Expert matching and management method and system based on modern database technology

Abstract

The present invention discloses an expert matching and management method and system based on modern database technology, belonging to the technical field of database analysis application. The present invention adds the comprehensive analysis capability of expert multi-dimensional information on the basis of the functions of basic information management and simple classification query, and solves the deficiencies of the existing system in multi-dimensional matching, dynamic information updating and security assurance. Through the expert matching and management of this solution, not only the most suitable experts can be quickly screened and matched, but also the real-time updating and secure storage of expert information can be ensured, thereby improving the efficiency and success rate of activity arrangements.

Description

Technical Field

[0001] The present invention belongs to the technical field of database analysis applications, and in particular relates to an expert matching and management method and system based on modern database technology. Background Art

[0002] Currently, the internet industry is developing rapidly against the backdrop of globalization and informatization. Since the widespread adoption of the internet and the increasing maturity of the market, the storage and organization of content and the collection of user behavioral data provided by products have become fundamental functions for a wide range of applications across various industries. The ever-changing internet, its ever-improving technology, and the influx of skilled professionals have also led to the development and upgrading of major search application platforms both domestically and internationally. Therefore, the rational utilization of expert resources is crucial to the success of various organizations and activities. However, traditional expert information management systems often rely on manual record-keeping and simple database queries, making it difficult to meet the requirements of multi-dimensional matching and rapid response. Existing technologies primarily focus on the management of basic information and simple categorized queries, lacking the ability to comprehensively analyze multi-dimensional expert information. Summary of the Invention

[0003] In response to the above-mentioned deficiencies in the prior art, this application provides an expert matching and management method and system based on modern database technology.

[0004] In the first aspect, the present application proposes an expert matching and management method based on modern database technology, comprising the following steps:

[0005] Acquiring expert data, and storing the expert data using database technology to obtain an expert database;

[0006] Obtain expert information from the expert database according to preset expert screening conditions, and perform preliminary fuzzy matching calculations on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list;

[0007] The preliminary expert matching list is weighted using a multi-level matching model, and the expert matching list after weighting is calculated using a pre-built precise matching algorithm to obtain a secondary screening expert list;

[0008] Obtaining demand information input by the user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy;

[0009] An expert information management platform is constructed to update the expert information in the final matching expert list in real time.

[0010] In some optional implementations of some embodiments, acquiring expert data and storing the expert data using database technology to obtain an expert database includes:

[0011] Select the appropriate database type based on the structured nature and query complexity of the expert data, including relational databases and NoSQL databases;

[0012] When the database type is a relational database, a table structure is designed for the relational database; when the database type is a NoSQL database, a document structure is designed for the NoSQL database;

[0013] Import the expert data according to the designed document structure / table structure;

[0014] A query statement is created for the relational database or NoSQL database according to the multi-dimensional matching requirements to ensure the query efficiency and data security of the expert data and complete the construction of the expert database.

[0015] In some optional implementations of some embodiments, obtaining expert information from the expert database according to preset expert screening conditions includes:

[0016] Obtain corresponding expert information based on the industry, nature of the unit, honorary title, technical field or expertise, professional title level, score, and past activity evaluation as the preset expert screening conditions;

[0017] The expert information corresponding to the industry is obtained as follows: the industry field in which the expert is located;

[0018] The expert information corresponding to the unit nature is obtained as follows: the type of unit to which the expert belongs;

[0019] The expert information corresponding to the honorary title is: the honors received by the expert;

[0020] The expert information corresponding to the technical field or expertise is obtained as follows: the expert's professional field or technical expertise;

[0021] The expert information corresponding to the professional title level is obtained as follows: the professional title of the expert;

[0022] The expert information corresponding to the ratings obtained is: the expert's rating based on past activities and peer reviews;

[0023] The expert information corresponding to the past activity evaluation is obtained as follows: the expert's performance evaluation in the activities he participated in the past.

[0024] In some optional implementations of some embodiments, performing a preliminary fuzzy matching calculation on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list includes:

[0025] Construct a weighted scoring model, the formula is:

[0026]

[0027] Assume that the initial weights of each expert information are , the similarity between expert information and project requirements is ;

[0028] Calculate the similarity scores of all experts based on expert information After that, set the threshold τ to retain the similarity score Experts with numbers greater than τ are used as the preliminary expert matching list.

[0029] In some optional implementations of some embodiments, the weighting of the preliminary expert matching list using a multi-level matching model includes:

[0030] Determine the relative importance of each expert information in the preliminary expert matching list through the hierarchical analysis method, and determine the weight of each level through expert scoring and consistency test ;

[0031] Among them, the first level For basic information: nature of industry and unit;

[0032] Second level Professional ability: technical field or expertise and professional title level;

[0033] The third level For historical performance: honorary titles, ratings, and evaluations of past activities.

[0034] In some optional implementations of some embodiments, the weighted expert matching list is calculated using a pre-built exact matching algorithm to obtain a secondary screening expert list, including:

[0035] For each expert information after hierarchical assignment, calculate the similarity between the expert information and the requirements:

[0036]

[0037] in, Indicates the The similarity of the levels, represents the total amount of expert information in each level, Indicates the Expert information, and Represents expert information and demand in The eigenvalues of the expert information;

[0038] The similarity of each level is weighted by the hierarchical structure to obtain the comprehensive similarity :

[0039]

[0040] in, represents the comprehensive similarity, represents the weight of each level, For the Based on the similarity of the levels, a preset number of experts are selected in descending order according to the comprehensive similarity to enter the secondary screening expert list.

[0041] In some optional implementations of some embodiments, obtaining demand information input by a user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy, includes:

[0042] The demand information input by the user is segmented into sentences or words as basic processing units. The basic processing units are used as nodes to build a node connection graph. The similarity between the basic processing units is used as the weight of the connection edge. The TextRank value of the basic processing unit is calculated through loop iteration, and then the selection is made based on the TextRank value. The calculation formula is:

[0043]

[0044] in the formula Points to the node The node set of is a node The set of all nodes pointed to: is a node With node The weight of the edge between them; Is the damping coefficient, which represents the probability of the current node jumping to any other node and ensures that the weight can be stably transferred to convergence. Finally, the weight of each word is calculated and sorted. According to the demand, the first N words in the sequence are selected as N user demand keywords and output. Indicates the nodes;

[0045] Based on the short text characteristics of the user demand keywords and the secondary screening expert list, the Bert-TextCNN combined model is used to perform text vectorization processing to obtain a demand-expert vector set;

[0046] The final list of matching experts is determined from the clustering results through the demand expert matching strategy, and the steps of the demand expert matching strategy are as follows:

[0047] Step 1: Input the demand-expert vector set and calculate the similarity distance between each vector;

[0048] Step 2: Determine the thresholds of local density and relative distance, calculate the local density and relative distance of any vector, and use the point where the local density and relative distance of the vector are greater than the threshold as the initial cluster center point;

[0049] Step 3: Complete the division of the data in the set according to the initial cluster center point, and initialize the membership matrix and membership weight matrix;

[0050] Step 4: Iteratively update the cluster clusters according to the membership matrix and the membership weight matrix to obtain the clustering results;

[0051] Step 5: Obtain a list of experts that belong to the same cluster as the demand information from the clustering results, take the intersection of the expert list and the secondary screening expert list, and finally randomly sample the candidates in the intersection. According to the sampling results, dynamically obtain any number of experts from the candidates as the final matching expert list.

[0052] In the second aspect, this application proposes an expert matching and management system based on modern database technology, including an expert database construction module, a preliminary matching module, a secondary screening module, a final matching module and an information update module;

[0053] The expert database construction module is used to obtain expert data and store the expert data using database technology to obtain an expert database;

[0054] The preliminary matching module is used to obtain expert information from the expert database according to preset expert screening conditions, and perform preliminary fuzzy matching calculations on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list;

[0055] The secondary screening module is used to assign weights to the preliminary expert matching list through a multi-level matching model, and calculate the expert matching list after weight assignment using a pre-built accurate matching algorithm to obtain a secondary screening expert list;

[0056] The final matching module is used to obtain demand information input by the user, perform keyword extraction on the demand information to obtain user demand keywords, use a text vectorization model to vectorize the user demand keywords and the secondary screening expert list to obtain a demand-expert vector set, use the demand-expert vector set as clustering input data to perform clustering iterative optimization to obtain a clustering result, and determine the final matching expert list from the clustering result through the demand-expert matching strategy;

[0057] The information updating module is used to construct an expert information management platform to update the expert information in the final matching expert list in real time.

[0058] In a third aspect, the present application proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0059] In a fourth aspect, the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0060] Beneficial effects of the present invention:

[0061] On the basis of the functions of basic information management and simple classification query, the comprehensive analysis capability of multi-dimensional information of experts is added to solve the shortcomings of the existing system in multi-dimensional matching, dynamic information update and security assurance. Through the expert matching and management of this solution, not only can the most suitable experts be quickly screened and matched, but also the real-time update and secure storage of expert information can be ensured, thereby improving the efficiency and success rate of event arrangements. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is the overall flow chart of the present invention.

[0063] Figure 2 This is the system schematic diagram of this device. DETAILED DESCRIPTION

[0064] The following describes exemplary embodiments of the present disclosure in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein; rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0065] On the one hand, this application proposes an expert matching and management method based on modern database technology, such as Figure 1 As shown, the following steps are included:

[0066] S100: Acquire expert data, and store the expert data using database technology to obtain an expert database;

[0067] In some optional implementations of some embodiments, acquiring expert data and storing the expert data using database technology to obtain an expert database includes:

[0068] Select the appropriate database type based on the structured nature and query complexity of the expert data, including relational databases and NoSQL databases;

[0069] When the database type is a relational database, a table structure is designed for the relational database; when the database type is a NoSQL database, a document structure is designed for the NoSQL database;

[0070] Import the expert data according to the designed document structure / table structure;

[0071] A query statement is created for the relational database or NoSQL database according to the multi-dimensional matching requirements to ensure the query efficiency and data security of the expert data and complete the construction of the expert database.

[0072] When designing an expert database, the choice of storage structure is crucial. First, relational databases (such as MySQL and PostgreSQL), with their structured data storage, are well-suited for handling data with fixed schemas. In such databases, we can design a database architecture consisting of multiple tables, each corresponding to a different attribute of an expert. For example, an "Expert Information" table might contain fields such as "Expert ID," "Name," "Industry," and "Type of Organization." By using foreign key relationships, we can link these tables to facilitate complex data queries and updates.

[0073] On the other hand, NoSQL databases (such as MongoDB and Cassandra) offer a more flexible data model suitable for handling unstructured or semi-structured data. In NoSQL databases, we can use document storage, where each document contains all the information about an expert, stored as key-value pairs. This approach allows us to flexibly add or modify fields without changing the overall database structure.

[0074] In order to implement the multi-dimensional expert matching algorithm, the database needs to support complex query operations. This can be achieved by combining indexing technology and query optimization. In a relational database, SQL statements can be used for complex queries, for example:

[0075]

[0076] Furthermore, to ensure the security and privacy of data storage, data encryption and access control policies must be implemented. At the database level, data encryption can be enabled to ensure data confidentiality during storage and transmission. Furthermore, roles and permissions can be set to restrict users who can access and manipulate data. These measures can be implemented through the database's built-in security features or supplemented by application-layer security mechanisms.

[0077] S200: Obtaining expert information from the expert database according to preset expert screening conditions, performing preliminary fuzzy matching calculations on the expert information using a pre-built multi-dimensional matching algorithm, and obtaining a preliminary expert matching list;

[0078] In some optional implementations of some embodiments, obtaining expert information from the expert database according to preset expert screening conditions includes:

[0079] Obtain corresponding expert information based on the industry, nature of the unit, honorary title, technical field or expertise, professional title level, score, and past activity evaluation as the preset expert screening conditions;

[0080] The expert information corresponding to the industry is obtained as follows: the industry field in which the expert is located;

[0081] The expert information corresponding to the unit nature is obtained as follows: the type of unit to which the expert belongs;

[0082] The expert information corresponding to the honorary title is: the honors received by the expert;

[0083] The expert information corresponding to the technical field or expertise is obtained as follows: the expert's professional field or technical expertise;

[0084] The expert information corresponding to the professional title level is obtained as follows: the professional title of the expert;

[0085] The expert information corresponding to the ratings obtained is: the expert's rating based on past activities and peer reviews;

[0086] The expert information corresponding to the past activity evaluation is obtained as follows: the expert's performance evaluation in the activities he participated in the past.

[0087] Among other things, ensure that all expert-related data fields in the database are accurately populated. These fields include but are not limited to:

[0088] Industry: The industry sector in which the expert works.

[0089] Nature of unit: the type of unit to which the expert belongs (such as university, enterprise, research institute, etc.).

[0090] Honorary title: honors received by experts (such as academician, outstanding young scientist, etc.).

[0091] Technical Field / Expertise: The expert's area of expertise or technical specialization.

[0092] Professional title level: the expert’s professional title (such as professor, associate professor, researcher, etc.).

[0093] Rating: Experts are rated based on past activities and peer reviews.

[0094] Past activity evaluation: evaluation of the expert’s performance in past activities.

[0095] In some optional implementations of some embodiments, performing a preliminary fuzzy matching calculation on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list includes:

[0096] Construct a weighted scoring model, the formula is:

[0097]

[0098] Assume that the initial weights of each expert information are , the similarity between expert information and project requirements is ;

[0099] Calculate the similarity scores of all experts based on expert information After that, set the threshold τ to retain the similarity score Experts with numbers greater than τ are used as the preliminary expert matching list.

[0100] Implementation Example

[0101] Suppose we need to find experts for a project in the field of artificial intelligence. The target requirements are as follows:

[0102] Industry: Information Technology

[0103] Unit nature: University

[0104] Technology area / expertise: Artificial intelligence

[0105] Professional title: at least professor

[0106] Rating: at least 4.0

[0107] Calculate similarity: For each expert, calculate the similarity between him and the target requirement.

[0108] Weight setting: Assuming that the weights of industry and technology fields are higher (such as 0.3), while the weights of unit nature and professional title level are lower (such as 0.2), the weight of the score is 0.2.

[0109] Similarity calculation: For example, an expert in the information technology industry, whose technical field is machine learning (related to artificial intelligence), works at a university, has the title of associate professor, and has a score of 4.5. The similarity calculation is as follows:

[0110] Industry similarity: 1.0 (perfect match)

[0111] Unit Property Similarity: 1.0 (Exact Match)

[0112] Technical field similarity: 0.8 (related fields)

[0113] Professional title similarity: 0.5 (not yet professor)

[0114] Rating similarity: 1.0 (rating satisfied)

[0115] Comprehensive similarity for:

[0116]

[0117] Screening: Experts with a similarity greater than 0.8 are included in the preliminary candidate list.

[0118] Through this method, multi-dimensional expert matching can be effectively carried out to ensure that the selected experts best meet the project requirements.

[0119] S300: weighting the preliminary expert matching list using a multi-level matching model, and calculating the expert matching list after weighting using a pre-built precise matching algorithm to obtain a secondary screening expert list;

[0120] In some optional implementations of some embodiments, the weighting of the preliminary expert matching list using a multi-level matching model includes:

[0121] Determine the relative importance of each expert information in the preliminary expert matching list through the hierarchical analysis method, and determine the weight of each level through expert scoring and consistency test ;

[0122] Among them, the first level For basic information: nature of industry and unit;

[0123] Second level Professional ability: technical field or expertise and professional title level;

[0124] The third level For historical performance: honorary titles, ratings, and evaluations of past activities.

[0125] In some optional implementations of some embodiments, the weighted expert matching list is calculated using a pre-built exact matching algorithm to obtain a secondary screening expert list, including:

[0126] For each expert information after hierarchical assignment, calculate the similarity between the expert information and the requirements:

[0127]

[0128] in, Indicates the The similarity of the levels, represents the total amount of expert information in each level, Indicates the Expert information, and Represents expert information and demand in The eigenvalues of the expert information;

[0129] The similarity of each level is weighted by the hierarchical structure to obtain the comprehensive similarity :

[0130]

[0131] in, represents the comprehensive similarity, represents the weight of each level, For the Based on the similarity of the levels, a preset number of experts are selected in descending order according to the comprehensive similarity to enter the secondary screening expert list.

[0132] Implementation example:

[0133] Suppose you need to find experts for a new energy project with the following target requirements:

[0134] Industry: Energy

[0135] Unit nature: Research institution

[0136] Technical field / expertise: New energy technology

[0137] Professional title: at least senior researcher

[0138] Rating: At least 4.5

[0139] Construct feature vector: Construct a feature vector A=(a1,a2,…,an)A=(a1,a2,…,an) for each expert and compare it with the feature vector B=(b1,b2,…,bn)B=(b1,b2,…,bn) of the requirement.

[0140] Similarity calculation: The similarity of each expert is calculated using cosine similarity.

[0141] For example, if the feature vector of an expert is (1, 0.9, 0.8, 0.6, 1.0) (1, 0.9, 0.8, 0.6, 1.0) and the feature vector of the requirement is (1, 1, 1, 1, 1) (1, 1, 1, 1, 1), then the similarity is:

[0142]

[0143] Comprehensive similarity calculation: Combine the hierarchical weights to calculate the comprehensive similarity.

[0144] Final screening: Select the experts with the highest similarity as the final candidates.

[0145] Through this multi-level, multi-angle precise screening algorithm, the system can quickly and accurately find the most suitable experts in complex and changing needs, thereby improving the success rate and efficiency of event arrangements.

[0146] S400: Obtaining demand information input by a user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy;

[0147] In some optional implementations of some embodiments, obtaining demand information input by a user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy, includes:

[0148] The demand information input by the user is segmented into sentences or words as basic processing units. The basic processing units are used as nodes to build a node connection graph. The similarity between the basic processing units is used as the weight of the connection edge. The TextRank value of the basic processing unit is calculated through loop iteration, and then the selection is made based on the TextRank value. The calculation formula is:

[0149]

[0150] in the formula Points to the node The node set of is a node The set of all nodes pointed to: is a node With node The weight of the edge between them; Is the damping coefficient, which represents the probability of the current node jumping to any other node and ensures that the weight can be stably transferred to convergence. Finally, the weight of each word is calculated and sorted. According to the demand, the first N words in the sequence are selected as N user demand keywords and output. Indicates the nodes;

[0151] Among them, when the user inputs the demand information: "Jiangsu XX Shuangshun Machinery Manufacturing Co., Ltd. is a professional manufacturer of cement pipe machinery and series molds in China. The main demand description and requirements: 1. Use the double-excitation variable amplitude radial extrusion method to perform extrusion molding in the mold cavity. The extrusion device provides an exciter with variable diameter axial and radial vibration to achieve extrusion. 2. Design and improve the molding equipment set, mainly targeting the process level and control quality of the equipment, efficiently allocate materials for each process flow, adjust parameters, and improve performance. 3. Automatic and intelligent production process, safe and efficient production environment, real-time monitoring of production equipment operation, production scheduling process control detection and intelligence of important links. 4. Modular design of various main systems and auxiliary systems, mainly including production processes, mechanical components, various drive devices, control system hardware, control system software, etc."

[0152] The user demand keywords obtained by the formula calculation of the TextRank value are: Jiangsu Shuangshun Machinery Manufacturing Co., Ltd., double-excitation variable amplitude radial extrusion, pipe making process, machinery manufacturing;

[0153] Based on the short text characteristics of the user demand keywords and the secondary screening expert list, the Bert-TextCNN combined model is used to perform text vectorization processing to obtain a demand-expert vector set;

[0154] Furthermore, the BERT-TextCNN combined model is suitable for short text classification tasks and can also be used for feature extraction from short text data. Furthermore, the distributed text representation model combined with CNN-derived models can generate high-quality semantic encodings of text. The BERT model, constructed by stacking multiple Transformer encoders, obtains high-quality text vectors through the MLM and NSP tasks. The TextCNN model primarily consists of embedding, convolutional, pooling, and fully connected layers. This network architecture effectively extracts text features through a multi-channel approach, demonstrating strong local feature capture capabilities.

[0155] The final list of matching experts is determined from the clustering results through the demand expert matching strategy, and the steps of the demand expert matching strategy are as follows:

[0156] Step 1: Input the demand-expert vector set and calculate the similarity distance between each vector;

[0157] Furthermore, the cosine similarity after probability distribution is used to represent the similarity distance between vectors, specifically:

[0158]

[0159] in, express, It represents the cosine similarity between vectors i and j, μ is the mean, σ is the standard deviation, and the general value is μ=-1, 0=1. It mainly reflects the mapping of the cosine similarity of a vector i with the rest of the vectors. The probability distribution maps the data range of cosine similarity from [-1, 1] to a smaller positive range, standardizing the value range and improving the computational efficiency of the model.

[0160] Step 2: Determine the thresholds of local density and relative distance, calculate the local density and relative distance of any vector, and use the point where the local density and relative distance of the vector are greater than the threshold as the initial cluster center point;

[0161] Calculating vectors The local density and relative distance , a hybrid kernel function of radial basis function and polynomial function is used to calculate the local density based on K nearest neighbors. The calculation of the radial basis function depends on the distance between the sample point and the center point. The radial basis function value of the sample point with the same distance to the center point is the same. The vector inner product in the polynomial is determined by the vector modulus and the cosine value of the included angle, which implies the similarity relationship of the vectors. To ensure the adaptability of the initial cluster division on data sets with different distribution shapes, the two functions are combined to obtain a more hierarchical local density, which is calculated as follows:

[0162]

[0163]

[0164] in represents the mixed kernel function, β, q and σ represent the weight, order and function width respectively, and is the vector A vector point within the K nearest neighbors.

[0165] Determine local density and relative distance The value of , takes the point whose local density and relative distance are both greater than the threshold as the initial cluster center point. The value of is determined as follows:

[0166]

[0167]

[0168] Step 3: Complete the division of the data in the set according to the initial cluster center point, and initialize the membership matrix and membership weight matrix;

[0169] in and Represent the values of local density and relative distance respectively, Indicates the calculation of the median, N is the number of sample points, and It is a parameter. A reasonable value can screen out obvious cluster centers. The median reflects the concentration trend of local density and relative distance. According to the assumption of DPC clustering, the cluster center should have relatively high local density and relative distance. Based on the above heuristic method, we can determine and The values are: 0.75 and 1.57 respectively.

[0170] Step 4: Iteratively update the cluster clusters according to the membership matrix and the membership weight matrix to obtain the clustering results;

[0171] Initialize the membership matrix and membership weight matrix , the calculation formula is:

[0172]

[0173]

[0174] in, Indicates the membership of sample vector i to cluster j, M is the number of clusters, b is a constant, generally 2, The information entropy of the membership set of sample vector i is expressed. The smaller the information entropy is, the greater the amount of information provided is. The weight value is The bigger.

[0175] Updating the clusters includes: first, dividing the sample vectors into clusters according to the maximum membership, and taking the vector mean point as the latest cluster center point. Then, according to the formula:

[0176]

[0177] Recalculate the membership of sample vector i by The function updates the membership weight matrix. When the membership matrix converges and the error is less than 1e-6, the iteration ends and the clustering result is returned, completing the clustering.

[0178] Step 5: Obtain a list of experts that belong to the same cluster as the demand information from the clustering results, take the intersection of the expert list and the secondary screening expert list, and finally randomly sample the candidates in the intersection. According to the sampling results, dynamically obtain any number of experts from the candidates as the final matching expert list. Random sampling can be processed using the random.sample function in Python.

[0179] S500: Construct an expert information management platform to update the expert information in the final matching expert list in real time.

[0180] To implement a dynamic information update mechanism, enabling experts to update their personal information through a secure online platform while also providing administrators with manual update capabilities, the following solution can be designed:

[0181] Expert information management platform design:

[0182] 1. User Authentication and Authorization: Multi-factor authentication (MFA) is used to ensure user identity authenticity. Experts are required to provide their username, password, and dynamic verification code (sent via email or SMS) when logging in to ensure account security.

[0183] 2. Data encryption: During data transmission, all data streams should be encrypted using Transport Layer Security (TLS) to prevent data from being intercepted or tampered with during network transmission. At the same time, all stored data should also be encrypted using the Advanced Encryption Standard (AES) to ensure the security of the data during storage.

[0184] 3. Update interface design: Provide a user-friendly interface so that experts can easily update their personal information. The interface should include editable fields such as contact information, technical field, title, honorary title, etc.

[0185] Administrator manual update function:

[0186] 1. Administrator authority management: Provide administrators with different levels of authority to control access to and modification of expert information. Only authorized administrators can perform manual update operations.

[0187] 2. Logging and auditing: Every update operation by the administrator should be recorded in the log, including the update time, update content and operator information, to facilitate subsequent auditing and problem tracing.

[0188] 3. Anomaly detection and alarm: The system should have anomaly detection capabilities to promptly identify and alarm any abnormal update behavior, such as frequent update requests or suspicious update content.

[0189] Implementation example:

[0190] Suppose an expert needs to update his contact information and title information. Here are the specific steps:

[0191] Login and Verification: Experts log in to the secure online platform by entering their username and password, then receive a dynamic verification code via email and enter the verification code on the login screen to complete the login.

[0192] Information update: On the update interface, the expert finds the contact information and title fields, updates the old contact information to the new phone number, and updates the title from "Researcher" to "Senior Researcher".

[0193] Submit update: Click the "Save" button, the system will prompt you to confirm the updated information, and the expert will submit the update request after confirming that it is correct.

[0194] Data encryption and storage: The system encrypts the updated data with AES and stores it in the database to ensure data security.

[0195] Manual update by administrator: If an expert encounters information that cannot be updated by himself, such as the need to provide relevant certificates due to a change in professional title, the administrator can manually update the relevant information in the background system and record a detailed log of the operation.

[0196] Log audit: Dedicated auditors regularly check logs to ensure the legality and compliance of all update operations.

[0197] Through this mechanism, the system not only ensures timely updates of expert information, but also effectively safeguards the security and accuracy of the data. The flexibility and security of this design meet the needs of dynamic management of expert information, while also providing the necessary support for manual updates in special circumstances.

[0198] In the second aspect, this application proposes an expert matching and management system based on modern database technology, such as Figure 2 As shown, it includes expert database construction module, preliminary matching module, secondary screening module, final matching module and information update module;

[0199] The expert database construction module is used to obtain expert data and store the expert data using database technology to obtain an expert database;

[0200] The preliminary matching module is used to obtain expert information from the expert database according to preset expert screening conditions, and perform preliminary fuzzy matching calculations on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list;

[0201] The secondary screening module is used to assign weights to the preliminary expert matching list through a multi-level matching model, and calculate the expert matching list after weight assignment using a pre-built accurate matching algorithm to obtain a secondary screening expert list;

[0202] The final matching module is used to obtain demand information input by the user, perform keyword extraction on the demand information to obtain user demand keywords, use a text vectorization model to vectorize the user demand keywords and the secondary screening expert list to obtain a demand-expert vector set, use the demand-expert vector set as clustering input data to perform clustering iterative optimization to obtain a clustering result, and determine the final matching expert list from the clustering result through the demand-expert matching strategy;

[0203] The information updating module is used to construct an expert information management platform to update the expert information in the final matching expert list in real time.

[0204] In some optional implementations of some embodiments, an activity notification and feedback module is further included, which is used to inform experts of activity information through SMS, email or in-application notification, while recording attendance and collecting feedback.

[0205] Furthermore, the information update module includes a one-click notification function, which notifies experts of event information via SMS, email or in-app notification, while also recording attendance and collecting feedback.

[0206] One-click notification function design:

[0207] 1. Notification channel selection: Design a flexible notification system that supports SMS, email, and in-app notifications for expert activity information. The system should allow users to select the most appropriate notification channel based on the expert's preferences and contact information.

[0208] 2. Templated Notices: Create reusable notice templates to ensure consistency and accuracy. The template should include the event name, time, location, organizer information, and any special notes.

[0209] 3. Batch sending mechanism: The system should support batch notification sending function, and administrators can send notifications to selected expert groups with one-click operation, saving time and improving efficiency.

[0210] 4. Delivery Status Tracking: Monitor the delivery status of notifications in real time, including success, failure, and unread status, so that you can take timely measures to resolve any delivery issues.

[0211] Attendance Records and Feedback Collection:

[0212] 1. Attendance Recording: The system should support expert check-in at the event site or through the online platform. Check-in data is automatically recorded in the system for subsequent attendance analysis.

[0213] 2. Feedback Collection Mechanism: After the event, the system automatically sends a feedback form to participating experts to collect their satisfaction, suggestions, and opinions on the event. The feedback form should be simple and easy to use to increase the completion rate of experts.

[0214] 3. Data Analysis and Report Generation: The system should have analytical tools that can perform statistical analysis on the collected feedback data and generate detailed analysis reports. The reports can be used to improve the organization and scheduling of future activities.

[0215] Security and Privacy Protection:

[0216] 1. Data encryption and access control: All activity notifications and feedback data should be encrypted and stored to ensure the security of information during transmission and storage. Only authorized users can access and process this data.

[0217] 2. Privacy terms and consent mechanism: Before sending notifications and collecting feedback, the system should inform experts of relevant privacy terms and obtain their consent to comply with data protection laws and regulations.

[0218] Implementation example:

[0219] Assume that an industry seminar is about to be held and that you need to notify participating experts and collect feedback after the event. The implementation steps are as follows:

[0220] 1. Prepare a notification template: The administrator creates a notification template in the system that contains the seminar details.

[0221] 2. Select notification channel: Based on the expert's preference, choose to send activity information through email and in-app notification.

[0222] 3. Send notification: The administrator selects the expert list and clicks the "One-click Send" button, and the system automatically sends the notification to all selected experts.

[0223] 4. Sign-in and record: After arriving at the venue, experts sign in by scanning the QR code or clicking the sign-in button in the app, and the system automatically records their attendance.

[0224] 5. Feedback collection: After the workshop, the system automatically sends a feedback form link. Experts receive the link via email or in-app notification and fill in the feedback.

[0225] 6. Data Analysis: The system analyzes the feedback data and generates reports for reference by administrators and organizers to improve the organization of future events.

[0226] In a third aspect, the present application proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0227] In a fourth aspect, the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0228] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0229] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0230] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0231] In the embodiments provided in the present disclosure, it should be understood that the disclosed apparatus / computer equipment and methods can be implemented in other ways. For example, the apparatus / computer equipment embodiments described above are merely schematic. For example, the division of modules or units is merely a logical function division. In actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection of the apparatus or unit, which may be electrical, mechanical or other forms.

[0232] Units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0233] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0234] If the integrated module / unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present disclosure can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program can include computer program code, which can be in source code form, object code form, executable file, or some intermediate form. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, removable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunications signals, and software distribution media. It should be noted that the content included in computer-readable media can be appropriately increased or decreased based on the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, based on legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunications signals.

[0235] The above are only preferred embodiments of the present invention. It should be pointed out that various modifications and improvements made by those skilled in the art without departing from the present technical solution should also be deemed to fall within the scope of protection required by this solution.

Claims

1. Expert matching and management method based on modern database technology, characterized by: The following steps are involved: Acquiring expert data, and storing the expert data using database technology to obtain an expert database; According to the preset expert screening conditions, expert information is obtained from the expert database, and a preliminary fuzzy matching calculation is performed on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list, including: Construct a weighted scoring model, the formula is: Assume that the initial weights of each expert information are , the similarity between expert information and project requirements is ; Calculate the similarity scores of all experts based on expert information After that, set the threshold τ to retain the similarity score Experts with numbers greater than τ are used as the preliminary expert matching list; The preliminary expert matching list is weighted using a multi-level matching model, including: Determine the relative importance of each expert information in the preliminary expert matching list through the hierarchical analysis method, and determine the weight of each level through expert scoring and consistency test ; in, is the weight of the first-level basic information, which includes industry and unit nature; The weight of the second-level professional ability, which includes technical fields or expertise and professional title level; The third level is the weight of historical performance, which includes honorary titles, ratings, and past activity evaluations; The pre-built precise matching algorithm is used to calculate the expert matching list after weight allocation to obtain the secondary screening expert list, including: For each expert information after hierarchical assignment, calculate the similarity between the expert information and the requirements: in, Indicates the The similarity of the levels, represents the total amount of expert information in each level, Indicates the Expert information, and Represents expert information and demand in The eigenvalues of the expert information; The similarity of each level is weighted by the hierarchical structure to obtain the comprehensive similarity : in, represents the comprehensive similarity, represents the weight of each level, For the Based on the similarity of the levels, a preset number of experts are selected in descending order according to the comprehensive similarity to enter the secondary screening expert list; Obtaining demand information input by the user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy; An expert information management platform is constructed to update the expert information in the final matching expert list in real time.

2. The method according to claim 1, wherein: The step of acquiring expert data and storing the expert data using database technology to obtain an expert database includes: Select the appropriate database type based on the structured nature and query complexity of the expert data, including relational databases and NoSQL databases; When the database type is a relational database, a table structure is designed for the relational database; when the database type is a NoSQL database, a document structure is designed for the NoSQL database; Import the expert data according to the designed document structure / table structure; A query statement is created for the relational database or NoSQL database according to the multi-dimensional matching requirements to ensure the query efficiency and data security of the expert data and complete the construction of the expert database.

3. The method according to claim 2, wherein: The obtaining of expert information from the expert database according to the preset expert screening conditions includes: Obtain corresponding expert information based on the industry, nature of the unit, honorary title, technical field or expertise, professional title level, score, and past activity evaluation as the preset expert screening conditions; The expert information corresponding to the industry is obtained as follows: the industry field in which the expert is located; The expert information corresponding to the unit nature is obtained as follows: the type of unit to which the expert belongs; The expert information corresponding to the honorary title is: the honors received by the expert; The expert information corresponding to the technical field or expertise is obtained as follows: the expert's professional field or technical expertise; The expert information corresponding to the professional title level is obtained as follows: the professional title of the expert; The expert information corresponding to the ratings obtained is: the expert's rating based on past activities and peer reviews; The expert information corresponding to the past activity evaluation is obtained as follows: the expert's performance evaluation in the activities he participated in the past.

4. The method according to claim 3, wherein: The method includes obtaining demand information input by a user, performing keyword extraction on the demand information to obtain user demand keywords, vectorizing the user demand keywords and the secondary screening expert list using a text vectorization model to obtain a demand-expert vector set, performing clustering iterative optimization using the demand-expert vector set as clustering input data to obtain a clustering result, and determining a final matching expert list from the clustering result using a demand-expert matching strategy, including: The demand information input by the user is segmented into sentences or words as basic processing units. The basic processing units are used as nodes to build a node connection graph. The similarity between the basic processing units is used as the weight of the connection edge. The TextRank value of the basic processing unit is calculated through loop iteration, and then the selection is made based on the TextRank value. The calculation formula is: in the formula Points to the node The node set of is a node The set of all nodes pointed to: is a node With node The weight of the edge between them; Is the damping coefficient, which represents the probability of the current node jumping to any other node and ensures that the weight can be stably transferred to convergence. Finally, the weight of each word is calculated and sorted. According to the demand, the first N words in the sequence are selected as N user demand keywords and output. Indicates the nodes; Based on the short text characteristics of the user demand keywords and the secondary screening expert list, the Bert-TextCNN combined model is used to perform text vectorization processing to obtain a demand-expert vector set; The final list of matching experts is determined from the clustering results through the demand expert matching strategy, and the steps of the demand expert matching strategy are as follows: Step 1: Input the demand-expert vector set and calculate the similarity distance between each vector; Step 2: Determine the thresholds of local density and relative distance, calculate the local density and relative distance of any vector, and use the point where the local density and relative distance of the vector are greater than the threshold as the initial cluster center point; Step 3: Complete the division of the data in the set according to the initial cluster center point, and initialize the membership matrix and membership weight matrix; Step 4: Iteratively update the cluster clusters according to the membership matrix and the membership weight matrix to obtain the clustering results; Step 5: Obtain a list of experts that belong to the same cluster as the demand information from the clustering results, take the intersection of the expert list and the secondary screening expert list, and finally randomly sample the candidates in the intersection. According to the sampling results, dynamically obtain any number of experts from the candidates as the final matching expert list.

5. Expert matching and management system based on modern database technology, characterized by: It includes expert database construction module, preliminary matching module, secondary screening module, final matching module and information update module; The expert database construction module is used to obtain expert data and store the expert data using database technology to obtain an expert database; The preliminary matching module is used to obtain expert information from the expert database according to preset expert screening conditions, and perform preliminary fuzzy matching calculations on the expert information using a pre-built multi-dimensional matching algorithm to obtain a preliminary expert matching list, including: Construct a weighted scoring model, the formula is: Assume that the initial weights of each expert information are , the similarity between expert information and project requirements is ; Calculate the similarity scores of all experts based on expert information After that, set the threshold τ to retain the similarity score Experts with numbers greater than τ are used as the preliminary expert matching list; The secondary screening module is used to assign weights to the preliminary expert matching list through a multi-level matching model, including: Determine the relative importance of each expert information in the preliminary expert matching list through the hierarchical analysis method, and determine the weight of each level through expert scoring and consistency test ; in, is the weight of the first-level basic information, which includes industry and unit nature; The weight of the second-level professional ability, which includes technical fields or expertise and professional title level; The third level is the weight of historical performance, which includes honorary titles, ratings, and past activity evaluations; The pre-built precise matching algorithm is used to calculate the expert matching list after weight allocation to obtain the secondary screening expert list, including: For each expert information after hierarchical assignment, calculate the similarity between the expert information and the requirements: in, Indicates the The similarity of the levels, represents the total amount of expert information in each level, Indicates the Expert information, and Represents expert information and demand in The eigenvalues of the expert information; The similarity of each level is weighted by the hierarchical structure to obtain the comprehensive similarity : in, represents the comprehensive similarity, represents the weight of each level, For the Based on the similarity of the levels, a preset number of experts are selected in descending order according to the comprehensive similarity to enter the secondary screening expert list; The final matching module is used to obtain demand information input by the user, perform keyword extraction on the demand information to obtain user demand keywords, use a text vectorization model to vectorize the user demand keywords and the secondary screening expert list to obtain a demand-expert vector set, use the demand-expert vector set as clustering input data to perform clustering iterative optimization to obtain a clustering result, and determine the final matching expert list from the clustering result through the demand-expert matching strategy; The information updating module is used to construct an expert information management platform to update the expert information in the final matching expert list in real time.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images