasp the key features of each project and the potential relationship between personnel and matching by manual labor alone.

[0004] It can be seen that how to optimize the matching of engineering project processes and personnel to achieve matching of target personnel and engineering projects has become a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0005] The present application provides a method and system for matching the process and personnel of an engineering project to solve the technical problem of how to optimize the matching of the process and personnel of an engineering project.

[0006] In order to solve the above technical problems, the embodiment of the present application provides a method for matching process and personnel of an engineering project, including: Performing content recognition on the acquired historical engineering project files, and processing the content recognition results based on a natural language processing algorithm to obtain a first sample data set; Performing feature extraction on the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; Constructing a personnel-project graph network according to the target personnel and the process characteristic data, inputting the personnel-project graph network into a pre-trained first graph neural network model, and extracting implicit state information output by the first graph neural network model as the empirical characteristic data of the target personnel; wherein the first graph neural network model is designed to obtain the project result data according to the personnel-project graph network; Constructing personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data; In the actual data processing process, the acquired target process characteristic data of the target engineering project and the personnel characteristic data are input into the pre-trained second graph neural network model for prediction, and a matching result corresponding to the target engineering project is generated according to the prediction result.

[0007] As one preferred solution, the historical engineering project files include text format files and non-text format files; The content recognition result is processed based on the natural language processing algorithm to obtain a first sample data set, including: Performing text extraction, format conversion and segmentation labeling operations on the non-text format file in sequence to obtain standardized data; According to a natural language processing algorithm, entity recognition and relationship extraction are performed on the standardized data in sequence to obtain non-text data features; According to the natural language processing algorithm, entity recognition and relationship extraction are performed on the text format file in sequence to obtain text data features; The non-text data features and the text data features are fused to obtain the first sample data set.

[0008] As one preferred solution, before inputting the personnel-project graph network into the pre-trained first graph neural network model, the method further includes: Divide the personnel-project graph network into time windows based on preset time intervals to obtain time series graph data; Inputting the time series graph data into the first graph neural network model, and constructing a first loss function according to the output result of the first graph neural network model and the project result data; Based on the Adam optimization algorithm and the first loss function, the parameters of the first graph neural network model are optimized to obtain a trained first graph neural network model.

[0009] As one of the preferred solutions, the step of extracting implicit state information of the output of the first graph neural network model as the experience feature data of the target person includes: Input the time series graph data into the first graph neural network model to perform forward propagation calculation, obtain the average implicit state of the target person in each time window, and use the average implicit state as the window experience feature of the target person in the time window; Constructing a time weight curve corresponding to the time series graph data, wherein the time weight curve reflects the importance of the experience feature data in different time periods; Convert the window experience features of the target person in each time period into a vector representation to obtain a window experience feature vector; A weighted average calculation is performed on the window experience feature vector to obtain the experience feature data of the target person, wherein the weight coefficient of the weighted average calculation is the weight curve value of the corresponding time period.

[0010] As one preferred solution, the empirical characteristic data is expressed as: in, is the experience characteristic data of the target person, For target personnel exist The empirical eigenvector at time , is the weight coefficient.

[0011] As one preferred solution, the step of constructing a personnel-project graph network based on the target personnel and the process characteristic data includes: Extracting target personnel data corresponding to the process characteristic data based on a preset time interval; The personnel-project graph network is constructed by taking the target personnel data and the process characteristic data as nodes and the participation relationship between the target personnel data and the process characteristic data as edges.

[0012] As one of the preferred solutions, the step of constructing the personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data includes: Preprocessing the basic feature data, and performing correlation analysis on the preprocessed basic feature data to obtain target basic features; Quantifying and encoding the target basic features to obtain basic feature vectors; The empirical feature data and the basic feature vector are concatenated based on a weighted summation method to obtain the personnel feature data.

[0013] As one of the preferred solutions, before inputting the acquired target process feature data of the target engineering project and the personnel feature data into the pre-trained second graph neural network model for prediction, the method further includes: Constructing a second personnel-project graph network by taking the personnel characteristic data and the process characteristic data as nodes and the participation relationship between the personnel characteristic data and the process characteristic data as edges; Inputting the second person-project graph network into the second graph neural network model, and constructing a second loss function according to the model output result and the project result data; Determine a matching relationship matrix between the target person and the project result data based on the first sample data set, and optimize the second loss function according to the matching relationship matrix to obtain a target loss function; Based on the Adam optimization algorithm and the target loss function, the parameters of the second graph neural network model are optimized to obtain a trained second graph neural network model.

[0014] As one preferred solution, after generating a matching result corresponding to the target engineering project according to the prediction result, the method further includes: Analyze the matching results to obtain risk assessment results of the target engineering project; Generating a training plan for personnel corresponding to the target engineering project based on the risk assessment results; Training the personnel according to the training plan, updating the personnel's experience feature data according to the training results, inputting the updated personnel feature data and the target process feature data into the second graph neural network model for prediction, and obtaining a second matching result; If the second matching result is greater than a preset threshold, the operation of the target engineering project is performed according to the second matching result.

[0015] Another embodiment of the present application provides a system for matching process and personnel of an engineering project, including: An identification module is used to perform content identification on the acquired historical engineering project files, and process the content identification results based on a natural language processing algorithm to obtain a first sample data set; An extraction module, used to extract features from the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; An experience extraction module, used for constructing a personnel-project graph network according to the target personnel and the process characteristic data, inputting the personnel-project graph network into a pre-trained first graph neural network model, and extracting implicit state information output by the first graph neural network model as the experience characteristic data of the target personnel; wherein the first graph neural network model is designed to obtain the project result data according to the personnel-project graph network; A splicing module, used to construct personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data; The processing module is used to input the acquired target process characteristic data of the target engineering project and the personnel characteristic data into the pre-trained second graph neural network model for prediction in the actual data processing process, and generate a matching result corresponding to the target engineering project according to the prediction result.

[0016] Compared with the prior art, the beneficial effects of the embodiments of the present application are at least one of the following: 1) This application can accurately extract process feature data and personnel feature data through in-depth processing and analysis of historical engineering project files. In actual applications, the feature data and personnel feature data of the target engineering project are input into the second graph neural network model for prediction, and the generated matching results can provide enterprises with accurate personnel and project matching suggestions. This helps enterprises to quickly and accurately select the most suitable personnel to participate in the project before the project starts, greatly improving decision-making efficiency and reducing project risks caused by improper staffing.

[0017] 2) This application extracts the experience feature data of the target personnel from the first sample data set, so that the enterprise can fully explore and utilize the implicit knowledge accumulated by employees in past projects. Through the learning of the first graph neural network model, the experience of personnel in different projects is converted into quantifiable and usable experience features. This not only helps enterprises better understand the capabilities and potential of employees, but also provides data support for employees' career development planning.

[0018] 3) After generating the matching results, this application analyzes them to obtain risk assessment results, which can discover in advance the risks that may exist in the personnel matching of the target engineering project, such as project delays or quality problems caused by mismatched personnel skills, etc., which helps project managers take measures in advance to avoid risks and ensure the smooth progress of the project. In addition, a training plan is generated based on the risk assessment results, and targeted training is provided to personnel to improve personnel capabilities and thus reduce project risks. At the same time, the personnel experience feature data is updated based on the training results and matching predictions are performed again. If the second matching result is greater than the preset threshold, the project operation is carried out, realizing the dynamic optimization of personnel and project matching, and continuously improving the quality and efficiency of project execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of a method for matching process and personnel of an engineering project in one embodiment of the present application; Figure 2 It is a schematic diagram of a process and personnel matching system for an engineering project in one embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0022] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used in this article are only for illustrative purposes, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application. The term "and / or" used in this article includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0023] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0024] An embodiment of the present application provides a method for matching the process and personnel of an engineering project. For details, see Figure 1 , Figure 1 The flowchart of the method for matching the process and personnel of an engineering project in one embodiment of the present application is shown, which includes steps S1-S5: S1: performing content recognition on the acquired historical engineering project files, and processing the content recognition results based on a natural language processing algorithm to obtain a first sample data set; Among them, historical engineering project files are database files or paper archives of various storage systems within the enterprise. The collected file types include common text format files, such as Word documents (.doc, .docx), plain text files (.txt), Excel tables (.xlsx), etc., as well as non-text format files, such as PDF files (.pdf), scanned image files (.jpg, .png, etc.).

[0025] Preferably, in one embodiment of the present application, the historical engineering project files include text format files and non-text format files; The content recognition result is processed based on a natural language processing algorithm to obtain a first sample data set, including: Perform text extraction, format conversion and segmentation labeling operations on non-text format files in sequence to obtain standardized data; According to the natural language processing algorithm, entity recognition and relationship extraction are performed on the standardized data in sequence to obtain non-text data features; According to the natural language processing algorithm, entity recognition and relationship extraction are performed on the text format files in sequence to obtain text data features; The non-text data features and the text data features are fused to obtain a first sample data set.

[0026] Specifically, for PDF files, use professional OCR (optical character recognition) tools, such as Tesseract, ABBYY FineReader, etc. For scanned image files, also use OCR technology to pre-process the image first, such as grayscale, noise reduction, binarization, etc., to improve image quality to improve recognition accuracy, and then perform text extraction.

[0027] Convert the extracted text into a unified format that is easy to process later. Plain text format (.txt) is usually chosen because it is simple and easy to be read by various natural language processing tools. For some texts with complex formats (such as text with formatting tags extracted from PDF), use a text processing library (such as the re module in Python for regular expression processing) to remove unnecessary format information and only keep the plain text content.

[0028] Segment the text based on its logical structure and language habits. For example, for project reports, segment the text based on chapter titles, paragraph spacing, content themes, etc. Use natural language processing toolkits (such as NLTK, SpaCy, etc.) to perform entity recognition on standardized non-text data.

[0029] In one embodiment of the present application, the information about the project participants extracted from the non-text data and the project time node information extracted from the text data are merged together to form a comprehensive data set containing various aspects of the project. Ensure that during the fusion process, there will be no duplication or conflict for entities and relationships of the same type. If duplication occurs, it is screened and merged based on the reliability and completeness of the data. The fused data set finally obtained is the first sample data set, which provides a rich and structured data foundation for subsequent feature extraction, model training and other steps.

[0030] S2: extracting features from the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; In this step, the word vector model is used to vectorize the text description part contained in the data set (such as project introduction, personnel responsibilities description, etc.). Taking Word2Vec as an example, each word is mapped to a vector of fixed dimension by training on a large-scale text corpus (which can be project-related text data). For a sentence or a text, the vector representation of the text is obtained by averaging or other aggregation operations on the vectors of each word in it.

[0031] Design a CNN model suitable for project feature extraction, which includes multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer uses convolution kernels of different sizes to convolve the input data to extract local features. For example, for the vector representation of project text data, the convolution kernel can capture the local semantic features formed by the combination of adjacent words in the text. Pooling layers (such as maximum pooling or average pooling) are used to reduce the data dimension while retaining important features. The fully connected layer integrates the features output by the pooling layer to obtain the final project feature representation.

[0032] After the trained CNN model processes the input data, the process feature data and project result data are obtained from the specific layer of the model (such as the output of the fully connected layer). The process feature data includes process duration, process sequence, time interval, resource input and process qualification rate; the project result data includes project progress characteristics, project quality characteristics and project cost characteristics.

[0033] S3: constructing a personnel-project graph network according to the target personnel and process feature data, inputting the personnel-project graph network into a pre-trained first graph neural network model, and extracting implicit state information output by the first graph neural network model as the empirical feature data of the target personnel; wherein the first graph neural network model is designed to obtain project result data according to the personnel-project graph network; Preferably, in one embodiment of the present application, a personnel-project graph network is constructed according to target personnel and process characteristic data, including: Extracting target personnel data corresponding to process characteristic data based on a preset time interval; A personnel-project graph network is constructed with target personnel data and process characteristic data as nodes and the participation relationship between target personnel data and process characteristic data as edges.

[0034] For projects with short cycles and tight tasks, such as small software development projects, time intervals are set in days or weeks to more carefully capture the dynamic changes in personnel and project characteristics during the project progress. For large infrastructure construction projects, which have long cycles and obvious phased characteristics, the preset time intervals are months or quarters. For each preset time interval, information corresponding to the target personnel is extracted from the project-related data.

[0035] Preferably, in one embodiment of the present application, before inputting the person-project graph network into the pre-trained first graph neural network model, it also includes: Divide the personnel-project graph network into time windows based on preset time intervals to obtain time series graph data; Input the time series graph data into the first graph neural network model, and construct a first loss function according to the output result of the first graph neural network model and the project result data; Based on the Adam optimization algorithm and the first loss function, the parameters of the first graph neural network model are optimized to obtain the trained first graph neural network model.

[0036] The size of the time window needs to be determined based on the preset time interval and the level of detail of the project analysis. If the preset time interval is a week, for situations where detailed analysis of short-term project dynamics is required, the time window may be set to 3-4 consecutive weeks to capture the trend of changes in personnel and project characteristics in the short term.

[0037] The personnel-project graph network is divided along the time axis in sequence with the set time window. Starting from the time point when the project starts, the personnel nodes, project feature nodes and the edges between them in the first time window are extracted to form a subgraph. Then the time window is moved backward in sequence, each time by a preset time interval, and the corresponding subgraphs are continuously extracted until the entire project cycle is covered, thereby obtaining a series of time series graph data.

[0038] The divided time series graph data is input into the pre-trained first graph neural network model in sequence. After the model processes the graph data of each time window, it will output the corresponding prediction result, which includes the prediction of the status of the project in the time window (such as whether the project progress is normal, whether there are risks, etc.).

[0039] Compare the output of the model with the actual project result data (such as the actual progress of the project, whether it was successful in the end, etc.). Construct the first loss function based on the comparison results. The cross entropy loss function (such as predicting whether the project was successful) can be used to quantify the difference between the model prediction and the actual project results by calculating the loss function.

[0040] Specifically, in one embodiment of the present application, the constructed first loss function is substituted into the Adam optimization algorithm. In each iterative training, the Adam optimization algorithm first calculates the gradient of the model parameters, and updates the parameters of the model according to the gradient and the current learning rate. Specifically, it adjusts the update step size of each parameter according to the estimated values ​​of the first-order moment and second-order moment of the gradient. For example, for the weight parameters in the model, the update formula is, where is the learning rate, is the first-order moment estimate of the gradient, is the second-order moment estimate of the gradient, and is a very small constant used to prevent the denominator from being zero. After multiple rounds of iterative training, the parameters of the model are continuously adjusted so that the value of the first loss function gradually decreases, and finally the trained first graph neural network model is obtained, which can accurately process the time series graph data of the personnel-project graph network.

[0041] Preferably, in one embodiment of the present application, the implicit state information of the output of the first graph neural network model is extracted as the experience feature data of the target person, including: Input the time series graph data into the first graph neural network model for forward propagation calculation to obtain the average implicit state of the target person in each time window, and use the average implicit state as the window experience feature of the target person in the time window; Construct a time weight curve corresponding to the time series graph data, which reflects the importance of the empirical feature data in different time periods; Convert the window experience features of the target person in each time period into a vector representation to obtain a window experience feature vector; The window experience feature vector is weighted averaged to obtain the experience feature data of the target person, wherein the weight coefficient of the weighted average calculation is the weight curve value of the corresponding time period.

[0042] Specifically, the obtained time series graph data is input into the trained first graph neural network model in sequence according to the time window. Inside the model, the data is forward propagated through a series of graph convolution layers, activation function layers, etc. In the calculation process of each layer, the neuron will perform weighted summation based on the input data and its own weight, and perform nonlinear transformation through the activation function. After these complex calculations, the model will output the implicit state information in each time window. This implicit state information contains the model's understanding and abstract representation of the personnel-project graph network data in the time window. It integrates various project-related characteristics of the target personnel during the time period, such as the depth of participation in the project, the ability to solve problems, and the effect of collaboration with team members.

[0043] For each time window, the implicit states related to the target person output by the model are averaged to obtain the average implicit state. The average implicit state integrates the target person's performance in all aspects within the time window, and can more comprehensively reflect the project-related experience accumulated by the target person during this period of time, and is used as the window experience feature of the target person within the time window. For example, within a four-week time window, the implicit states of the target person at different time points are calculated by the model, and these implicit states are averaged. The result is the window experience feature within the time window, which represents the comprehensive experience accumulated by the target person in participating in the project during these four weeks.

[0044] In one embodiment of the present application, the importance of different time periods is evaluated based on the characteristics of the project and historical data, and the evaluation and data statistical results are integrated to form a continuous time weight curve. The weight value corresponding to each time point on the curve reflects the relative importance of the empirical feature data within the time period.

[0045] The window experience characteristics of the target personnel in each time period are converted into vector representation, and the window experience feature vector is weighted averaged according to the weight coefficient corresponding to the window experience feature vector and the time weight curve. The vector finally calculated is the experience feature data of the target personnel, which comprehensively considers the experience of the target personnel in different time windows and the importance of these experiences in the project, and provides key data support for subsequent tasks such as project matching and personnel capability assessment based on personnel characteristics.

[0046] Preferably, in one embodiment of the present application, the empirical characteristic data is expressed as: in, is the experience characteristic data of the target person, For target personnel exist The empirical eigenvector at time , is the weight coefficient S4: constructing personnel characteristic data based on the acquired basic characteristic data and experience characteristic data of the target personnel; Preferably, in one embodiment of the present application, constructing personnel characteristic data based on the acquired basic characteristic data and experience characteristic data of the target personnel includes: Preprocessing the basic feature data, and performing correlation analysis on the preprocessed basic feature data to obtain the target basic features; Quantify and encode the basic features of the target to obtain the basic feature vector; Based on the weighted summation method, the empirical feature data and the basic feature vector are concatenated to obtain the personnel feature data.

[0047] Among them, basic characteristic data include personal information, educational background and work experience. Preprocessing of the basic characteristic data includes removing duplicate records, correcting erroneous information in the data, and filling in the mean, median or mode if there are few missing values ​​and the feature is important for the overall analysis. Standardize the basic characteristic data of different dimensions to make them have a unified scale.

[0048] Use statistical analysis methods to perform correlation analysis on the preprocessed basic feature data. Set a correlation threshold, such as 0.5 (which can be adjusted according to actual conditions). For basic features whose correlation with the project relevance (or with the final target to be analyzed, such as the matching degree between personnel and project) is higher than the threshold, determine them as target basic features.

[0049] Specifically, for continuous values ​​in the target basic characteristics, such as years of work experience, they can be directly used as quantified values. For categorical target basic characteristics, such as majors, continue to use the one-hot encoding method. Based on experience or data analysis, weights are determined for the experience feature data and basic feature vectors respectively. The experience feature data and basic feature vectors are weighted and summed according to their respective weights. The final vector is the personnel feature data, which integrates the basic characteristics and experience characteristics of the target personnel, and provides a comprehensive and valuable data foundation for subsequent tasks such as personnel and project matching analysis and personnel capability assessment.

[0050] S5: In the actual data processing process, the target process feature data and personnel feature data of the target engineering project are input into the pre-trained second graph neural network model for prediction, and a matching result corresponding to the target engineering project is generated according to the prediction result.

[0051] Preferably, in one embodiment of the present application, before inputting the acquired target process feature data and personnel feature data of the target engineering project into the pre-trained second graph neural network model for prediction, the method further includes: The second personnel-project graph network is constructed by taking the personnel characteristic data and the process characteristic data as nodes and the participation relationship between the personnel characteristic data and the process characteristic data as edges; Inputting the second person-project graph network into the second graph neural network model, and constructing a second loss function based on the model output results and the project result data; Determine a matching relationship matrix between target personnel and project result data based on the first sample data set, and optimize the second loss function according to the matching relationship matrix to obtain a target loss function; Based on the Adam optimization algorithm and the target loss function, the parameters of the second graph neural network model are optimized to obtain the trained second graph neural network model.

[0052] Specifically, the constructed second person-project graph network is input into the pre-trained second graph neural network model. The model performs complex calculations and analyses on the nodes and edges in the graph network and outputs predictions on project results. The output of the model is compared with known project result data (from the first sample data set or other reliable data sources).

[0053] Based on the first sample data set, the intrinsic connection between the target personnel and the project result data is deeply analyzed to determine the matching relationship matrix. The elements in the matrix represent the strength of the association between the target personnel and the project results. For example, for a certain target person, if he has played an important role in multiple successful projects in the past, the element value corresponding to the successful project results in the matrix is ​​higher; conversely, if he participates more in failed projects, the corresponding element value may be lower. The matrix element value can be determined by statistically analyzing factors such as the frequency of participation and contribution of personnel in different project results. For example, the proportion of the number of times the target person plays a key role in successful projects to the total number of projects is counted as the basis for calculating the matrix element value. The matching relationship matrix is ​​converted into a regularized loss formula.

[0054] In one embodiment of the present application, optimizing the second loss function according to the matching relationship matrix includes: The matching relationship matrix is ​​converted into a regularized loss formula, and the second loss function is optimized according to the regularized loss formula, where the regularized loss formula is expressed as: in, represents the regularization loss, is the node feature matrix, is the adjacency matrix, is the matrix norm.

[0055] Then the regularized loss formula is combined with the second loss function to optimize the second loss function and obtain the target loss function. This optimized loss function can better constrain the model, making it pay more attention to the matching relationship between people and projects during the learning process, and improving the accuracy of model prediction.

[0056] Substitute the target loss function into the Adam optimization algorithm. In each iterative training, first calculate the gradient of the parameters of the second graph neural network model. According to the first-order moment and second-order moment estimation of the gradient, use the update formula of the Adam optimization algorithm to update the model parameters. Through continuous iteration, gradually adjust the model parameters so that the value of the target loss function gradually decreases. After multiple rounds of training, the trained second graph neural network model is obtained.

[0057] The acquired target process characteristic data and personnel characteristic data of the target project are input into the trained second graph neural network model. Based on the relationship pattern between personnel and project processes that the model has learned, the model predicts the degree of match between personnel and target project. The prediction result is expressed as a matching score in numerical form, and the higher the score, the better the match.

[0058] According to the prediction results of the model, the matching results corresponding to the target engineering project are generated. Different thresholds can be set to divide the matching between personnel and projects into different levels. For example, a matching score greater than 0.8 is a high match, 0.5-0.8 is a medium match, and less than 0.5 is a low match. The matching level of each target person and target engineering project is then organized into a report form to provide project decision makers with clear information on the matching of personnel and projects, helping them make reasonable personnel allocation decisions and improve the success rate and efficiency of the project.

[0059] Preferably, in one embodiment of the present application, after generating a matching result corresponding to the target engineering project according to the prediction result, the method further includes: Analyze the matching results to obtain the risk assessment results of the target engineering project; Generate training plans for personnel corresponding to the target engineering project based on risk assessment results; Training the personnel according to the training plan, updating the personnel's experience feature data according to the training results, inputting the updated personnel feature data and target process feature data into the second graph neural network model for prediction, and obtaining a second matching result; If the second matching result is greater than a preset threshold, the target engineering project is operated according to the second matching result.

[0060] First, conduct an in-depth analysis of the matching results, and identify possible risks based on the specific requirements, environment, and historical data of the target project. A variety of methods can be used, such as using historical project experience to create a risk checklist to check possible risk points.

[0061] Evaluate the identified risks to determine the probability of risk occurrence and the extent of the impact on the project. You can use qualitative evaluation methods, such as dividing the risk level into low, medium, and high levels; or use quantitative evaluation methods to calculate the probability of risk occurrence and the possible losses through mathematical models and statistical data.

[0062] Based on the results of the risk assessment, clearly identify the problems that need to be solved and the goals that need to be achieved through training. For example, if the risk assessment finds that there is a risk that personnel in the target project are not familiar with new technologies, then the training goal can be to enable relevant personnel to master the application of the new technology. Design specific training content based on the knowledge, skills, and operating specifications involved in the risk points. For example, for engineering projects with safety risks, the training content can include safe operating procedures, accident emergency response methods, safety awareness education, etc.; for risks involving professional technology, training courses on relevant professional knowledge and skills can be arranged, such as the use of engineering design software, learning of new construction processes, etc.

[0063] According to the training effect evaluation results, the experience characteristic data of the personnel will be updated. If the personnel performs well in the training and has mastered the relevant knowledge and skills, the weight and score of this aspect can be increased accordingly in their experience characteristic data; if the personnel still have deficiencies in certain aspects, their experience characteristic data can be appropriately adjusted to more accurately reflect their actual ability and level.

[0064] The updated personnel feature data and target process feature data are used as inputs to the pre-trained second graph neural network model. The second graph neural network model calculates and analyzes the input data, uses its existing training weights and algorithms, and re-predicts the degree of match between the personnel and the target project, and outputs the second matching result. Specifically, the model outputs a value between 0 and 1, representing the degree of match between the personnel and the target project. The closer the value is to 1, the higher the degree of match.

[0065] Analyze and interpret the second matching result to determine whether the personnel are more suitable for participating in the target project. If the second matching result is greater than the preset threshold, it means that after training and updating of experience feature data, the matching degree between the personnel and the target project has been improved and the requirements for project operations have been met; if the second matching result still does not reach the preset threshold, further analysis of the reasons may be required, and more in-depth training of the personnel or adjustment of personnel allocation may be required.

[0066] When the second matching result is greater than the preset threshold, personnel are reasonably arranged to participate in the work of the target engineering project according to the matching result. Specific work tasks and responsibilities can be assigned according to the expertise and capabilities of the personnel to ensure the smooth progress of the project.

[0067] An embodiment of the present application provides a system for matching process and personnel of an engineering project. For details, see Figure 2 , Figure 2 The schematic diagram of a system for matching process and personnel of an engineering project in one embodiment of the present application is shown, which includes: The identification module 11 is used to perform content identification on the acquired historical engineering project files, and process the content identification results based on a natural language processing algorithm to obtain a first sample data set; An extraction module 12 is used to extract features from the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; The experience extraction module 13 is used to construct a personnel-project graph network according to the target personnel and process feature data, input the personnel-project graph network into a pre-trained first graph neural network model, and extract implicit state information output by the first graph neural network model as the experience feature data of the target personnel; wherein the first graph neural network model is designed to obtain project result data according to the personnel-project graph network; A splicing module 14 is used to construct personnel characteristic data based on the acquired basic characteristic data and experience characteristic data of the target personnel; The processing module 15 is used to input the acquired target process characteristic data and personnel characteristic data of the target engineering project into the pre-trained second graph neural network model for prediction during the actual data processing process, and generate a matching result corresponding to the target engineering project according to the prediction result.

[0068] Compared with the prior art, the beneficial effects of the embodiments of the present application are at least one of the following: 1) This application can accurately extract process feature data and personnel feature data through in-depth processing and analysis of historical engineering project files. In actual applications, the feature data and personnel feature data of the target engineering project are input into the second graph neural network model for prediction, and the generated matching results can provide enterprises with accurate personnel and project matching suggestions. This helps enterprises to quickly and accurately select the most suitable personnel to participate in the project before the project starts, greatly improving decision-making efficiency and reducing project risks caused by improper staffing.

[0069] 2) This application extracts the experience feature data of the target personnel from the first sample data set, so that the enterprise can fully explore and utilize the implicit knowledge accumulated by employees in past projects. Through the learning of the first graph neural network model, the experience of personnel in different projects is converted into quantifiable and usable experience features. This not only helps enterprises better understand the capabilities and potential of employees, but also provides data support for employees' career development planning.

[0070] 3) After generating the matching results, this application analyzes them to obtain risk assessment results, which can discover in advance the risks that may exist in the personnel matching of the target engineering project, such as project delays or quality problems caused by mismatched personnel skills, etc., which helps project managers take measures in advance to avoid risks and ensure the smooth progress of the project. In addition, a training plan is generated based on the risk assessment results, and targeted training is provided to personnel to improve personnel capabilities and thus reduce project risks. At the same time, the personnel experience feature data is updated based on the training results and matching predictions are performed again. If the second matching result is greater than the preset threshold, the project operation is carried out, realizing the dynamic optimization of personnel and project matching, and continuously improving the quality and efficiency of project execution.

[0071] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations 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 present application shall be subject to the attached claims.

Claims

1. A method for matching process and personnel of an engineering project, characterized in that: include: Performing content recognition on the acquired historical engineering project files, and processing the content recognition results based on a natural language processing algorithm to obtain a first sample data set; Performing feature extraction on the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; Constructing a personnel-project graph network according to the target personnel and the process characteristic data, inputting the personnel-project graph network into a pre-trained first graph neural network model, and extracting implicit state information output by the first graph neural network model as the empirical characteristic data of the target personnel; wherein the first graph neural network model is designed to obtain the project result data according to the personnel-project graph network; Constructing personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data; In the actual data processing process, the acquired target process characteristic data of the target engineering project and the personnel characteristic data are input into the pre-trained second graph neural network model for prediction, and a matching result corresponding to the target engineering project is generated according to the prediction result.

2. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: The historical engineering project files include text format files and non-text format files; The content recognition result is processed based on the natural language processing algorithm to obtain a first sample data set, including: Performing text extraction, format conversion and segmentation labeling operations on the non-text format file in sequence to obtain standardized data; According to a natural language processing algorithm, entity recognition and relationship extraction are performed on the standardized data in sequence to obtain non-text data features; According to the natural language processing algorithm, entity recognition and relationship extraction are performed on the text format file in sequence to obtain text data features; The non-text data features and the text data features are fused to obtain the first sample data set.

3. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: Before inputting the person-project graph network into the pre-trained first graph neural network model, the method further includes: Divide the personnel-project graph network into time windows based on preset time intervals to obtain time series graph data; Inputting the time series graph data into the first graph neural network model, and constructing a first loss function according to the output result of the first graph neural network model and the project result data; Based on the Adam optimization algorithm and the first loss function, the parameters of the first graph neural network model are optimized to obtain a trained first graph neural network model.

4. The method for matching process and personnel of an engineering project according to claim 3, characterized in that: The step of extracting implicit state information of the output of the first graph neural network model as the experience feature data of the target person includes: Input the time series graph data into the first graph neural network model to perform forward propagation calculation, obtain the average implicit state of the target person in each time window, and use the average implicit state as the window experience feature of the target person in the time window; Constructing a time weight curve corresponding to the time series graph data, wherein the time weight curve reflects the importance of the experience feature data in different time periods; Convert the window experience features of the target person in each time period into a vector representation to obtain a window experience feature vector; A weighted average calculation is performed on the window experience feature vector to obtain the experience feature data of the target person, wherein the weight coefficient of the weighted average calculation is the weight curve value of the corresponding time period.

5. The method for matching process and personnel of an engineering project according to claim 4, characterized in that: The empirical characteristic data is expressed as: in, is the experience characteristic data of the target person, For target personnel exist The empirical eigenvector at time , is the weight coefficient.

6. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: The step of constructing a personnel-project graph network according to the target personnel and the process characteristic data includes: Extracting target personnel data corresponding to the process characteristic data based on a preset time interval; The personnel-project graph network is constructed by taking the target personnel data and the process characteristic data as nodes and the participation relationship between the target personnel data and the process characteristic data as edges.

7. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: The constructing of personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data includes: Preprocessing the basic feature data, and performing correlation analysis on the preprocessed basic feature data to obtain target basic features; Quantifying and encoding the target basic features to obtain basic feature vectors; The empirical feature data and the basic feature vector are concatenated based on a weighted summation method to obtain the personnel feature data.

8. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: Before inputting the acquired target process feature data of the target engineering project and the personnel feature data into the pre-trained second graph neural network model for prediction, the method further includes: Constructing a second personnel-project graph network by taking the personnel characteristic data and the process characteristic data as nodes and the participation relationship between the personnel characteristic data and the process characteristic data as edges; Inputting the second person-project graph network into the second graph neural network model, and constructing a second loss function according to the model output result and the project result data; Determine a matching relationship matrix between the target person and the project result data based on the first sample data set, and optimize the second loss function according to the matching relationship matrix to obtain a target loss function; Based on the Adam optimization algorithm and the target loss function, the parameters of the second graph neural network model are optimized to obtain a trained second graph neural network model.

9. The method for matching process and personnel of an engineering project according to claim 1, characterized in that: After generating a matching result corresponding to the target engineering project according to the prediction result, the method further includes: Analyze the matching results to obtain risk assessment results of the target engineering project; Generating a training plan for personnel corresponding to the target engineering project based on the risk assessment results; Training the personnel according to the training plan, updating the personnel's experience feature data according to the training results, inputting the updated personnel feature data and the target process feature data into the second graph neural network model for prediction, and obtaining a second matching result; If the second matching result is greater than a preset threshold, the operation of the target engineering project is performed according to the second matching result.

10. A system for matching process and personnel of an engineering project, characterized in that: include: An identification module is used to perform content identification on the acquired historical engineering project files, and process the content identification results based on a natural language processing algorithm to obtain a first sample data set; An extraction module, used to extract features from the first sample data set to obtain process feature data and corresponding project result data of each engineering project in which the target personnel participate; An experience extraction module, used for constructing a personnel-project graph network according to the target personnel and the process characteristic data, inputting the personnel-project graph network into a pre-trained first graph neural network model, and extracting implicit state information output by the first graph neural network model as the experience characteristic data of the target personnel; wherein the first graph neural network model is designed to obtain the project result data according to the personnel-project graph network; A splicing module, used to construct personnel characteristic data based on the acquired basic characteristic data of the target person and the experience characteristic data; The processing module is used to input the acquired target process characteristic data of the target engineering project and the personnel characteristic data into the pre-trained second graph neural network model for prediction in the actual data processing process, and generate a matching result corresponding to the target engineering project according to the prediction result.

Images