Object recognition method and device, electronic equipment and storage medium
By constructing an object relationship graph and utilizing graph convolution and object evaluation networks to identify key subgraphs, this approach addresses the problem of subjective factors affecting key team identification methods in existing technologies, achieving a more accurate assessment of team member contributions.
Patent Information
- Application Number
- CN202511221300.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-28
- Publication Date
- 2025-11-14
AI Technical Summary
Existing methods for identifying key groups are heavily influenced by subjective factors, leading to inaccurate evaluations.
By acquiring evaluation data from reference projects and attribute data from team members, an object relationship graph is constructed. Key subgraphs are identified using graph convolution and object evaluation networks, and the model is optimized to improve evaluation accuracy.
Effectively assess the contributions of team members and improve the accuracy of key team identification.
Smart Images
Figure CN120952630A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, applicable to the financial sector, and particularly to an object recognition method and apparatus, electronic device, and storage medium. Background Technology
[0002] Key team identification refers to identifying the most contributing groups within a team. For example, in the property insurance scenario within fintech, it's necessary to evaluate frontline property insurance teams to determine the most contributing groups. Current key team identification methods typically involve the human resources department manually collecting and analyzing data to generate scores, which are then used to identify key teams within the team. This manual evaluation method is heavily influenced by subjective factors, leading to inaccurate assessments. Therefore, improving the accuracy of key team evaluation is a pressing issue. Summary of the Invention
[0003] The main objective of this application is to provide an object recognition method and apparatus, electronic device and storage medium, aimed at improving the evaluation accuracy of key teams.
[0004] To achieve the above objectives, a first aspect of this application proposes an object recognition method, the method comprising:
[0005] Obtain reference evaluation data for the reference project, obtain reference object attribute data for the reference team, and obtain reference work record data for the reference team; wherein, the reference work record data is the record data recorded by the reference team when completing the reference project;
[0006] A graph is constructed based on the reference work record data to obtain a reference object relationship graph;
[0007] The reference object feature information is obtained by performing graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network.
[0008] Based on the feature information of the reference objects, key subgraphs are identified in the relationship graph of the reference objects to obtain a reference key relationship graph;
[0009] The reference key relationship graph and the reference object attribute data are evaluated by the preset object evaluation network to obtain the predicted object evaluation vector;
[0010] The target object evaluation network is obtained by optimizing the preset object evaluation network using the predicted object evaluation vector and the reference evaluation data.
[0011] Obtain target work record data of the target team, obtain target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and the target object attribute data.
[0012] In some embodiments, the reference work record data includes collaboration record data between reference objects; the step of constructing a graph based on the reference work record data to obtain a reference object relationship graph includes:
[0013] Graph nodes are constructed based on the reference object to obtain relational graph nodes;
[0014] Based on the cooperation record data between the reference objects, construct the edges between each node of the relationship graph to obtain the relationship graph edges;
[0015] A graph is constructed based on the nodes and edges of the relation graph to obtain a reference original relation graph;
[0016] Adjacent edge mining is performed on the original reference relationship graph to obtain the reference object relationship graph.
[0017] In some embodiments, the reference original relationship graph includes a first node, a second node, and a third node; the step of performing adjacency edge mining on the reference original relationship graph to obtain the reference object relationship graph includes:
[0018] If there is a connecting edge between the first node and the second node, and a connecting edge between the second node and the third node, the updated connecting edge is obtained by constructing an edge based on the first node and the third node;
[0019] The original reference relationship graph is updated based on the updated connection edges to obtain the reference object relationship graph.
[0020] In some embodiments, the step of performing graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network to obtain reference object feature information includes:
[0021] The number of connecting edges of each node in the reference object relationship graph is counted to obtain a node connecting edge count matrix.
[0022] The reference object feature information is obtained by performing graph convolution on the reference object relationship graph, the node connection edge number matrix, and the reference object attribute data through the preset object evaluation network.
[0023] In some embodiments, the step of identifying key subgraphs of the reference object relationship graph based on the reference object feature information to obtain a reference key relationship graph includes:
[0024] The reference object feature information is nonlinearly mapped to obtain mapped feature information;
[0025] Nodes with connecting edges in the reference object relationship graph are obtained to obtain selected node pairs; wherein, the selected node pair includes a first selected node and a second selected node;
[0026] The correlation degree is calculated based on the mapping feature information corresponding to the first selected node and the mapping feature information corresponding to the second selected node to obtain the node correlation degree between the first selected node and the second selected node.
[0027] Bernoulli relaxation sampling is performed on the node correlation to obtain the sampling probability;
[0028] If the sampling probability is less than a preset sampling threshold, the connection edges corresponding to the node correlation degree in the reference object relationship graph are deleted to obtain the reference key relationship graph.
[0029] In some embodiments, the step of evaluating the reference key relationship graph and the reference object attribute data through the preset object evaluation network to obtain a predicted object evaluation vector includes:
[0030] The key object feature information is obtained by performing graph convolution on the reference key relationship graph and the reference object attribute data through the preset object evaluation network.
[0031] The key object feature information is averaged to obtain the predicted object evaluation vector.
[0032] In some embodiments, optimizing the preset object evaluation network using the predicted object evaluation vector and the reference evaluation data to obtain the target object evaluation network includes:
[0033] The first loss value is obtained by calculating the divergence loss value based on the evaluation vector of the predicted object.
[0034] The predicted object evaluation vector is used to perform object data calculation to obtain the predicted object evaluation data;
[0035] The mean squared error is calculated based on the evaluation data of the predicted object and the attribute data of the reference object to obtain the second loss value;
[0036] The preset object evaluation network is optimized based on the first loss value and the second loss value to obtain the target object evaluation network.
[0037] To achieve the above objectives, a second aspect of this application provides an object recognition device, the device comprising:
[0038] The data acquisition module is used to acquire reference evaluation data of the reference project, reference object attribute data of the reference team, and reference work record data of the reference team; wherein, the reference work record data is the record data recorded by the reference team when completing the reference project;
[0039] The graph construction module is used to construct a graph based on the reference work record data to obtain a reference object relationship graph.
[0040] The graph convolution module is used to perform graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network to obtain reference object feature information.
[0041] The first identification module is used to identify key subgraphs of the reference object relationship graph based on the feature information of the reference object, and obtain a reference key relationship graph.
[0042] The object evaluation module is used to evaluate the reference key relationship graph and the reference object attribute data through the preset object evaluation network to obtain the predicted object evaluation vector;
[0043] The model optimization module is used to optimize the preset object evaluation network using the predicted object evaluation vector and the reference object attribute data to obtain the target object evaluation network.
[0044] The second identification module is used to acquire target work record data of the target team, acquire target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and the target object attribute data.
[0045] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.
[0046] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.
[0047] This application proposes an object recognition method, apparatus, electronic device, and storage medium, which acquires reference evaluation data of a reference project, reference object attribute data of a reference team, and reference work record data of the reference team, wherein the reference work record data is the record data recorded by the reference team when completing the reference project; constructs a graph based on the reference work record data to obtain a reference object relationship graph; performs graph convolution on the reference object relationship graph and reference object attribute data through a preset object evaluation network to obtain reference object feature information; identifies key subgraphs of the reference object relationship graph based on the reference object feature information to obtain a reference key relationship graph; evaluates objects based on the reference key relationship graph and reference object attribute data through a preset object evaluation network to obtain a predicted object evaluation vector; optimizes the preset object evaluation network using the predicted object evaluation vector and reference object attribute data to obtain a target object evaluation network; acquires target work record data and target object attribute data of the target team, constructs a graph based on the target work record data to obtain a target object graph, and identifies key subgraphs of the target object graph based on the target object evaluation network and target object attribute data. Therefore, by performing graph convolution on the reference object relationship graph and identifying key subgraphs, the embodiments of this application can effectively evaluate the contributions of team members, thereby improving the accuracy of key team identification. Attached Figure Description
[0048] Figure 1 This is a flowchart of the object recognition method provided in the embodiments of this application;
[0049] Figure 2 yes Figure 1 The flowchart of step S102 in the document;
[0050] Figure 3 yes Figure 2 The flowchart of step S204 in the process;
[0051] Figure 4 yes Figure 1 The flowchart of step S103 in the process;
[0052] Figure 5 yes Figure 1 The flowchart of step S104 in the process;
[0053] Figure 6 yes Figure 1 The flowchart of step S105 in the process;
[0054] Figure 7 yes Figure 1 The flowchart of step S106 in the process;
[0055] Figure 8 This is a schematic diagram of the structure of the object recognition device provided in the embodiments of this application;
[0056] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation
[0057] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0058] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.
[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0060] Key team identification refers to identifying the most contributing groups within a team. For example, in the property insurance scenario within fintech, it's necessary to evaluate frontline property insurance teams to determine the most contributing groups. Current key team identification methods typically involve the human resources department manually collecting and analyzing data to generate scores, which are then used to identify key teams within the team. This manual evaluation method is heavily influenced by subjective factors, leading to inaccurate assessments. Therefore, improving the accuracy of key team evaluation is a pressing issue.
[0061] Based on this, embodiments of this application provide an object recognition method and apparatus, an electronic device and a storage medium, aimed at improving the evaluation accuracy of key teams.
[0062] This application provides an object recognition method, apparatus, electronic device, and storage medium, which are specifically described through the following embodiments. First, the object recognition method in this application is described.
[0063] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0064] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.
[0065] The object recognition method provided in this application relates to the field of data processing technology and is applicable to the financial sector. The object recognition method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the object recognition method, but is not limited to the above forms.
[0066] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0067] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.
[0068] Figure 1 This is an optional flowchart of the object recognition method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S107.
[0069] Step S101: Obtain reference evaluation data for the reference project, obtain reference object attribute data for the reference team, and obtain reference work record data for the reference team; wherein, the reference work record data is the record data recorded by the reference team when completing the reference project.
[0070] Step S102: Construct a graph based on the reference work record data to obtain a reference object relationship graph;
[0071] Step S103: Perform graph convolution on the reference object relationship graph and reference object attribute data through a preset object evaluation network to obtain reference object feature information;
[0072] Step S104: Based on the feature information of the reference object, identify the key subgraphs of the reference object relationship graph to obtain the reference key relationship graph;
[0073] Step S105: The reference key relationship graph and reference object attribute data are evaluated by a preset object evaluation network to obtain the predicted object evaluation vector.
[0074] Step S106: Optimize the preset object evaluation network by using the predicted object evaluation vector and reference object attribute data to obtain the target object evaluation network;
[0075] Step S107: Obtain the target work record data of the target team, obtain the target object attribute data of the target team, construct a graph based on the target work record data to obtain the target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and the target object attribute data.
[0076] Steps S101 to S107 as illustrated in this embodiment involve acquiring reference evaluation data for the reference project, reference object attribute data for the reference team, and reference work record data for the reference team, wherein the reference work record data is the record data recorded by the reference team when completing the reference project; constructing a graph based on the reference work record data to obtain a reference object relationship graph; performing graph convolution on the reference object relationship graph and reference object attribute data using a preset object evaluation network to obtain reference object feature information; identifying key subgraphs of the reference object relationship graph based on the reference object feature information to obtain a reference key relationship graph; evaluating objects using the reference key relationship graph and reference object attribute data using a preset object evaluation network to obtain a predicted object evaluation vector; optimizing the preset object evaluation network using the predicted object evaluation vector and reference object attribute data to obtain a target object evaluation network; acquiring target work record data and target object attribute data of the target team; constructing a graph based on the target work record data to obtain a target object graph; and identifying key subgraphs of the target object graph based on the target object evaluation network and target object attribute data. Therefore, by performing graph convolution on the reference object relationship graph and identifying key subgraphs, the embodiments of this application can effectively evaluate the contributions of team members, thereby improving the accuracy of key team identification.
[0077] In step S101 of some embodiments, the reference project is a specific project, such as a periodic project in a property insurance scenario. The reference evaluation data is the completion rate of indicators related to the project or the results of manual scoring. The reference team is the team that completed the reference project. The reference object attribute data is the identity attribute information of the team members, including but not limited to position, work experience, etc. The reference work record data is the collaboration record between employees; for example, if employee A and employee B jointly completed a sub-project, this collaboration will generate a triplet record, recording employee A, employee B, and the jointly completed sub-project.
[0078] Please see Figure 2 In some embodiments, the reference work record data includes collaboration record data between reference objects; step S102 may include, but is not limited to, steps S201 to S204:
[0079] Step S201: Construct graph nodes based on the reference object to obtain relation graph nodes;
[0080] Step S202: Based on the cooperation record data between reference objects, construct the edges between each relationship graph node to obtain the relationship graph edges;
[0081] Step S203: Construct a graph based on the nodes and edges of the relation graph to obtain a reference original relation graph;
[0082] Step S204: Perform adjacent edge mining on the original reference relationship graph to obtain the reference object relationship graph.
[0083] Steps S201 to S204, as shown in this embodiment, involve constructing graph nodes based on reference objects to obtain relationship graph nodes; then, constructing edges between each relationship graph node based on the cooperation record data between reference objects to obtain relationship graph edges; next, constructing a graph based on the relationship graph nodes and relationship graph edges to obtain a reference original relationship graph; finally, performing adjacency edge mining on the reference original relationship graph to obtain a reference object relationship graph. Thus, this embodiment constructs relationship graph nodes based on reference objects and relationship graph edges based on the cooperation records between reference objects, thereby constructing complex relationships between objects using a reference original relationship graph, and further optimizing the obtained reference object relationship graph through adjacency edge mining.
[0084] In step S201 of some embodiments, the reference object is each employee in the team, and each employee is represented as a node in the graph by their unique identification information (such as employee number, name, etc.). Each employee is treated as a separate node, forming an initial set of nodes. For example, assuming there are employees A, B, and C, a node will be created for each employee, and no relationships have yet been established between the nodes.
[0085] In step S202 of some embodiments, if two employees have previously completed a task or project together, an edge is created between the two employees in the graph to represent their collaborative relationship. For example, if employee A and employee B have jointly completed a small goal, an edge is created between employee A and employee B in the graph, forming a connection in the graph.
[0086] In step S203 of some embodiments, a complete relationship graph is generated by combining nodes and edges to accurately reflect all cooperative relationships between employees. For example, if there is a cooperative relationship between employee A and employee B, and between employee B and employee C, these nodes and edges are used to construct a graph containing employees A, B, C and their corresponding edges, forming a reference to the original relationship graph.
[0087] Please see Figure 3 In some embodiments, the original relational graph includes a first node, a second node, and a third node; step S204 may include, but is not limited to, steps S301 to S302:
[0088] Step S301: If there is a connecting edge between the first node and the second node, and a connecting edge between the second node and the third node, construct the edge based on the first node and the third node to obtain the updated connecting edge;
[0089] Step S302: Update the original reference relationship graph according to the updated connection edges to obtain the reference object relationship graph.
[0090] In the embodiments of this application, steps S301 to S302 are performed as follows: if there is a connecting edge between the first node and the second node, and a connecting edge between the second node and the third node, the edges are constructed based on the first node and the third node to obtain updated connecting edges. Then, the original reference relationship graph is updated based on the updated connecting edges to obtain a reference object relationship graph, thereby uncovering potential cooperative relationships and further optimizing the expressive power of the graph.
[0091] In steps S301 to S302 of some embodiments, it is assumed that there are three employees A, B, and C, and that there is a cooperative relationship between employee A and employee B (i.e., there is an edge between node A and node B), and also a cooperative relationship between employee B and employee C (i.e., there is an edge between node B and node C). Since there is an edge between employee A and employee B, and also between employee B and employee C, a new connection edge is added between employee A and employee C, i.e., the connection edge is updated.
[0092] Next, the edge between employee A and employee C, i.e., the newly added connecting edge, is added to the original reference relationship graph to obtain the reference object relationship graph.
[0093] It should be noted that existing graph constructions often rely on direct collaborations between employees, neglecting the importance of indirect collaborations. This makes it difficult to accurately represent deep collaborations within a project and reduces the quality of the identified key collaboration groups.
[0094] Please see Figure 4 In some embodiments, step S103 may include, but is not limited to, steps S401 to S402:
[0095] Step S401: Count the number of connecting edges of each node in the reference object relationship graph to obtain the node connecting edge count matrix.
[0096] Step S402: Perform graph convolution on the reference object relationship graph, the node connection edge number matrix, and the reference object attribute data through a preset object evaluation network to obtain the reference object feature information.
[0097] In step S401 of some embodiments, the node connection edge number matrix refers to the degree matrix of each node in the reference object relationship graph. The degree matrix represents the number of edges connected to each node. The degree of each node represents the "activity level" of that node in the graph, that is, the number of direct relationships with other nodes. For example, in a relationship graph composed of employees, if employee A has a cooperative relationship with employees B and C, then employee A has a degree of 2 (connected to B and C by edges).
[0098] In step S402 of some embodiments, the preset object evaluation network is a Graph Convolutional Network (GCN). GCN is a neural network model specifically designed for graph data, capable of effectively learning features from nodes in a graph. By performing convolution operations on the adjacency information of nodes, GCN can capture the local structural information of nodes, as shown in equations (1) and (2).
[0099]
[0100] There are multiple reference items, and the reference object relationship diagram corresponds to the reference items. Each reference item has a subscript i, H. i That is, the reference object feature information of the i-th reference project, G i Let X be the reference object relationship graph for the i-th reference object, X be the reference object attribute data for the i-th reference project, A be the reference object relationship graph, and D be the matrix of the number of nodes connected by edges.
[0101] Please see Figure 5 In some embodiments, step S104 includes, but is not limited to, steps S501 to S505:
[0102] Step S501: Perform nonlinear mapping on the feature information of the reference object to obtain the mapped feature information;
[0103] Step S502: Obtain nodes with connecting edges in the reference object relationship graph to obtain selected node pairs; wherein, the selected node pairs include a first selected node and a second selected node;
[0104] Step S503: Calculate the correlation degree based on the mapping feature information corresponding to the first selected node and the mapping feature information corresponding to the second selected node to obtain the node correlation degree between the first selected node and the second selected node.
[0105] Step S504: Perform Bernoulli relaxation sampling on the node correlation degree to obtain the sampling probability;
[0106] Step S505: If the sampling probability is less than the preset sampling threshold, delete the connection edges corresponding to the node correlation degree in the reference object relationship graph to obtain the reference key relationship graph.
[0107] Steps S501 to S505 of this embodiment involve performing a nonlinear mapping on the feature information of the reference object to obtain mapped feature information; obtaining nodes with connecting edges in the reference object relationship graph to obtain selected node pairs; wherein, the selected node pairs include a first selected node and a second selected node; calculating the correlation degree based on the mapping feature information corresponding to the first selected node and the second selected node to obtain the node correlation degree between the first selected node and the second selected node; performing Bernoulli relaxation sampling on the node correlation degree to obtain the sampling probability; if the sampling probability is less than a preset sampling threshold, deleting the connecting edges corresponding to the node correlation degree in the reference object relationship graph to obtain a reference key relationship graph. Thus, this embodiment solves the problem of too many edges in the graph leading to a decrease in model accuracy by deleting unimportant connecting edges, making the graph more concise and highlighting the relationships between key nodes, thereby improving the efficiency and accuracy of graph analysis.
[0108] In some embodiments, steps S501 to S505 involve nonlinear mapping as shown in equation (3):
[0109] Z(v i ) = MLP(h i (3),
[0110] Where i is the object index, i.e., the i-th employee, Z(v i ) represents the mapping feature information, h i For reference object feature information.
[0111] First, all nodes with connecting edges in the reference object relationship graph are selected. Then, these nodes are paired according to the relationships of the connecting edges to form selected node pairs. Each pair includes a first selected node and a second selected node. For example, in an employee collaboration relationship graph, employee A has a collaboration relationship with employee B, and employee B has a collaboration relationship with employee C. Employee A and employee B form a selected node pair, and employee B and employee C also form a selected node pair.
[0112] For example, consider a team relationship graph with nodes representing employees A, B, C, and D, and edges connecting them representing collaborative relationships. If employee A collaborates with employee B, employee B collaborates with employee C, and employee C collaborates with employee D, then the selected node pairs would be: (Employee A, Employee B), (Employee B, Employee C), and (Employee C, Employee D).
[0113] The correlation degree is calculated as shown in equation (4):
[0114] p ij =sigmoid(Z(v i )Z(v j ) T(4),
[0115] p ij Z(v) represents the node correlation degree. i Z(v) represents the mapping feature information corresponding to the first selected node. j ) represents the mapping feature information corresponding to the second selected node.
[0116] Bernoulli relaxation sampling is shown in equation (5):
[0117]
[0118] Among them, b ij Let t be the sampling probability, and t be a freely settable temperature parameter with a positive value, ∈ ~Uniform(0,1).
[0119] In step S504 of some embodiments, the connection edge deletion is as shown in equation (6):
[0120]
[0121] Where, q ij This is the mask for the connecting edges. When the mask is 1, it means to keep the edge; when the mask is 0, it means to delete the connecting edge.
[0122] It should be noted that in step S204, adjacency edge mining is performed on the original reference relationship graph to obtain the reference object relationship graph, which further enriches the graph's expressive power. It not only mines the direct cooperative relationships between employees, but also the indirect cooperative relationships between employees, thereby accurately representing the deep cooperative relationships within the project.
[0123] Then, in steps S501 to S505, key groups within the team are identified by sequentially applying nonlinear mapping, correlation calculation, Bernoulli relaxation sampling, and threshold comparison. First, nonlinear mapping transforms the feature information of the reference objects, capturing more complex correlation patterns between nodes. Next, by calculating the correlation between each node, the cooperation strength between employees is quantified, further revealing which nodes (employees) have high influence or criticality in their cooperative relationships. Subsequently, Bernoulli relaxation sampling samples and adjusts edges according to a certain probability, thereby filtering out unimportant or low-correlational connection edges. Finally, by comparing with a preset sampling threshold, connection edges with low correlation are eliminated, forming the final reference key relationship graph.
[0124] Through this in-depth analysis and gradual screening, key groups with high levels of cooperation within the team are identified, redundant or unnecessary connections are removed, and the key groups in the diagram are accurately identified.
[0125] Please see Figure 6In some embodiments, step S105 includes, but is not limited to, steps S601 to S602:
[0126] Step S601: Perform graph convolution on the reference key relationship graph and reference object attribute data through a preset object evaluation network to obtain key object feature information;
[0127] Step S602: Average the key object feature information to obtain the predicted object evaluation vector.
[0128] In some embodiments, the principle of the graph convolution step in step S601 is similar to that of step S103, and will not be repeated here.
[0129] In step S602 of some embodiments, the average calculation is to calculate the average value of all key object feature information to obtain the predicted object evaluation vector.
[0130] Please see Figure 7 In some embodiments, step S106 may include, but is not limited to, steps S701 to S704:
[0131] Step S701: Calculate the divergence loss value based on the evaluation vector of the predicted object to obtain the first loss value;
[0132] Step S702: Calculate the object data for the predicted object evaluation vector to obtain the predicted object evaluation data;
[0133] Step S703: Calculate the mean square error based on the prediction object evaluation data and the reference evaluation data to obtain the second loss value;
[0134] Step S704: Optimize the preset object evaluation network based on the first loss value and the second loss value to obtain the target object evaluation network.
[0135] In steps S701 to S704 of some embodiments, the object data calculation for the predicted object evaluation vector is shown in equation (7):
[0136]
[0137] Among them, g i To evaluate the vector for the predicted object, Evaluate the data for the predicted object.
[0138] The calculation of the first loss value and the second loss value is shown in equation (8):
[0139]
[0140] in, The second loss value, y iFor reference evaluation data.
[0141] The source is the key object feature information obtained by convolving the reference key relationship graph and the reference object attribute data graph. This key object feature information is used to represent the nodes, i.e., the key object feature information of employees. Then, the key object feature information is averaged to obtain the predicted object evaluation vector, which is the mean of all employees. Finally, the predicted object evaluation vector is calculated through the object data to obtain the predicted object evaluation data. In other words, the meaning of the predicted object evaluation data is the average employee evaluation data, which should approximate the project score, i.e., the reference evaluation data.
[0142] β is a preset hyperparameter, p(Hsub_i|Gi)=μ+∑⊙γ, μ is the first k-dimensional vector of gi, representing the mean. ∑ is the last k-dimensional vector of gi, representing the variance, and k is set to 1 / 2 of the total dimensions of gi. γ is noise independently sampled from a Gaussian distribution. r(Hsub_i) is a standard normal distribution.
[0143] The divergence loss value reduces the divergence between predicted and real data by maximizing the similarity between the predicted and reference distributions. This makes the target object evaluation network perform more consistently on different training data, improves the generalization ability of the target object evaluation network, and enables the target object evaluation network to handle a wide variety of stage relationship graphs and a wide variety of object attribute data.
[0144] Please see Figure 8 This application also provides an object recognition device that can implement the above-described object recognition method. The device includes:
[0145] The data acquisition module 801 is used to acquire reference evaluation data of the reference project, reference object attribute data of the reference team, and reference work record data of the reference team; wherein, the reference work record data is the record data of the reference team when completing the reference project.
[0146] Graph construction module 802 is used to construct a graph based on reference work record data to obtain a reference object relationship graph.
[0147] Graph convolution module 803 is used to perform graph convolution on the reference object relationship graph and reference object attribute data through a preset object evaluation network to obtain reference object feature information;
[0148] The first identification module 804 is used to identify key subgraphs of the reference object relationship graph based on the feature information of the reference object, and obtain a reference key relationship graph.
[0149] The object evaluation module 805 is used to evaluate the reference key relationship graph and reference object attribute data through a preset object evaluation network to obtain the predicted object evaluation vector.
[0150] The model optimization module 806 is used to optimize the preset object evaluation network by using the predicted object evaluation vector and the reference object attribute data to obtain the target object evaluation network.
[0151] The second identification module 807 is used to acquire target work record data of the target team, acquire target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and target object attribute data.
[0152] The specific implementation of this object recognition device is basically the same as the specific implementation of the object recognition method described above, and will not be repeated here.
[0153] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described object recognition method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.
[0154] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:
[0155] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.
[0156] The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the object recognition method of the embodiments of this application.
[0157] The input / output interface 903 is used to implement information input and output;
[0158] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).
[0159] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);
[0160] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.
[0161] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the object recognition method described above.
[0162] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0163] The object recognition method, object recognition device, electronic device, and storage medium provided in this application embodiment acquire reference evaluation data of a reference project, reference object attribute data of a reference team, and reference work record data of the reference team, wherein the reference work record data is the record data recorded by the reference team when completing the reference project; construct a graph based on the reference work record data to obtain a reference object relationship graph; perform graph convolution on the reference object relationship graph and reference object attribute data through a preset object evaluation network to obtain reference object feature information; identify key subgraphs of the reference object relationship graph based on the reference object feature information to obtain a reference key relationship graph; evaluate objects based on the reference key relationship graph and reference object attribute data through a preset object evaluation network to obtain a predicted object evaluation vector; optimize the preset object evaluation network based on the predicted object evaluation vector and reference object attribute data to obtain a target object evaluation network; acquire target work record data of the target team, acquire target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and target object attribute data. Therefore, by performing graph convolution on the reference object relationship graph and identifying key subgraphs, the embodiments of this application can effectively evaluate the contributions of team members, thereby improving the accuracy of key team identification.
[0164] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.
[0165] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.
[0166] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0167] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.
[0168] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0169] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.
[0170] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0171] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0172] Furthermore, the functional units in the various embodiments of this application 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 integrated unit can be implemented in hardware or as a software functional unit.
[0173] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, 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 multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0174] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.
Claims
1. An object recognition method, characterized in that, The method includes: Obtain reference evaluation data for the reference project, obtain reference object attribute data for the reference team, and obtain reference work record data for the reference team; wherein, the reference work record data is the record data recorded by the reference team when completing the reference project; A graph is constructed based on the reference work record data to obtain a reference object relationship graph; The reference object feature information is obtained by performing graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network. Based on the feature information of the reference objects, key subgraphs are identified in the relationship graph of the reference objects to obtain a reference key relationship graph; The reference key relationship graph and the reference object attribute data are evaluated by the preset object evaluation network to obtain the predicted object evaluation vector; The target object evaluation network is obtained by optimizing the preset object evaluation network using the predicted object evaluation vector and the reference evaluation data. Obtain target work record data of the target team, obtain target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and the target object attribute data.
2. The method according to claim 1, characterized in that, The reference work record data includes collaboration record data between reference objects; the step of constructing a graph based on the reference work record data to obtain a reference object relationship graph includes: Graph nodes are constructed based on the reference object to obtain relational graph nodes; Based on the cooperation record data between the reference objects, the edges between each node of the relationship graph are constructed to obtain the relationship graph edges; A graph is constructed based on the nodes and edges of the relation graph to obtain a reference original relation graph; Adjacent edge mining is performed on the original reference relationship graph to obtain the reference object relationship graph.
3. The method according to claim 2, characterized in that, The reference original relationship graph includes a first node, a second node, and a third node; the process of performing adjacency edge mining on the reference original relationship graph to obtain the reference object relationship graph includes: If there is a connecting edge between the first node and the second node, and a connecting edge between the second node and the third node, the updated connecting edge is obtained by constructing an edge based on the first node and the third node; The original reference relationship graph is updated based on the updated connection edges to obtain the reference object relationship graph.
4. The method according to claim 1, characterized in that, The step of performing graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network to obtain reference object feature information includes: The number of connecting edges of each node in the reference object relationship graph is counted to obtain a node connecting edge count matrix. The reference object feature information is obtained by performing graph convolution on the reference object relationship graph, the node connection edge number matrix, and the reference object attribute data through the preset object evaluation network.
5. The method according to claim 1, characterized in that, The step of identifying key subgraphs of the reference object relationship graph based on the feature information of the reference object to obtain a reference key relationship graph includes: The reference object feature information is nonlinearly mapped to obtain mapped feature information; Nodes with connecting edges in the reference object relationship graph are obtained to obtain selected node pairs; wherein, the selected node pair includes a first selected node and a second selected node; The correlation degree is calculated based on the mapping feature information corresponding to the first selected node and the mapping feature information corresponding to the second selected node to obtain the node correlation degree between the first selected node and the second selected node. Bernoulli relaxation sampling is performed on the node correlation to obtain the sampling probability; If the sampling probability is less than a preset sampling threshold, the connection edges corresponding to the node correlation degree in the reference object relationship graph are deleted to obtain the reference key relationship graph.
6. The method according to claim 1, characterized in that, The step of evaluating the reference key relationship graph and the reference object attribute data through the preset object evaluation network to obtain the predicted object evaluation vector includes: The key object feature information is obtained by performing graph convolution on the reference key relationship graph and the reference object attribute data through the preset object evaluation network. The key object feature information is averaged to obtain the predicted object evaluation vector.
7. The method according to any one of claims 1 to 6, characterized in that, The step of optimizing the preset object evaluation network using the predicted object evaluation vector and the reference evaluation data to obtain the target object evaluation network includes: The first loss value is obtained by calculating the divergence loss value based on the evaluation vector of the predicted object. The predicted object evaluation vector is used to perform object data calculation to obtain the predicted object evaluation data; The mean squared error is calculated based on the evaluation data of the predicted object and the attribute data of the reference object to obtain the second loss value; The preset object evaluation network is optimized based on the first loss value and the second loss value to obtain the target object evaluation network.
8. An object recognition device, characterized in that, The device includes: The data acquisition module is used to acquire reference evaluation data of the reference project, reference object attribute data of the reference team, and reference work record data of the reference team; wherein, the reference work record data is the record data recorded by the reference team when completing the reference project; The graph construction module is used to construct a graph based on the reference work record data to obtain a reference object relationship graph. The graph convolution module is used to perform graph convolution on the reference object relationship graph and the reference object attribute data through a preset object evaluation network to obtain reference object feature information. The first identification module is used to identify key subgraphs of the reference object relationship graph based on the feature information of the reference object, and obtain a reference key relationship graph. The object evaluation module is used to evaluate the reference key relationship graph and the reference object attribute data through the preset object evaluation network to obtain the predicted object evaluation vector; The model optimization module is used to optimize the preset object evaluation network using the predicted object evaluation vector and the reference object attribute data to obtain the target object evaluation network. The second identification module is used to acquire target work record data of the target team, acquire target object attribute data of the target team, construct a graph based on the target work record data to obtain a target object graph, and identify key subgraphs of the target object graph based on the target object evaluation network and the target object attribute data.
9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the object recognition method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the object recognition method according to any one of claims 1 to 7.