Working platform and method based on big data

Abstract

The invention is suitable for the technical field of job hunting and recruitment, and provides a employment platform and employment method based on big data, and the method comprises the steps: uploading job hunting information; uploading recruitment information; matching all the job hunting information with all the recruitment information to generate a job hunter-recruiter recommendation list; receiving bad credit employer information uploaded by the job application account, adding label information to the bad credit employer information, constructing an employer comment area, and adding a reason of the bad credit employer to the employer comment area for display; and determining a credit score of each employer identity according to the poor credit employer information, adjusting the job seeker-recruiter recommendation list according to the credit scores, and determining a push sequence of recruitment information. The employer comment area is constructed, and the credit score of each employer identity is determined according to the bad credit employer information, so that comprehensive employer evaluation and credit information are provided for the job seeker, the job seeker is helped to avoid the bad credit employer, and the job hunting safety is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of job hunting and recruitment, and in particular to an employment platform and an employment method based on big data. Background Art

[0002] With the rapid development of the economy and the continuous adjustment of the industrial structure, the flexible employment model has gradually become an important part of the labor market. Traditional employment methods often have problems such as information asymmetry, low matching efficiency, and high recruitment costs. The flexible employment model has been widely welcomed by companies and job seekers for its flexibility, efficiency, and low cost. However, the existing flexible employment platform still has many shortcomings in information processing, matching efficiency, and security.

[0003] First, the information posted by job seekers and employers on the platform often lacks depth and breadth, resulting in inaccurate matching results. Secondly, the platform lacks efficient data processing and analysis capabilities when processing a large amount of job search and recruitment information, making the matching process time-consuming and inefficient. In addition, the existing platforms also have obvious shortcomings in protecting the rights and interests of job seekers and preventing bad credit employers. Job seekers often find it difficult to obtain comprehensive evaluations and credit information of employers, which increases the risk of job hunting. Therefore, it is necessary to provide an employment platform and employment method based on big data to solve the above problems. Summary of the invention

[0004] In view of the deficiencies in the prior art, the purpose of the present invention is to provide an employment platform and an employment method based on big data to solve the problems existing in the above-mentioned background technology.

[0005] The present invention is implemented as follows: a method for employing workers based on big data, the method comprising the following steps: Upload job application information through a job application account, including the job applicant's identity, type of work, type of machine that can be operated, dominant hand, age and job application location; Upload recruitment information through a recruitment account, including the employer's identity, type of work, required machine type, dominant hand, age and work location; Match all job search information with all recruitment information, conduct preliminary screening based on job types and operating hand preference, and obtain screening results; Dynamic scoring is calculated based on the distance between the two parties, the rarity of the machine type, and the age requirements in the screening results. The current labor market supply and demand status is considered in the dynamic scoring calculation process to generate a matching score, and a job seeker-employer recommendation list is generated based on the matching score; Receiving bad credit employer information uploaded by a job-seeking account, wherein the bad credit employer information includes the employer's identity and the reason for the bad credit employer; Add label information to the bad credit employer information, build an employer comment area, add the bad credit employer reasons to the employer comment area, and display the label information and the bad credit employer reasons to job seekers; Determine the credit score of each employer based on the bad credit employer information, adjust the job seeker-employer recommendation list based on the credit score, and determine the order of pushing recruitment information; Finally, conduct a comprehensive analysis of all job search and recruitment information to determine employment trend information.

[0006] Another object of the present invention is to provide a big data-based employment platform, the platform comprising: A job search information uploading module is used to upload job search information through a job search account, wherein the job search information includes the job seeker's identity, type of work, type of machine that can be operated, dominant hand, age and job search location; A recruitment information uploading module is used to upload recruitment information through a recruitment account, wherein the recruitment information includes the employer's identity, type of work, required machine type, operating dominant hand, age and work location; The job search and recruitment matching module is used to match all job search information with all recruitment information and generate a job seeker-recruiter recommendation list; A bad credit employer information module is used to receive bad credit employer information uploaded by a job-seeking account, wherein the bad credit employer information includes the employer identity and the reason for the bad credit employer; The employer comment area module is used to add label information to the bad credit employer information, build an employer comment area, add the bad credit employer reasons to the employer comment area, and display the label information and the bad credit employer reasons to job seekers; A credit score determination module is used to determine the credit score of each employer based on the information of bad credit employers, adjust the job seeker-employer recommendation list based on the credit score, and determine the order of pushing recruitment information; The employment trend information module is used to conduct a comprehensive analysis of all job search information and recruitment information to determine employment trend information.

[0007] Compared with the prior art, the present invention has the following beneficial effects: In the present invention, job seekers can upload information about bad credit employers, add label information to the bad credit employer information, and build an employer comment area, add the bad credit employer reasons to the employer comment area for display, and determine the credit score of each employer identity based on the bad credit employer information, providing job seekers with comprehensive employer evaluation and credit information, which helps job seekers avoid bad credit employers and enhance job search security. The push order of recruitment information will also be determined based on the credit score, so that employers with good credit can get the attention of job seekers first, improving the overall credibility and user experience of the platform. It is possible to conduct a comprehensive analysis of all job search information and recruitment information, determine employment trend information, and help companies and job seekers better grasp market dynamics and make more informed decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 The figure is a flowchart of a labor employment method based on big data.

[0009] Figure 2 This is a structural diagram of a big data-based employment platform. DETAILED DESCRIPTION

[0010] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0011] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0012] like Figure 1 As shown, an embodiment of the present invention provides a method for employing workers based on big data, and the method includes the following steps: S100, uploading job search information through a job search account, wherein the job search information includes the job seeker's identity, type of work, type of machine that can be operated, dominant hand, age, and job search location; S200, uploading recruitment information through a recruitment account, wherein the recruitment information includes the employer's identity, type of work, required machine type, dominant hand, age and work location; S300, matching all job search information with all recruitment information, and performing preliminary screening according to the type of work and the dominant hand to obtain the screening results; Dynamic scoring is calculated based on the distance between the two parties, the rarity of the machine type, and the age requirements in the screening results. The current labor market supply and demand status is considered in the dynamic scoring calculation process to generate a matching score, and a job seeker-employer recommendation list is generated based on the matching score; S400, receiving bad credit employer information uploaded by a job-seeking account, wherein the bad credit employer information includes the employer identity and the reason for the bad credit employer; S500, adding label information to the bad credit employer information, and constructing an employer comment area, adding the bad credit employer reason to the employer comment area, and displaying the label information and the bad credit employer reason to the job seeker; S600, determining the credit score of each employer based on the information of employers with bad credit, adjusting the job seeker-employer recommendation list based on the credit score, and determining the order of pushing recruitment information; S700 conducts comprehensive analysis of all job search and recruitment information to determine employment trend information.

[0013] It should be noted that the existing flexible employment platforms still have many deficiencies in information processing, matching efficiency and security. First, the information posted by job seekers and employers on the platform often lacks depth and breadth, resulting in inaccurate matching results. Secondly, when processing a large amount of job search and recruitment information, the platform lacks efficient data processing and analysis capabilities, making the matching process time-consuming and inefficient. In addition, the existing platforms also have obvious shortcomings in protecting the rights and interests of job seekers and preventing employers with bad credit. Job seekers often find it difficult to obtain comprehensive evaluations and credit information of employers, thereby increasing the risk of job hunting. The embodiments of the present invention are intended to solve the above problems.

[0014] In the embodiment of the present invention, first, the job seeker needs to register a job search account, and the recruiter needs to register a recruitment account, and then upload the job search information through the job search account, and upload the recruitment information through the recruitment account. The job search information includes the identity of the job seeker, the type of work, the type of machine that can be operated, the operating hand, the age and the job search location, and the recruitment information includes the identity of the employer, the type of work, the type of machine required, the operating hand, the age and the work location. The information is more specific and comprehensive, making the matching process more accurate and meeting the personalized needs of employers and job seekers. Then all job search information and all recruitment information will be matched to generate a job seeker-recruiter recommendation list, which greatly shortens the matching time and improves the matching efficiency. In addition, the job seeker can also upload the bad credit employer information, the bad credit employer information includes the employer identity and the bad credit employer cause, add label information to the bad credit employer information, and construct an employer comment area, add the bad credit employer cause to the employer comment area for display, and determine the credit score of each employer identity according to the bad credit employer information, providing job seekers with comprehensive employer evaluation and credit information, helping job seekers avoid bad credit employers and enhancing job search security. The order of job postings will also be determined based on credit scores, so that employers with good credit can get priority attention from job seekers, improving the overall credibility of the platform and user experience. Finally, all job search and recruitment information will be comprehensively analyzed to determine employment trend information, which will help companies and job seekers better grasp market dynamics and make more informed decisions.

[0015] As a preferred embodiment of the present invention, the steps of matching all job search information with all recruitment information, performing preliminary screening according to the type of work and the dominant hand, and obtaining screening results; performing dynamic score calculation according to the distance between the two parties, the rarity of the machine type, and the age requirement in the screening results, and taking into account the current labor market supply and demand status in the dynamic score calculation process to generate a matching score, and generating a job seeker-recruiter recommendation list according to the matching score, specifically include: S30201, obtaining a job classification code, matching the job classification code of the job seeker with the job classification code issued by the employer, and obtaining a first matching result in which the job classification codes are consistent; S30202, comparing the operating hand declared by the job seeker in the first matching result with the operating hand requirement of the position in the recruiter to obtain a screening result; S30203, calculating the actual physical distance between the job-seeking location of the job seeker and the job location of the employer according to the latitude and longitude coordinates of the two locations in the screening results, and applying an exponential decay function to the actual physical distance to obtain an exponential decay coefficient; S30204, obtain the administrative levels of the job-seeking location and the work location, assign different weights according to the administrative levels, and obtain the affiliation weights: For example: Completely the same street: the membership degree is 1; Different streets in the same district: the degree of affiliation is reduced to 0.8; Different districts in the same city: the degree of affiliation is reduced to 0.6; By analogy, the degree of affiliation is lowest when matching across provinces.

[0016] S30205, weighted fusion of the exponential decay coefficient and the membership weight to generate a spatial matching score; S30206, calculating the intersection and union of the machine types that the job seeker can operate and the machine types required by the employer in the screening results; S30207, count the frequency of occurrence of various machine types in the recruitment market, and assign corresponding scarcity weights to each machine type according to the rarity of each machine type; S30208, calculating weighted similarity according to the scarcity weight of the intersection machine type and the scarcity weight of the union machine type to obtain a device matching score; S30209, obtaining the age requirements of the employer, constructing an S-curve function, and using the age requirements of the employer as the central transition zone of the S-curve function, substituting the actual age of the job seeker into the S-curve function, and obtaining the corresponding age matching score according to the slope of the S-curve function; S30210, according to the current labor market supply and demand status, dynamically assign weights to the spatial matching score, the equipment matching score, and the age matching score to obtain a spatial matching comprehensive score, an equipment matching comprehensive score, and an age matching comprehensive score, respectively, and weight the spatial matching comprehensive score, the equipment matching comprehensive score, and the age matching comprehensive score according to the dynamic weights to obtain a final matching score; S30211, sort all job seekers for the same recruitment position in descending order according to the final matching score, generate a recruitment position matching list, count the recruitment position matching lists of all positions, and generate a job seeker-recruiter recommendation list.

[0017] In the embodiment of the present invention, invalid matches are quickly screened out through hard conditions such as job types and operators (for example, a right-handed job will not recommend left-handed job seekers) to avoid wasting resources; at the same time, flexible algorithms such as geographic location attenuation and age fuzzy matching are introduced to achieve more refined priority sorting through quantitative scoring after key constraints are met. In addition, industry scarcity weights are used in machine type matching (for example, precision instruments have higher weights than ordinary machine tools) to reflect the real supply and demand relationship in the employment market and solve the problem of "the same equipment labels but large differences in actual skill requirements" in traditional matching.

[0018] In addition, in order to adapt to market supply and demand, dynamic weights are designed to automatically adjust the weights of each dimension according to the supply and demand ratio of the job type (such as relaxing geographical restrictions and increasing equipment matching weights when there is a labor shortage), so as to avoid rigid rules that may lead to an increase in matching failure rates.

[0019] As a preferred embodiment of the present invention, the steps of adding label information to the bad credit employer information, constructing an employer comment area, adding the bad credit employer reasons to the employer comment area, and displaying the label information and the bad credit employer reasons to job seekers specifically include: S501, classifying all bad credit employer information according to employer identity, where the employer identity in each category is the same; S502, extracting high-frequency keywords of the reasons for bad credit of employers in each category, and adding label information to each employer identity according to the high-frequency keywords; S503, constructing an employer comment area, and adding the bad credit employer reasons to the corresponding employer comment area; S504, displaying bad credit employer label information and bad credit employer reasons to serve as identification information for job seekers during the job search process.

[0020] In an embodiment of the present invention, in order to obtain an employer comment area, all bad credit employer information is classified according to the employer identity, and then high-frequency keywords of the bad credit employer reasons in each category are extracted based on natural language processing (NLP) technology, label information is added to each employer identity according to the high-frequency keywords, and an employer comment area is constructed, and the bad credit employer reasons are added to the corresponding employer comment area, so that job seekers can have a more comprehensive understanding of the employer.

[0021] As a preferred embodiment of the present invention, the step of classifying all bad credit employer information according to the employer identity, wherein the employer identities in each category are the same, specifically includes: S50101, import the standard business registration dictionary into the enterprise credit information disclosure system, which includes the industry classification standard word list and the enterprise business scope standard expression word library; S50102, employer identity includes employer name, business registration number and unified social credit code. A bidirectional LSTM-CRF neural network model is used to perform fine-grained segmentation on the employer name to obtain the word segmentation result; S50103, matching the word segmentation results with the standard business registration dictionary to extract core words and obtain the font size and industry characteristic words; S50104, establishing a synonym mapping table, and using the synonym mapping table to perform standardized conversion on the industry characteristic words to obtain standardized industry characteristic words; S50105, using the font sizes of all employers to construct a global corpus, using TF-IDF weight analysis to calculate the TF-IDF word frequency of the font size in the global corpus, and obtain the TF-IDF weight; S50106, using the pre-trained industry-specific word vector model, calculate the semantic similarity between the standardized industry characteristic words and the standard business registration dictionary to obtain the part-of-speech tag; S50107, perform Luhn algorithm verification on the business registration number, verify the legitimacy of the code and generate a binary check digit feature, parse the business registration number according to the binary check digit feature, and obtain the registration area code; S50108, performs a SHA-256 hash operation on the unified social credit code to generate a fixed-length cryptographic feature summary; S50109, combining the binary check bit feature with the cryptographic feature summary to obtain a coding check feature vector; S50110, according to the frequency of occurrence of industry characteristic words in the standard business registration dictionary, the corresponding industry classification weight is given, the TF-IDF weight, part-of-speech tag and industry classification weight are used as text feature vectors, and the text feature vectors are orthogonally spliced ​​with the coding check feature vectors to form a multi-dimensional hybrid feature representation; S50111, create independent clusters for each employer, each cluster contains a multi-dimensional mixed feature representation of a single employer; S50112, according to the clustering time period, the dynamic threshold is set in an incremental manner; S50113, according to the set dynamic threshold, the Jaccard-Wasserstein composite metric is used to calculate the feature vector similarity between two clusters; S50114, when the similarity between two clusters exceeds a dynamic threshold, the two clusters are merged to obtain a clustering result, in which each cluster represents a category of employer identity.

[0022] In the embodiment of the present invention, a multimodal feature fusion method is used to integrate structured coding information and unstructured text features, breaking through the limitations of traditional text matching, and using the Jaccard-Wasserstein composite metric: it not only considers the set similarity of text features, but also evaluates the distribution distance of coding features, thereby ensuring the accuracy of clustering. In order to balance efficiency and accuracy, a dynamic threshold mechanism is set according to the time period of clustering. In the early stage, a loose threshold is used to promote large-category aggregation, and the threshold is increased in the later stage to ensure fine-grained distinction.

[0023] Taking the employer name: XX Mechanical Equipment Co., Ltd., business registration number: 12345678, unified social credit code: ******* as an example, the above scheme is explained as follows: The bidirectional LSTM-CRF neural network model is used to perform fine-grained segmentation of the employer name, and the segmentation results are obtained ["XX", "machinery", "equipment", "limited company"]; Match the word segmentation results with the standard business registration dictionary, retain "XX" (font size) and "machinery" (industry characteristic word) as core words; identify "machinery equipment" as a standard business scope statement; The TF-IDF weight analysis is used to calculate the TF-IDF frequency of "XX" (font size) in the global corpus, and the TF-IDF value is 0.85; The semantic similarity between the standardized industry characteristic word "machinery" and the standard industrial and commercial registration dictionary is 0.92; Perform Luhn algorithm verification on the business registration number 12345678. The business registration number verification digit is: 1 (valid). The registration area code is parsed from the business registration number. Perform hash operation on the social credit code *******, hash value: a1b2c3d4...; Orthogonally concatenate the TF-IDF value, semantic similarity, check digit, and hash value to obtain a mixed feature vector: [0.85, 0.92, 1, a1b2c3d4...]; The Jaccard-Wasserstein composite metric is used to calculate the similarity of the feature vectors between two clusters: Similarity with "XX Machinery Manufacturing Co., Ltd.": 0.89; Text similarity: 0.85 (font size matches, industry is similar); Registration area code similarity: 0.93 (same registration location); Similarity with "XX Mechanical and Electrical Equipment Company": 0.72; Text similarity: 0.68 (difference in industry characteristics); Registration area code similarity: 0.76 (different registration locations); The current dynamic threshold is 0.85; If the similarity with "XX Machinery Manufacturing Co., Ltd." is 0.89 > the dynamic threshold 0.85, the merge operation is triggered; Monte Carlo validation: 10 sampling similarities are all > 0.85, confirming the merger; Generate classification label: "Mechanical equipment manufacturing enterprises"; record classification basis: core feature words: "XX", "machinery"; registration place feature: *Hai; business scope: mechanical equipment manufacturing.

[0024] As a preferred embodiment of the present invention, the step of extracting high-frequency keywords of the reasons for bad credit employers in each category and adding label information to each employer identity according to the high-frequency keywords specifically includes: S50201, translate the text of bad credit employers into English and build a bilingual parallel corpus; S50202, compares the semantic differences between Chinese and English texts and extracts cross-language vectors that are not affected by language; S50203, use the pre-trained Chinese BERT model to perform deep semantic capture on the text of the bad credit employer’s reasons to obtain a semantic vector containing contextual information; S50204, using the pre-trained BiLSTM model to analyze the word order relationship of the bad credit employer's cause text to extract word order features, and then using the attention mechanism to perform weighted calculation on the word order features to obtain a feature vector with time-order weights; S50205, dimensionality alignment of the semantic vector of context information, the feature vector with temporal weights, and the cross-language vector through a fully connected layer; S50206, weighted fusion is performed on the semantic vector of the context information after dimension alignment, the feature vector with temporal weight, and the cross-language vector. In the weighted fusion process, the weights corresponding to the semantic vector of the context information and the feature vector with temporal weight are adjusted according to the text length of the text of the employer's reasons for bad credit, so as to obtain a multimodal fusion vector; S50207, obtain the labor violation word library, search for the number and frequency of labor violation words in the text of the bad credit employer, and obtain the corresponding risk weight according to the number and frequency; S50208, perform TF-IDF analysis on the text of the employer's reasons for bad credit, obtain the TF-IDF value, and weight the TF-IDF value with the risk weight to obtain the improved value; S50209, LDA is used to perform topic mining on multimodal fusion vectors to generate a topic-vocabulary distribution matrix; S50210, based on the improvement value, select the first several high-frequency words as network nodes to construct a graph structure; and combine the topic-vocabulary distribution matrix to form a semantic network graph; S50211, based on the co-occurrence strength of word pairs in the topic-vocabulary distribution matrix, a graph convolutional network is used to analyze the association strength between nodes in the graph structure, and a weighted semantic network graph is constructed, where nodes are keywords and edge weights are associations; S50212, using density peak-based algorithm to identify core nodes in weighted semantic network graphs; S50213, performing semantic clustering according to the core nodes, and in the clustering process, dynamically adjusting the clustering radius according to the edge weight in the weighted semantic network graph to obtain semantic clusters; S50214, calculating the eccentricity according to the improvement value and the centrality of the weighted semantic network graph, and selecting the first several words from the semantic cluster as the label information of the employer identity according to the eccentricity.

[0025] In the embodiment of the present invention, a cross-dimensional feature fusion method is adopted to overcome the inherent defect of contextual ambiguity in single text analysis through the complementary features of semantic depth and temporal association, and the weight distribution is automatically optimized according to the text features. Compared with the fixed ratio fusion method, it can effectively improve the accuracy.

[0026] It also uses "explicit + implicit" dual analysis, obtains surface word frequency through TF-IDF, obtains potential topics through LDA, and uses surface word frequency and potential topics for collaborative application, thereby capturing high-frequency words such as wage arrears and identifying implicit associations such as wage arrears and labor arbitration. And by dynamically adjusting the clustering radius, it can accurately segment in high-density areas and avoid over-segmentation in low-density areas, effectively assigning corresponding keywords to bad credit employers, so that job seekers can have a more comprehensive understanding of employers.

[0027] Taking the text of the bad credit employer as an example, "Employees of a construction company reported that they had not received their wages for six consecutive months and had not signed a written labor contract. After many unsuccessful negotiations, they complained to the relevant agency", the above scheme is explained as follows: First, change the text range to English, and get the English text "Employees of a construction company reported unpaid wages for six consecutive months, with no written labor contract"; By extracting cross-language common features: "unpaid wages" and "labor contract" are locked as language-independent core elements; The bad credit employer reason text is input into BERT, and BERT captures the strong correlation between "six consecutive months" and "not issued" (similarity 0.88); Then, the text of the employer’s reasons for bad credit is input into BiLSTM and weighted by the attention mechanism to identify the key time series pattern: [Six consecutive months] No salary paid (Attention weight 0.94) → [No labor contract signed] (0.89) → [Complaint] (0.82); The length of the text of the employer's bad credit is 49 words, and the BERT semantic feature weight should be assigned 80%, and the BiLSTM temporal feature weight should be assigned 20%, and then fused to generate a fusion vector. For example: highlight the composite features of "wage arrears" and "contract missing"; Improvement value calculation is performed on the text of the employer's reasons for bad credit, such as: "Unpaid wages" basic value 23.5 × risk factor 3 → final score 70.5; "Labor contract" basic value 18.2 × risk factor 3 → 54.6; LDA is used to perform topic mining on the multimodal fusion vector and two potential topics are found, for example: Topic 1: Salary payment | Hours | Negotiation (probability 0.63); Topic 2: Contract signing | Legal procedures | Complaints (probability 0.57); According to the improvement value, words such as "unpaid wages", "labor contract", and "complaint" are selected as nodes; The “failure to issue wages-labor contracts” is cross-topic and co-occurs 2 times → basic edge weight is 0.4, and the “wages-complaints” is in the same topic and co-occurs 3 times → edge weight is 0.4+0.3=0.7; a weighted semantic network graph is formed; The density peak algorithm is used to identify the core nodes of the weighted semantic network graph. Finally, "unpaid wages" becomes the core node (density value 8.7). Then clustering is performed to generate the semantic clusters: unpaid wages | arrears duration | negotiation records. According to the eccentricity, unpaid wages|arrears duration (six months) are selected as the label information of the employer's identity.

[0028] As a preferred embodiment of the present invention, the step of determining the credit score of each employer identity according to the bad credit employer information specifically includes: S601, analyzing all bad credit employer information of each employer identity to determine the credit type corresponding to the bad credit employer cause; S602, determining the credit deduction points for each employer with bad credit information according to the credit type, and summarizing all the credit deduction points to obtain the credit score of the corresponding employer identity.

[0029] In an embodiment of the present invention, a credit score is also calculated. Specifically, all bad credit employer information of each employer identity is analyzed to determine the credit type corresponding to the bad credit employer cause. The credit types include wage arrears, wage deductions, non-compliance with contractual agreements, etc. Each credit type has corresponding credit deductions in advance. In this way, the credit deductions for each bad credit employer information are determined according to the credit type, and the credit score of the corresponding employer identity can be obtained by summarizing all the credit deductions.

[0030] As a preferred embodiment of the present invention, the step of comprehensively analyzing all job search information and recruitment information to determine employment trend information specifically includes: S701, determining the recruitment quantity for each type of work, the recruitment quantity for each machine type, the job seeker quantity for each type of work, and the job seeker quantity for each machine type according to the job seeker information and the recruitment information; S702, determining the employment trend of each job type according to the recruitment volume of each job type and the job seeker volume of each job type; S703, determining the employment trend of each machine type according to the recruitment volume of each machine type and the job seeker volume of each machine type.

[0031] In an embodiment of the present invention, all job search information and recruitment information are integrated to determine the recruitment quantity of each type of work, the recruitment quantity of each machine type, the job seekers of each type of work and the job seekers of each machine type. Then, the employment trend of each type of work is determined based on the recruitment quantity of each type of work and the job seekers of each type of work, and the employment trend of each machine type is determined based on the recruitment quantity of each machine type and the job seekers of each machine type. The employment trend indicates an increase in demand or a decrease in demand.

[0032] like Figure 2 As shown, an embodiment of the present invention further provides a big data-based employment platform, the platform comprising: A job search information uploading module 100 is used to upload job search information through a job search account, wherein the job search information includes the job seeker's identity, type of work, type of machine that can be operated, dominant hand, age and job search location; A recruitment information uploading module 200 is used to upload recruitment information through a recruitment account, wherein the recruitment information includes the employer's identity, type of work, required machine type, dominant hand, age and work location; The job search and recruitment matching module 300 is used to match all job search information with all recruitment information and generate a job seeker-recruiter recommendation list; The bad credit employer information module 400 is used to receive the bad credit employer information uploaded by the job-seeking account, wherein the bad credit employer information includes the employer identity and the bad credit employer reason; The employer comment area module 500 is used to add label information to the bad credit employer information, and to construct an employer comment area, to add the bad credit employer reasons to the employer comment area, and to display the label information and the bad credit employer reasons to job seekers; A credit score determination module 600 is used to determine the credit score of each employer based on the bad credit employer information, adjust the job seeker-employer recommendation list based on the credit score, and determine the order of pushing recruitment information; The employment trend information module 700 is used to comprehensively analyze all job search information and recruitment information to determine employment trend information.

[0033] The above only describes in detail the preferred embodiments of the present invention, which is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0034] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0035] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0036] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A method of employing workers based on big data, characterized in that: The method comprises the following steps: Upload job application information through a job application account, including the job applicant's identity, type of work, type of machine that can be operated, dominant hand, age and job application location; Upload recruitment information through a recruitment account, including the employer's identity, type of work, required machine type, dominant hand, age and work location; Match all job search information with all recruitment information, conduct preliminary screening based on job types and operating hand preference, and obtain screening results; Dynamic scoring is calculated based on the distance between the two parties, the rarity of the machine type, and the age requirements in the screening results. The current supply and demand status of the labor market is considered in the dynamic scoring calculation process to generate a matching score, and a job seeker-employer recommendation list is generated based on the matching score; Receiving bad credit employer information uploaded by a job-seeking account, wherein the bad credit employer information includes the employer's identity and the reason for the bad credit employer; Add label information to the bad credit employer information, build an employer comment area, add the bad credit employer reasons to the employer comment area, and display the label information and the bad credit employer reasons to job seekers; Determine the credit score of each employer based on the bad credit employer information, adjust the job seeker-employer recommendation list based on the credit score, and determine the order of pushing recruitment information; Comprehensively analyze all job search and recruitment information to determine employment trend information.

2. The employment method based on big data according to claim 1, characterized in that: Match all job search information with all recruitment information, perform preliminary screening based on job types and operating hand preference to obtain screening results; perform dynamic score calculation based on the distance between the two parties, the rarity of machine types, and age requirements in the screening results, and consider the current labor market supply and demand status in the dynamic score calculation process to generate a matching score, and generate a job seeker-recruiter recommendation list based on the matching score, specifically including: Obtaining a job classification code, matching the job classification code of the job seeker with the job classification code issued by the employer, and obtaining a first matching result in which the job classification codes are consistent; The operating hand declared by the job seeker in the first matching result is screened with the operating hand requirement of the position in the recruiter to obtain a screening result; According to the latitude and longitude coordinates of the job-seeking location of the job seeker and the job location of the recruiter in the screening results, the actual physical distance between the two locations is calculated, and an exponential decay function is applied to the actual physical distance to calculate an exponential decay coefficient; Get the administrative levels of the job-seeking location and the work location, assign different weights according to the administrative level, and get the affiliation weight: The exponential decay coefficient is weighted and fused with the membership weight to generate a spatial matching score; Calculate the intersection and union of the machine types that the job seeker can operate and the machine types required by the employer in the screening results; Count the frequency of each machine type in the recruitment market, and assign a corresponding scarcity weight to each machine type based on its rarity; The weighted similarity is calculated according to the scarcity weight of the intersection machine type and the scarcity weight of the union machine type, thereby obtaining the device matching score; Obtain the age requirements of the employer, construct an S-curve function, and use the age requirements of the employer as the central transition zone of the S-curve function, substitute the actual age of the job seeker into the S-curve function, and determine the corresponding age matching score according to the slope of the S-curve function; According to the current supply and demand status of the labor market, dynamic weights are assigned to the spatial matching score, equipment matching score and age matching score to obtain the spatial matching comprehensive score, equipment matching comprehensive score and age matching comprehensive score respectively. The spatial matching comprehensive score, equipment matching comprehensive score and age matching comprehensive score are weighted and fused according to the dynamic weights to obtain the final matching score. All job seekers for the same recruitment position are sorted in descending order according to the final matching score to generate a recruitment position matching list, and the recruitment position matching lists of all positions are counted to generate a job seeker-recruiter recommendation list.

3. The employment method based on big data according to claim 2 is characterized in that: The steps of adding label information to the bad credit employer information, constructing an employer comment area, adding the bad credit employer reasons to the employer comment area, and displaying the label information and the bad credit employer reasons to job seekers specifically include: All bad credit employer information is classified according to the employer identity, and the employer identity in each category is the same; Extract high-frequency keywords of bad credit employers in each category, and add label information to each employer based on the high-frequency keywords; Build an employer review area and add the bad credit employer reasons to the corresponding employer review area; The bad credit employer label information and the bad credit employer reasons are displayed to serve as identification information for job seekers during the job search process.

4. The employment method based on big data according to claim 3 is characterized in that: The step of classifying all bad credit employer information according to the employer identity, wherein the employer identities in each category are the same, specifically includes: Import the standard business registration dictionary into the enterprise credit information disclosure system, which includes the industry classification standard vocabulary and the enterprise business scope standard expression vocabulary; The employer identity includes the employer name, business registration number, and unified social credit code. The bidirectional LSTM-CRF neural network model is used to perform fine-grained segmentation on the employer name to obtain the word segmentation result; Match the word segmentation results with the standard business registration dictionary to extract core words and obtain the font size and industry characteristic words; Establish a synonym mapping table, and use the synonym mapping table to perform standardized conversion on industry characteristic words to obtain standardized industry characteristic words; A global corpus is constructed using the font sizes of all employers, and the TF-IDF weight analysis is used to calculate the TF-IDF word frequency of the font size in the global corpus to obtain the TF-IDF weight; Use the pre-trained industry-specific word vector model to calculate the semantic similarity between the standardized industry feature words and the standard business registration dictionary to obtain the part-of-speech tags; Perform Luhn algorithm verification on the business registration number to verify the legitimacy of the code and generate a binary check digit feature. Parse the business registration number based on the binary check digit feature to obtain the registration area code. Perform SHA-256 hash operation on the unified social credit code to generate a fixed-length cryptographic feature summary; Combine the binary check bit feature with the cryptographic feature summary to obtain a coding check feature vector; According to the frequency of occurrence of industry characteristic words in the standard business registration dictionary, the corresponding industry classification weight is given, and the TF-IDF weight, part-of-speech tag and industry classification weight are used as text feature vectors. The text feature vectors are orthogonally spliced ​​with the coding check feature vectors to form a multi-dimensional hybrid feature representation; Create independent clusters for each employer, each cluster contains a multi-dimensional mixed feature representation of a single employer; According to the clustering time period, the dynamic threshold is set in an incremental manner; According to the set dynamic threshold, the Jaccard-Wasserstein composite metric is used to calculate the feature vector similarity between two clusters; When the similarity of the feature vectors between two clusters exceeds a dynamic threshold, the two clusters are merged to obtain a clustering result, in which each cluster represents a category of employer identity.

5. The employment method based on big data according to claim 4 is characterized in that: The step of extracting high-frequency keywords of the reasons for bad credit employers in each category and adding label information to each employer identity according to the high-frequency keywords specifically includes: Translate the text of bad credit employers into English and build a bilingual parallel corpus; Compare the semantic differences between Chinese and English texts and extract cross-language vectors that are not affected by language; Use the pre-trained Chinese BERT model to perform deep semantic capture on the text of the employer's reasons for bad credit, and obtain a semantic vector containing contextual information; The pre-trained BiLSTM model is used to analyze the word order relationship of the bad credit employer's cause text to extract word order features, and then the word order features are weighted by the attention mechanism to obtain a feature vector with time-order weights; The semantic vector of context information, the feature vector with temporal weights, and the cross-language vector are dimensionally aligned through a fully connected layer; The semantic vector of the context information after dimension alignment, the feature vector with time series weight and the cross-language vector are weightedly fused. In the weighted fusion process, the weights corresponding to the semantic vector of the context information and the feature vector with time series weight are adjusted according to the text length of the text of the employer with bad credit, so as to obtain a multimodal fusion vector. Multi-dimensional analysis and clustering operations are performed on the multimodal fusion vector and the text of the employer’s reasons for bad credit to obtain the label information of the employer’s identity.

6. The employment method based on big data according to claim 5 is characterized in that: The steps of performing multi-dimensional analysis and clustering operations on the multimodal fusion vector and the text of the employer's reasons for bad credit to obtain the label information of the employer's identity include: Obtain a labor violation word library, search for the number and frequency of labor violation words in the text of the bad credit employer's reasons, and obtain the corresponding risk weight according to the number and frequency; Perform TF-IDF analysis on the text of the employer's reasons for bad credit, obtain the TF-IDF value, and weight the TF-IDF value with risk weight to obtain the improved value; LDA is used to perform topic mining on multimodal fusion vectors to generate a topic-vocabulary distribution matrix; According to the improvement value, select the first several high-frequency words as network nodes to build a graph structure; and combine the topic-vocabulary distribution matrix to form a semantic network graph; According to the co-occurrence intensity of words in the topic-vocabulary distribution matrix, a graph convolutional network is used to analyze the association intensity between nodes in the graph structure and construct a weighted semantic network graph; where the nodes are keywords and the edge weights are the association degrees; The core nodes in the weighted semantic network graph are identified using a density peak algorithm; Semantic clustering is performed based on core nodes, and in the clustering process, the clustering radius is dynamically adjusted according to the edge weights in the weighted semantic network graph to obtain semantic clusters; The eccentricity is calculated based on the improvement value and the centrality of the weighted semantic network graph. According to the eccentricity, the first several words are selected from the semantic cluster as the label information of the employer identity.

7. The employment method based on big data according to claim 6 is characterized in that: The step of determining the credit score of each employer identity based on the bad credit employer information specifically includes: Analyze all bad credit employer information for each employer identity and determine the credit type corresponding to the bad credit employer reason; The credit deduction points for each employer with bad credit are determined according to the credit type, and all the credit deduction points are aggregated to obtain the credit score of the corresponding employer identity.

8. The employment method based on big data according to claim 7 is characterized in that: The steps of comprehensively analyzing all job search information and recruitment information to determine employment trend information specifically include: Determine the recruitment quantity for each type of work, the recruitment quantity for each machine type, the job application quantity for each type of work, and the job application quantity for each machine type based on the job application information and recruitment information; Determine the employment trend of each job type based on the number of job openings and job seekers for each job type; Determine the employment trend of each machine type based on the recruitment volume of each machine type and the job search volume of each machine type.

9. A big data-based employment platform, the platform is applied to the big data-based employment method according to any one of claims 1 to 8, characterized in that: The platform includes: A job search information uploading module is used to upload job search information through a job search account, wherein the job search information includes the job seeker's identity, type of work, type of machine that can be operated, dominant hand, age and job search location; A recruitment information uploading module is used to upload recruitment information through a recruitment account, wherein the recruitment information includes the employer's identity, type of work, required machine type, operating dominant hand, age and work location; The job search and recruitment matching module is used to match all job search information with all recruitment information and generate a job seeker-recruiter recommendation list; A bad credit employer information module is used to receive bad credit employer information uploaded by a job-seeking account, wherein the bad credit employer information includes the employer identity and the reason for the bad credit employer; The employer comment area module is used to add label information to the bad credit employer information, build an employer comment area, add the bad credit employer reasons to the employer comment area, and display the label information and the bad credit employer reasons to job seekers; A credit score determination module is used to determine the credit score of each employer based on the information of bad credit employers, adjust the job seeker-employer recommendation list based on the credit score, and determine the order of pushing recruitment information; The employment trend information module is used to conduct a comprehensive analysis of all job search information and recruitment information to determine employment trend information.

Images