A team optimization matching method and system based on a bidirectional matching mechanism

Abstract

This invention provides a team optimization matching method and system based on a two-way matching mechanism, relating to the field of intelligent collaboration and resource optimization technology. The method involves: uniformly encoding user-end nodes and cloud collaboration spaces; acquiring unstructured business description text and requirement description text, and using a Gaussian embedding model to extract probabilistic node business feature vectors and space requirement feature vectors; collecting user interaction attribute features of user-end nodes and matching constraints of the cloud collaboration space; calculating semantic similarity based on the idea of ​​two-way projection by using Mahalanobis distance and covariance shape differences between two probability distributions; quantifying interaction attribute features into interaction activity using a weighted normalization function; dynamically adjusting the fusion weights of semantic similarity and interaction activity using grey relational analysis to generate a comprehensive matching degree; selecting user-end nodes based on the comprehensive matching degree and creating a cloud resource orchestration interface for the cloud collaboration space, establishing network connections through a software-defined network.

Description

Technical Field

[0001] This invention relates to the field of intelligent collaboration and resource optimization technology, and more specifically, to a team optimization matching method and system based on a two-way matching mechanism. Background Technology

[0002] With the rapid development of cloud computing and distributed collaboration models, team building and optimal matching have become key factors affecting project success. Traditional team matching methods often rely on manual screening or simple keyword-based searches. This approach is not only time-consuming and labor-intensive, but also struggles to accurately capture the deep semantic connections between member capabilities and team needs. In particular, when the descriptive text is ambiguous or expressive, the matching results are often poor.

[0003] In recent years, academia and industry have begun exploring team formation methods based on recommender systems. For example, some studies have used collaborative filtering techniques to recommend members based on users' historical behavior or similarity. However, these methods typically assume that member preferences are static and ignore the dynamic nature of the two-way selection between team members and the team, resulting in recommendations lacking specificity. Other studies have proposed optimization models based on bilateral matching theory, solving matching schemes by constructing 0-1 integer programming or heuristic algorithms. However, these models often only focus on explicit attributes of members (such as skill tags and years of experience), failing to effectively handle the semantic ambiguity and polysemy inherent in unstructured text descriptions. This limits the matching results to the alignment of surface attributes, neglecting the semantic fit between members' actual abilities and team needs.

[0004] In semantic matching, existing technologies typically represent text as deterministic vectors and calculate similarity using cosine similarity or Euclidean distance. This deterministic representation assumes that each text corresponds to a unique semantic point. However, in reality, the same expression may have multiple interpretations, or semantic uncertainty may exist due to the speaker's language habits. For example, a member describing themselves as "skilled in agile development" may encompass various specific practices, while a team's requirement of "needing rapid iteration capabilities" may point to different technology stacks. Deterministic vectors cannot capture this degree of semantic discreteness and the correlation between dimensions, resulting in limited matching accuracy.

[0005] In addition, existing matching schemes, when integrating multi-dimensional indicators, often only consider static matching of skills or experience, failing to fully integrate the interactive attributes (such as communication frequency and feedback speed) exhibited by members in actual collaboration and the team's dynamic expectations for collaboration style.

[0006] In terms of weighted fusion, existing technologies typically employ fixed weights or simple weighted averages, failing to dynamically adjust the contribution ratios of different indicators based on specific matching scenarios. For example, when semantic similarity is unreliable due to data sparsity, fixed weights may still assign it a high weight, affecting the robustness of the overall matching score. This lack of adaptability makes it difficult to guarantee stable matching quality in complex and ever-changing service environments.

[0007] Finally, after matching is complete, achieving an efficient and secure network connection between member nodes and the cloud collaboration space is a pain point in practical applications. Traditional methods require manual network configuration, resulting in high latency and errors, which cannot meet the needs of real-time collaboration.

[0008] In summary, existing team matching technologies have significant shortcomings in semantic modeling, interaction attribute fusion, dynamic weight adjustment, and automatic networking. There is an urgent need for a team optimization matching scheme that can comprehensively consider semantic uncertainty, integrate dynamic interaction behavior, adaptively adjust weights, and achieve automatic network connection. Summary of the Invention

[0009] The purpose of this invention is to provide a team optimization matching method and system based on a two-way matching mechanism to solve the technical problems of inaccurate semantic modeling, ignoring member interaction behavior, lack of adaptability in weight fusion, and inability to automatically form a network after matching in existing team matching methods.

[0010] In view of the above-mentioned technical problems, the present invention provides a team optimization matching method and system based on a two-way matching mechanism.

[0011] In a first aspect, the present invention provides a team optimization matching method based on a two-way matching mechanism, the method comprising:

[0012] Step S1: Unify the coding of the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team.

[0013] Step S2: Obtain the unstructured business description text of the user-end node and the unstructured requirement description text of the cloud collaboration space. Use a Gaussian embedding model based on feature ensemble pre-training to extract features and obtain the node business feature vector and the space requirement feature vector in probabilistic form.

[0014] Step S3: Collect user interaction attribute features of each user terminal node based on the collaborative behavior log, and obtain the matching constraints based on interaction attributes for each cloud collaboration space;

[0015] Step S4: Based on the node business feature vector and spatial demand feature vector, calculate the semantic similarity between the user terminal node and the cloud collaboration space using the idea of ​​bidirectional projection, and quantify the user interaction attribute features into interaction activity.

[0016] Step S5: Use grey relational analysis to dynamically adjust the fusion weights of semantic similarity and interaction activity to generate a comprehensive matching degree;

[0017] Step S6: Select user-end nodes in descending order of overall matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end nodes and the corresponding cloud collaboration spaces.

[0018] Furthermore, the specific implementation process of step S1 includes:

[0019] The cloud service sharing center uses a unified encoding for the user-end nodes that initiate collaboration requests, denoting the i-th user-end node as... To form the user-end node library Where I represents the total number of user-end nodes;

[0020] The cloud service sharing center uses a unified coding system for cloud-based collaboration spaces that provide collaborative resources, denoting the g-th cloud-based collaboration space as... To form a cloud-based collaborative space library Where G represents the total number of cloud collaboration spaces;

[0021] In this system, one user-end node corresponds to one team member, and one cloud collaboration space corresponds to one team.

[0022] Furthermore, the specific implementation process of step S2 includes:

[0023] The cloud service sharing center responds to the registration requests of each user node, obtains the unstructured business description text fed back by the user node, and the unstructured requirement description text pre-configured in the cloud collaboration space;

[0024] A Gaussian embedding model based on feature ensemble pre-training is used to extract features from business description text and requirement description text, including:

[0025] Pre-training is performed by fusing internal sequence features of strokes, structure, and pinyin with sentence features to capture the multi-faceted semantics of business description texts and requirement description texts;

[0026] By fusing the internal sequence features with a Gaussian distribution through correlation Gaussian representation, a node business feature vector in probabilistic form with a center point vector and a covariance matrix, and a spatial demand feature vector in probabilistic form are generated.

[0027] Wherein, the node business feature vector is represented as The spatial demand feature vector is represented as In the formula, Represents the i-th user node The vector of the center point of the core semantics of the business description text in the high-dimensional semantic space. Represents the g-th cloud collaboration space The vector of the center point of the core semantics of the requirement description text in the high-dimensional semantic space. This represents the semantic uncertainty covariance matrix of the service description text of the i-th user node. Let represent the semantic uncertainty covariance matrix of the g-th cloud collaboration space requirement description text. The symbol for Gaussian distribution;

[0028] The dimension of the center point vector is d, and the semantic uncertainty covariance matrix is... A positive definite matrix, wherein the diagonal elements of the positive definite matrix are used to reflect the degree of dispersion of multi-semantic semantics, and the off-diagonal elements reflect the correlation between the dimensions of multi-semantic semantics.

[0029] Furthermore, the specific implementation process of step S3 includes:

[0030] The cloud service sharing center collects user interaction attribute characteristics of each user terminal node based on collaborative behavior logs. These user interaction attribute characteristics include the frequency of communication within the team and the speed of communication feedback.

[0031] In addition, the matching constraints based on interaction attributes of each cloud collaboration space are obtained, including the expected communication frequency range and the expected feedback speed threshold.

[0032] The i-th user node User interaction attribute features are represented as ,in, For the frequency of communication within the team, To improve the speed of communication and feedback;

[0033] The gth cloud collaboration space The matching constraint is expressed as This includes the expected communication frequency threshold. and expected feedback speed threshold ,in, and These are the lower and upper limits of the communication frequency, respectively. and These are the lower and upper limits of the feedback speed, respectively.

[0034] Furthermore, the specific implementation process of step S4 includes:

[0035] Based on node business feature vectors and spatial demand feature vector Based on the concept of bidirectional projection, the i-th user node is evaluated. With the gth cloud collaboration space Semantic similarity:

[0036] ;

[0037] In the formula, For semantic similarity, As a probability distribution distance metric based on the idea of ​​bidirectional projection, the first term measures the difference between the two mean vectors in the sense of Mahalanobis distance (weighted by joint covariance), and the second term measures the difference in shape between the two covariance matrices (by the ratio of their determinants).

[0038] A weighted normalization function is used to quantify user interaction attribute features into interaction activity:

[0039] ;

[0040] In the formula, and Let $\frac{g}{g}$ be the baseline value of the expected communication frequency for the $g$-th cloud collaboration space, taking the median of the expected communication frequency threshold, and $\frac{g}{g}$ be the baseline value of the feedback speed, taking the median of the expected feedback speed threshold. and All are normalized functions. Returning based on how close the frequency of team communication is to the baseline value. Fractions in an interval Returning based on the closeness of the feedback speed to the baseline value of the feedback speed. Fractions in an interval and To preset attribute weights, satisfying It can be set based on historical experience.

[0041] Furthermore, the specific implementation process of step S5 includes:

[0042] Grey relational analysis is used to dynamically adjust the fusion weights of semantic similarity and interaction activity:

[0043] Based on semantic similarity and interaction positivity To compare sequences, a reference sequence of the same dimension is constructed using an ideal value of 1, and the grey relational coefficient is used to evaluate semantic similarity. And the grey relational coefficient of interaction positivity :

[0044] ;

[0045] Determine dynamic weights based on the distribution of grey relational coefficients. and :

[0046] ;

[0047] Generate overall matching score:

[0048] ;

[0049] In the formula, This is the grayscale resolution coefficient (usually preset to 0.5). For user-side nodes cloud-based collaborative spaces The overall matching degree.

[0050] Furthermore, the specific implementation process of step S6 includes:

[0051] The cloud service sharing center selects user-end nodes based on the number of members configured within the team, in descending order of overall matching degree;

[0052] After selecting the user-end node, the cloud service sharing center creates a cloud resource orchestration interface for each cloud collaboration space, and determines the number of cloud resource orchestration interfaces for the corresponding cloud collaboration space of the team based on the number of members configured in the team.

[0053] After determining the number of cloud resource orchestration interfaces in the cloud collaboration space, the cloud service sharing center establishes cross-domain VPC peering connections or dedicated network channels between user-end nodes and the virtual private cloud to which the corresponding cloud collaboration space belongs through a software-defined network controller.

[0054] Secondly, the present invention also provides a team optimization matching system based on a two-way matching mechanism, the system comprising:

[0055] A readable storage medium is used to carry a computer program that can be executed by a processor to implement the steps of the team optimization matching method based on a two-way matching mechanism.

[0056] Furthermore, the system also includes:

[0057] One or more processors;

[0058] A cloud service sharing center is used to schedule one or more computer programs, which, when executed by one or more processors, enable the cloud service sharing center to implement the steps of the team optimization matching method based on a two-way matching mechanism.

[0059] Furthermore, the cloud service sharing center also includes:

[0060] The coding module is used to uniformly code the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and to build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team.

[0061] The feature extraction module is used to obtain unstructured business description text of user-end nodes and unstructured requirement description text of cloud collaboration space, and uses a Gaussian embedding model based on feature integration pre-training to extract features, resulting in probabilistic node business feature vectors and probabilistic space requirement feature vectors.

[0062] The attribute collection module collects user interaction attribute features from each user terminal node based on the collaborative behavior log, and obtains matching constraints based on interaction attributes for each cloud collaboration space.

[0063] The similarity calculation module, based on the node business feature vector and spatial demand feature vector, uses the idea of ​​bidirectional projection to calculate the semantic similarity between the user terminal node and the cloud collaboration space, and at the same time quantifies the user interaction attribute features into interaction activity.

[0064] The matching degree generation module is used to dynamically adjust the fusion weights of semantic similarity and interaction activity using grey relational analysis to generate a comprehensive matching degree.

[0065] The network connection module is used to select user-end nodes in descending order of comprehensive matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end node and the corresponding cloud collaboration space.

[0066] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0067] 1. This invention introduces a Gaussian embedding model based on feature ensemble pre-training to map unstructured business description text and requirement description text into probabilistic feature vectors containing centroid vectors and covariance matrices. Unlike existing technologies that represent text as deterministic vectors, this invention quantifies the semantic discreteness and inter-dimensional correlations through the covariance matrix, effectively capturing the inherent uncertainty and ambiguity in natural language expression. When the business description of a user-end node or the requirement description of a cloud-based collaborative space contains ambiguous expressions, this invention can preserve its multi-dimensional semantic information through a probability distribution. In subsequent similarity calculations, it not only compares the closeness of core semantics (centroid vectors) but also measures the morphological matching degree of the semantic distribution through the covariance matrix, thereby improving matching accuracy in semantically ambiguous scenarios and avoiding misjudgments caused by hard keyword matching.

[0068] 2. This invention constructs a probability distribution distance metric based on the concept of bidirectional projection to evaluate the semantic similarity between node business characteristics and spatial demand characteristics. By jointly calculating the Mahalanobis distance between the mean vectors (center point vectors) using the weighted average of the covariance matrices, and simultaneously introducing the determinant ratio to measure the difference in covariance shape between the two probability distributions, a comprehensive quantitative assessment of the similarity between two Gaussian distributions is achieved. Compared to traditional cosine similarity or Euclidean distance, this bidirectional projection metric can more precisely reflect the degree of distribution alignment in the semantic space. Even when the mean vectors are close, differences in covariance shape can identify differences in semantic connotation, providing a more reliable semantic foundation for subsequent comprehensive matching.

[0069] 3. This invention generates an interaction activity index by collecting user interaction attribute characteristics (including the frequency and speed of communication within the team) from user-end nodes and quantitatively matching them with the expected interaction constraints of the cloud-based collaboration space. Unlike existing technologies that only focus on static skill matching, this invention incorporates members' dynamic collaborative behaviors into the matching considerations. By normalizing and comparing actual interaction behaviors with the team's expected interaction patterns, this invention can identify members who not only have matching skills but also whose collaboration styles align with the team's ecosystem, thereby helping to improve the overall smoothness of team collaboration and member satisfaction.

[0070] 4. This invention employs grey relational analysis to dynamically adjust the fusion weights of semantic similarity and interaction positivity to generate a comprehensive matching degree. In grey relational analysis, using an ideal value of 1 as a reference sequence, the grey relational coefficients of semantic similarity and interaction positivity with the ideal value are calculated. This objectively assesses the performance of the two dimensions in the current matching scenario and dynamically allocates fusion weights accordingly. The dynamic weighting mechanism based on data-driven principles and grey system theory avoids the problem of insufficient adaptability of fixed weights in complex and changing environments. When the matching degree of a certain dimension deviates from the ideal value due to data noise or special circumstances, grey relational analysis can automatically reduce its weight contribution, enhancing the robustness and generalization ability of the comprehensive matching degree, ensuring stable and reliable matching results in different application scenarios.

[0071] 5. After matching is complete, this invention establishes a cross-domain VPC peering connection or dedicated network channel between the user-end node and the virtual private cloud to which the cloud collaboration space belongs through a software-defined network controller. This achieves a fully automated closed loop from "matching decision" to "network connection". Based on the number of members configured within the team, the system automatically creates a corresponding number of cloud resource orchestration interfaces and dynamically establishes a secure network channel based on the matching results, ensuring that selected team members can access the cloud collaboration environment instantly and securely, effectively improving deployment efficiency and collaboration experience after team formation.

[0072] The above description is merely an overview of the technical solution of the present invention. To better understand the technical means of the present invention and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of the present invention more apparent, specific embodiments of the present invention are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily apparent from the following description. Attached Figure Description

[0073] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0074] Figure 1 This is a schematic diagram illustrating the steps of a team optimization matching method based on a two-way matching mechanism according to the present invention.

[0075] Figure 2 This is a schematic diagram of the structure of a team optimization matching system based on a two-way matching mechanism according to the present invention. Detailed Implementation

[0076] This invention provides a team optimization matching method and system based on a bidirectional matching mechanism, solving the technical problems of inaccurate semantic modeling, neglect of member interaction behavior, lack of adaptability in weight fusion, and inability to automatically form a network after matching in existing team matching methods. By introducing probabilistic feature vectors to model semantic uncertainty, employing bidirectional projection to calculate semantic similarity, dynamically adjusting fusion weights using grey relational analysis, and establishing software-defined network connections, the invention achieves the technical effects of improving matching accuracy, enhancing team collaboration adaptability, improving matching robustness, and realizing automated network formation.

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, and not all of them.

[0078] Example 1, please refer to Figure 1 This paper provides a team optimization matching method based on a two-way matching mechanism, which includes:

[0079] Step S1: Unify the coding of the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team.

[0080] For example, the cloud service sharing center uses a unified encoding for the user-end nodes that initiate collaboration requests, denoting the i-th user-end node as... To form the user-end node library Where I represents the total number of user-end nodes;

[0081] The cloud service sharing center uses a unified coding system for cloud-based collaboration spaces that provide collaborative resources, denoting the g-th cloud-based collaboration space as... To form a cloud-based collaborative space library Where G represents the total number of cloud collaboration spaces;

[0082] In this system, one user-end node corresponds to one team member, and one cloud collaboration space corresponds to one team.

[0083] Step S2: Obtain the unstructured business description text of the user-end node and the unstructured requirement description text of the cloud collaboration space. Use a Gaussian embedding model based on feature ensemble pre-training to extract features and obtain the node business feature vector and the space requirement feature vector in probabilistic form.

[0084] For example, in response to the registration requests of each user node, the cloud service sharing center obtains the unstructured business description text fed back by the user node, and the unstructured requirement description text pre-configured in the cloud collaboration space;

[0085] A Gaussian embedding model based on feature ensemble pre-training is used to extract features from business description text and requirement description text, including:

[0086] Pre-training is performed by fusing internal sequence features of strokes, structure, and pinyin with sentence features to capture the multi-faceted semantics of business description texts and requirement description texts;

[0087] By fusing the internal sequence features with a Gaussian distribution through correlation Gaussian representation, a node business feature vector in probabilistic form with a center point vector and a covariance matrix, and a spatial demand feature vector in probabilistic form are generated.

[0088] Wherein, the node business feature vector is represented as The spatial demand feature vector is represented as In the formula, Represents the i-th user node The vector of the center point of the core semantics of the business description text in the high-dimensional semantic space. Represents the g-th cloud collaboration space The vector of the center point of the core semantics of the requirement description text in the high-dimensional semantic space. This represents the semantic uncertainty covariance matrix of the service description text of the i-th user node. Let represent the semantic uncertainty covariance matrix of the g-th cloud collaboration space requirement description text. The symbol for Gaussian distribution;

[0089] The dimension of the center point vector is d, and the semantic uncertainty covariance matrix is... A positive definite matrix, wherein the diagonal elements of the positive definite matrix are used to reflect the degree of dispersion of multi-semantic semantics, and the off-diagonal elements reflect the correlation between the dimensions of multi-semantic semantics.

[0090] Step S3: Collect user interaction attribute features of each user terminal node based on the collaborative behavior log, and obtain the matching constraints based on interaction attributes for each cloud collaboration space;

[0091] For example, the cloud service sharing center collects user interaction attribute characteristics of each user terminal node based on collaborative behavior logs. The user interaction attribute characteristics include the frequency of communication within the team and the speed of communication feedback.

[0092] In addition, the matching constraints based on interaction attributes of each cloud collaboration space are obtained, including the expected communication frequency range and the expected feedback speed threshold.

[0093] The i-th user node User interaction attribute features are represented as ,in, For the frequency of communication within the team, To improve the speed of communication and feedback;

[0094] The gth cloud collaboration space The matching constraint is expressed as This includes the expected communication frequency threshold. and expected feedback speed threshold ,in, and These are the lower and upper limits of the communication frequency, respectively. and These are the lower and upper limits of the feedback speed, respectively.

[0095] Step S4: Based on the node business feature vector and spatial demand feature vector, calculate the semantic similarity between the user terminal node and the cloud collaboration space using the idea of ​​bidirectional projection, and quantify the user interaction attribute features into interaction activity.

[0096] For example, based on node business feature vectors and spatial demand feature vector Based on the concept of bidirectional projection, the i-th user node is evaluated. With the gth cloud collaboration space Semantic similarity:

[0097] ;

[0098] In the formula, For semantic similarity, It is a distance metric for probability distributions based on the idea of ​​bidirectional projection;

[0099] A weighted normalization function is used to quantify user interaction attribute features into interaction activity:

[0100] ;

[0101] In the formula, and Let $\frac{g}{g}$ be the baseline value of the expected communication frequency for the $g$-th cloud collaboration space, taking the median of the expected communication frequency threshold, and $\frac{g}{g}$ be the baseline value of the feedback speed, taking the median of the expected feedback speed threshold. and All are normalized functions. Returning based on how close the frequency of team communication is to the baseline value. Fractions in an interval Returning based on the closeness of the feedback speed to the baseline value of the feedback speed. Fractions in an interval and To preset attribute weights, satisfying .

[0102] Step S5: Use grey relational analysis to dynamically adjust the fusion weights of semantic similarity and interaction activity to generate a comprehensive matching degree;

[0103] For example, grey relational analysis is used to dynamically adjust the fusion weights of semantic similarity and interaction activity:

[0104] Based on semantic similarity and interaction positivity To compare sequences, a reference sequence of the same dimension is constructed using an ideal value of 1, and the grey relational coefficient is used to evaluate semantic similarity. And the grey relational coefficient of interaction positivity :

[0105] ;

[0106] Determine dynamic weights based on the distribution of grey relational coefficients. and :

[0107] ;

[0108] Generate overall matching score:

[0109] ;

[0110] In the formula, The gray resolution coefficient is... For user-side nodes cloud-based collaborative spaces The overall matching degree.

[0111] Step S6: Select user-end nodes in descending order of comprehensive matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end node and the corresponding cloud collaboration space.

[0112] For example, the cloud service sharing center selects user-end nodes in descending order of overall matching degree based on the number of members configured within the team;

[0113] After selecting the user-end node, the cloud service sharing center creates a cloud resource orchestration interface for each cloud collaboration space, and determines the number of cloud resource orchestration interfaces for the corresponding cloud collaboration space of the team based on the number of members configured in the team.

[0114] After determining the number of cloud resource orchestration interfaces in the cloud collaboration space, the cloud service sharing center establishes cross-domain VPC peering connections or dedicated network channels between user-end nodes and the virtual private cloud to which the corresponding cloud collaboration space belongs through a software-defined network controller.

[0115] For example, a technology company plans to launch three cloud-based collaborative projects: Project A (development of an intelligent customer service system), Project B (building a big data analytics platform), and Project C (reconstruction of the mobile application front-end). The cloud service sharing center receives collaboration requests from 15 developers (client nodes C1 to C15) and resource requests from three project teams (cloud collaboration spaces S1, S2, and S3).

[0116] The cloud service sharing center selects client nodes based on the number of members and role requirements configured in each cloud collaboration space, ranking them from highest to lowest overall match. For S1 (Project A), two backend developers, one frontend developer, and one algorithm developer are required. All nodes applying for backend roles are sorted according to their overall match with S1, and the top two are selected, assuming C1 and C4 are chosen. Among the frontend nodes, C3 has the highest match, and among the algorithm nodes, C7 has the highest match. The selection process is similar for S2 and S3. After selection, the cloud service sharing center creates a cloud resource orchestration interface for each cloud collaboration space. Based on the four members configured for S1, four interfaces are created. Subsequently, through a software-defined network controller (SDN Controller), a cross-domain VPC peering connection is established between the selected client nodes C1, C4, C3, and C7 and the Virtual Private Cloud (VPC) to which S1 belongs. Each member's local development environment is interconnected with S1's cloud code repository, CI / CD pipeline, collaboration tools, etc., enabling team members to collaborate instantly.

[0117] Example 2, please refer to Figure 2 This paper provides a team optimization matching system based on a two-way matching mechanism, which includes:

[0118] A readable storage medium for carrying a computer program that can be executed by a processor to implement the steps of a team optimization matching method based on a bidirectional matching mechanism according to Embodiment 1 above;

[0119] One or more processors;

[0120] A cloud service sharing center is used to schedule one or more computer programs, which, when executed by one or more processors, enable the cloud service sharing center to implement the steps of the team optimization matching method based on a two-way matching mechanism.

[0121] The cloud service sharing center also includes:

[0122] The coding module is used to uniformly code the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and to build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team.

[0123] The feature extraction module is used to obtain unstructured business description text of user-end nodes and unstructured requirement description text of cloud collaboration space, and uses a Gaussian embedding model based on feature integration pre-training to extract features, resulting in probabilistic node business feature vectors and probabilistic space requirement feature vectors.

[0124] The attribute collection module collects user interaction attribute features from each user terminal node based on the collaborative behavior log, and obtains matching constraints based on interaction attributes for each cloud collaboration space.

[0125] The similarity calculation module, based on the node business feature vector and spatial demand feature vector, uses the idea of ​​bidirectional projection to calculate the semantic similarity between the user terminal node and the cloud collaboration space, and at the same time quantifies the user interaction attribute features into interaction activity.

[0126] The matching degree generation module is used to dynamically adjust the fusion weights of semantic similarity and interaction activity using grey relational analysis to generate a comprehensive matching degree.

[0127] The network connection module is used to select user-end nodes in descending order of comprehensive matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end node and the corresponding cloud collaboration space.

[0128] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. The team optimization matching method and specific examples based on a two-way matching mechanism in Embodiment 1 described above are also applicable to the team optimization matching system based on a two-way matching mechanism in this embodiment. Through the foregoing detailed description of the team optimization matching method based on a two-way matching mechanism, those skilled in the art can clearly understand the team optimization matching system based on a two-way matching mechanism in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As for the system disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be referred to in the method section.

[0129] In the several embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0130] 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. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0131] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0132] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory, random access memory, magnetic disks, or optical disks.

[0133] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0134] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0135] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. If such modifications and variations fall within the scope of this invention and its equivalents, then this invention is also intended to include such modifications and variations.

Claims

1. A team optimization matching method based on a two-way matching mechanism, characterized in that, The method includes the following steps: Step S1: Unify the coding of the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team. Step S2: Obtain the unstructured business description text of the user-end node and the unstructured requirement description text of the cloud collaboration space. Use a Gaussian embedding model based on feature ensemble pre-training to extract features and obtain the node business feature vector and the space requirement feature vector in probabilistic form. Step S3: Collect user interaction attribute features of each user terminal node based on the collaborative behavior log, and obtain the matching constraints based on interaction attributes for each cloud collaboration space; Step S4: Based on the node business feature vector and spatial demand feature vector, calculate the semantic similarity between the user terminal node and the cloud collaboration space using the idea of ​​bidirectional projection, and quantify the user interaction attribute features into interaction activity. Step S5: Use grey relational analysis to dynamically adjust the fusion weights of semantic similarity and interaction activity to generate a comprehensive matching degree; Step S6: Select user-end nodes in descending order of comprehensive matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end node and the corresponding cloud collaboration space. The specific implementation process of step S2 includes: The cloud service sharing center responds to the registration requests of each user node, obtains the unstructured business description text fed back by the user node, and the unstructured requirement description text pre-configured in the cloud collaboration space; A Gaussian embedding model based on feature ensemble pre-training is used to extract features from business description text and requirement description text, including: Pre-training is performed by fusing internal sequence features of strokes, structure, and pinyin with sentence features to capture the multi-faceted semantics of business description texts and requirement description texts; By fusing the internal sequence features with a Gaussian distribution through correlation Gaussian representation, a node business feature vector in probabilistic form with a center point vector and a covariance matrix, and a spatial demand feature vector in probabilistic form are generated. Wherein, the node business feature vector is represented as The spatial demand feature vector is represented as In the formula, Represents the i-th user node The vector of the center point of the core semantics of the business description text in the high-dimensional semantic space. Represents the g-th cloud collaboration space The vector of the center point of the core semantics of the requirement description text in the high-dimensional semantic space. This represents the semantic uncertainty covariance matrix of the service description text of the i-th user node. Let represent the semantic uncertainty covariance matrix of the g-th cloud collaboration space requirement description text. The symbol for Gaussian distribution; The dimension of the center point vector is d, and the semantic uncertainty covariance matrix is... A positive definite matrix, wherein the diagonal elements of the positive definite matrix are used to reflect the degree of dispersion of multi-semantic semantics, and the off-diagonal elements reflect the correlation between the dimensions of multi-semantic semantics; The specific implementation process of step S4 includes: Based on node business feature vectors and spatial demand feature vector Based on the concept of bidirectional projection, the i-th user node is evaluated. With the gth cloud collaboration space Semantic similarity: ； In the formula, For semantic similarity, It is a distance metric for probability distributions based on the idea of ​​bidirectional projection; A weighted normalization function is used to quantify user interaction attribute features into interaction activity: ； In the formula, Let be the expected baseline value for communication frequency in the g-th cloud collaboration space, where the baseline value is taken as the median of the expected communication frequency thresholds. Let be the expected baseline value for the feedback speed of the g-th cloud collaboration space, where the baseline value is taken as the median of the expected feedback speed threshold. and All are normalized functions. Returning based on how close the frequency of team communication is to the baseline value. Fractions in an interval Returning based on the closeness of the feedback speed to the baseline value of the feedback speed. Fractions in an interval and To preset attribute weights, satisfying .

2. The team optimization matching method based on a two-way matching mechanism according to claim 1, characterized in that, The specific implementation process of step S1 includes: The cloud service sharing center uses a unified encoding for the user-end nodes that initiate collaboration requests, denoting the i-th user-end node as... To form the user-end node library Where I represents the total number of user-end nodes; The cloud service sharing center uses a unified coding system for cloud-based collaboration spaces that provide collaborative resources, denoting the g-th cloud-based collaboration space as... To form a cloud-based collaborative space library Where G represents the total number of cloud collaboration spaces; In this system, one user-end node corresponds to one team member, and one cloud collaboration space corresponds to one team.

3. The team optimization matching method based on a two-way matching mechanism according to claim 1, characterized in that, The specific implementation process of step S3 includes: The cloud service sharing center collects user interaction attribute characteristics of each user terminal node based on collaborative behavior logs. These user interaction attribute characteristics include the frequency of communication within the team and the speed of communication feedback. In addition, the matching constraints based on interaction attributes of each cloud collaboration space are obtained, including the expected communication frequency range and the expected feedback speed threshold. The i-th user node User interaction attribute features are represented as ,in, For the frequency of communication within the team, To improve the speed of communication and feedback; The gth cloud collaboration space The matching constraint is expressed as This includes the expected communication frequency threshold. and expected feedback speed threshold ,in, and These are the lower and upper limits of the communication frequency, respectively. and These are the lower and upper limits of the feedback speed, respectively.

4. The team optimization matching method based on a two-way matching mechanism according to claim 1, characterized in that, The specific implementation process of step S5 includes: Grey relational analysis is used to dynamically adjust the fusion weights of semantic similarity and interaction activity: Based on semantic similarity and interaction positivity To compare sequences, a reference sequence of the same dimension is constructed using an ideal value of 1, and the grey relational coefficient is used to evaluate semantic similarity. And the grey relational coefficient of interaction positivity : ； Determine dynamic weights based on the distribution of grey relational coefficients. and : ； Generate overall matching score: ； In the formula, The gray resolution coefficient is... For user-side nodes cloud-based collaboration space The overall matching degree.

5. The team optimization matching method based on a two-way matching mechanism according to claim 1, characterized in that, The specific implementation process of step S6 includes: The cloud service sharing center selects user-end nodes based on the number of members configured within the team, in descending order of overall matching degree; After selecting the user-end node, the cloud service sharing center creates a cloud resource orchestration interface for each cloud collaboration space, and determines the number of cloud resource orchestration interfaces for the corresponding cloud collaboration space of the team based on the number of members configured in the team. After determining the number of cloud resource orchestration interfaces for the cloud collaboration space, the cloud service sharing center establishes cross-domain VPC peering connections or dedicated network channels between user-end nodes and the virtual private cloud to which the corresponding cloud collaboration space belongs through a software-defined network controller.

6. A team optimization matching system based on a two-way matching mechanism, comprising a built-in computer-readable storage medium, characterized in that, The readable storage medium is used to carry a computer program that can be executed by a processor to implement the steps of a team optimization matching method based on a bidirectional matching mechanism as described in any one of claims 1 to 5.

7. A team optimization matching system based on a two-way matching mechanism according to claim 6, characterized in that, include: One or more processors; A cloud service sharing center is used to schedule one or more computer programs, which, when executed by one or more processors, enable the cloud service sharing center to implement the steps of the team optimization matching method based on a two-way matching mechanism.

8. A team optimization matching system based on a two-way matching mechanism according to claim 7, characterized in that, The cloud service sharing center also includes: The coding module is used to uniformly code the user-end nodes that initiate collaboration requests and the cloud collaboration spaces that provide collaboration resources, and to build a user-end node library and a cloud collaboration space library, where each user-end node corresponds to a team member and each cloud collaboration space corresponds to a team. The feature extraction module is used to obtain unstructured business description text of user-end nodes and unstructured requirement description text of cloud collaboration space, and uses a Gaussian embedding model based on feature integration pre-training to extract features, resulting in probabilistic node business feature vectors and probabilistic space requirement feature vectors. The attribute collection module collects user interaction attribute features from each user terminal node based on the collaborative behavior log, and obtains matching constraints based on interaction attributes for each cloud collaboration space. The similarity calculation module, based on the node business feature vector and spatial demand feature vector, uses the idea of ​​bidirectional projection to calculate the semantic similarity between the user terminal node and the cloud collaboration space, and at the same time quantifies the user interaction attribute features into interaction activity. The matching degree generation module is used to dynamically adjust the fusion weights of semantic similarity and interaction activity using grey relational analysis to generate a comprehensive matching degree. The network connection module is used to select user-end nodes in descending order of comprehensive matching degree, create a cloud resource orchestration interface for each cloud collaboration space, and establish a network connection between the user-end node and the corresponding cloud collaboration space.

Images