Human resource information management system based on big data

Abstract

The invention relates to the technical field of human resource information management, in particular to a human resource information management system based on big data. The system comprises a matching combination unit, an evaluation filtering unit, an evaluation sorting unit and a credibility verification unit. The conversion and combination module carries out keyword matching scoring by utilizing post demand keywords and resume information, converts the resume information and the post demand information into semantic vectors of the resume information and semantic vectors of the post demand information through a pre-training language model, and then combines the semantic vectors with a similarity calculation function Similarity. According to the method, whether appropriate job seekers can be preliminarily screened out from the human resource information management library or not is judged, the defect of keyword matching can be overcome and the misjudgment rate can be reduced through the score of the semantic vector similarity, the situation that potential appropriate job seekers are missed due to the limitation of keyword matching is avoided, and the recruitment efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of human resource information management. Specifically, it relates to a human resource information management system based on big data. Background Art

[0002] Although significant progress has been made in many aspects of human resource information management functions, there are still many challenges in aspects such as data privacy, real-time decision-making, sentiment analysis, cross-cultural management, prediction accuracy, personalized development, legal compliance, system integration, and job screening. For example, a large number of job seeker resumes and recruiter requirement information are stored in the human resource information management database of a recruitment website. Many recruiters match and screen job seekers by setting job requirement keywords with the resume information in the human resource information management database. When the job requirement keywords are missing in the job seeker's resume information, the job position is relatively unpopular and it is impossible to screen out suitable job seekers through keyword matching, indicating that there are deficiencies and limitations in keyword matching, resulting in the loss of potential suitable job seekers due to the limitations of keyword matching, thus reducing the recruitment efficiency. At the same time, when there are limitations in keyword matching, potential candidates matching the job are missed due to fuzzy resume information, resulting in a decrease in the accuracy of screening. Therefore, we provide a human resource information management system based on big data. Summary of the Invention

[0003] The purpose of the present invention is to provide a human resource information management system based on big data to solve the problems raised in the above background art, including:

[0004] 1. Due to the deficiencies and limitations of keyword matching, potential suitable job seekers are missed due to the limitations of keyword matching, thus reducing the recruitment efficiency. Therefore, in this case, the resume information and job requirement information are converted into semantic vectors of the resume information and semantic vectors of the job requirement information, and then combined with the similarity calculation function Similarity to determine whether suitable job seekers can be initially screened out from the human resource information management database, avoiding the loss of potential suitable job seekers due to the limitations of keyword matching;

[0005] 2. Due to potential candidates matching the job being missed due to fuzzy resume information, the accuracy of screening is reduced. Therefore, in this case, the similarity between the resume information and the job requirement information is calculated using the semantic vectors of the resume information and the semantic vectors of the job requirement information, and then the collaborative filtering score of the job seekers is performed based on the similarity between the resume information and the job requirement information and the keyword matching score, screening out candidates for the job, avoiding potential candidates matching the job being missed due to fuzzy resume information;

[0006] To achieve the above object, the present invention provides a human resource information management system based on big data, including a matching and combining unit, an evaluation and filtering unit, an evaluation and sorting unit, and a credibility verification unit;

[0007] As a further improvement of this technical solution, the conversion and combination module (102) uses job requirement keywords and resume information to perform keyword matching scoring. Through keyword matching scoring, the recruitment system can quickly scan resumes, preliminarily screen resumes according to the keyword matching degree, and quickly exclude those resumes that obviously do not meet the basic job requirements, greatly reducing the number of resumes that recruiters need to view and saving a lot of time and effort;

[0008] Secondly, the resume information and job requirement information are converted into semantic vectors of the resume information and semantic vectors of the job requirement information through a pre-trained language model. Traditional keyword matching can only identify whether the words on the surface of the text are the same, while semantic vectors can capture the deep semantic information behind the text;

[0009] Finally, by combining the semantic vectors of the resume information and the semantic vectors of the job requirement information with the similarity calculation function Similarity, a score of the semantic vector similarity is obtained to determine whether a suitable job seeker can be preliminarily screened out from the human resource information management database. Using the semantic vector similarity score for preliminary screening, the computer can quickly process this information, evaluate a large number of resumes in a short time, and quickly exclude obviously unmatched job seekers, greatly shortening the screening cycle and making the recruitment process more efficient.

[0010] As a further improvement of this technical solution, the collaborative filtering module (202) calculates the similarity between the resume information and the job requirement information through the semantic vectors of the resume information and the semantic vectors of the job requirement information, and then performs collaborative filtering scoring of job seekers according to the similarity between the resume information and the job requirement information and the keyword matching score through the collaborative filtering algorithm. Using the collaborative filtering score of job seekers to screen out candidates for the position, and comprehensively evaluating the matching degree between candidates and the position through the collaborative filtering algorithm can more accurately screen out candidates who truly meet the job requirements. This reduces the situation of inviting inappropriate candidates to interviews, avoids the waste of time and resources of both the enterprise and the candidates, and makes the interview process more efficient.

[0011] Compared with the prior art, the beneficial effects of the present invention:

[0012] 1. In the human resource information management system based on big data, the conversion and combination module uses job requirement keywords and resume information for keyword matching scoring. The resume information and job requirement information are converted into semantic vectors of the resume information and semantic vectors of the job requirement information through a pre-trained language model, and then combined with the similarity calculation function Similarity to obtain the scoring of the semantic vector similarity, so as to judge whether suitable job seekers can be preliminarily screened out from the human resource information management database. Through the scoring of the semantic vector similarity, the deficiencies of keyword matching can be made up for and the misjudgment rate can be reduced, avoiding missing potential suitable job seekers due to the limitations of keyword matching, and improving the recruitment efficiency.

[0013] 2. In the human resource information management system based on big data, the collaborative filtering module calculates the similarity between the resume information and the job requirement information through the semantic vectors of the resume information and the semantic vectors of the job requirement information, and then conducts collaborative filtering scoring of job seekers according to the similarity between the resume information and the job requirement information and the keyword matching scoring. The candidates for the job are screened out by using the collaborative filtering scoring of job seekers. Through the collaborative filtering scoring of job seekers, it is possible to avoid missing candidates who match the job due to fuzzy resume information or relying solely on the limitations of keyword matching, and improve the accuracy of screening.

[0014] 3. In the human resource information management system based on big data, the evaluation and matching degree module evaluates the matching degree between candidates and jobs according to the keyword matching scoring, the scoring of the semantic vector similarity, and the collaborative filtering scoring of job seekers. By evaluating the matching degree between candidates and jobs, the most suitable candidates for the job requirements are recommended, providing more personalized job recommendations for candidates, enhancing the candidate experience, and timely discovering candidates who do not explicitly mention but actually possess relevant skills in the resume, mining potential candidates with high matching degrees, and improving the matching accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is the overall structure block diagram of the present invention;

[0016] Figure 2 It is the block diagram of the module unit of the present invention.

[0017] The meanings of the reference numerals in the figure are as follows:

[0018] 100. Matching and combining unit; 101. Keyword matching module; 102. Conversion and combining module;

[0019] 200. Evaluation and filtering unit; 201. Evaluation and matching degree module; 202. Collaborative filtering module;

[0020] 300. Evaluation and sorting unit; 400. Verification and credibility unit. DETAILED DESCRIPTION OF THE INVENTION

[0021] Next, in combination with the accompanying drawings in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Embodiment 1

[0023] The present invention provides a human resource information management system based on big data. Please refer to Figure 1 - Figure 2 , including a matching and combining unit 100, an evaluation and filtering unit 200, an evaluation and sorting unit 300, and a credibility verification unit 400;

[0024] The matching and combining unit 100 obtains job requirement information and extracts job requirement keywords for keyword matching scoring. It converts the job requirement information into a semantic vector of the job requirement information through a pre-trained language model, and then combines it with the similarity calculation function Similarity to determine whether suitable job seekers can be initially screened out;

[0025] The matching and combining unit 100 includes a keyword matching module 101 and a conversion and combination module 102;

[0026] The keyword matching module 101 collects the job requirement information J of the recruiter in real time through big data, extracts the job requirement keywords K from the job requirement information, records the total number n of job requirement keywords, and uses the job requirement keywords to perform keyword matching with the resume information in the human resource information management database. Its algorithm formula is: Match(K i , R); through keyword matching, resumes that obviously do not meet the job keyword requirements can be quickly filtered out. Among them, K i refers to the i-th job requirement keyword. Since Match(K i , R) is a custom matching function used to determine whether the i-th job requirement keyword K i appears in the resume information R, its return value is usually a boolean value, which is a dimensionless numerical value;

[0027] When the i-th job requirement keyword K i appears in the resume information R, Match(K i , R) takes a value of 1, indicating that the job requirement keyword matching is successful. For example, if the job requirement keyword is "nuclear physics direction" and the resume clearly mentions "understanding the nuclear physics direction", then for the job requirement keyword "nuclear physics direction", Match(K i , R) = 1, and the matching is successful;

[0028] When the i-th job requires keyword K i When it does not appear in the resume information R, Match(K i ,R) is 0, which means that the job requirement keyword fails to match; for example, the job requirement keyword is "nuclear physics direction", but the resume information does not mention "nuclear physics direction" at all, then for the job requirement keyword "nuclear physics direction", Match(K i ,R)=0, the match fails, then the position is relatively unpopular and lacks keywords;

[0029] When the conversion and combination module 102 receives a command from the keyword matching module 101 that the job requirement keyword matching fails, it means that the job position is relatively unpopular and it is impossible to match and select suitable job seekers from the human resources information management database through keywords;

[0030] First, the total number of job requirement keywords n, job requirement keywords K, and resume information R in the keyword matching module 101 are used to perform keyword matching scoring. Among them, S keyword The subscript "keyword" has no special physical meaning and is only used to distinguish K i Refers to the keyword required for the i-th position. Since Match(K i ,R) is a dimensionless value, so the keyword matching score is also a dimensionless value. Then the set keyword matching score threshold and the keyword matching score are used to determine whether to trigger the semantic vector generation function. When the set keyword matching score threshold is greater than the keyword matching score, the semantic vector generation function is triggered;

[0031] Secondly, when the command to trigger the semantic vector generation function is known, the conversion and combination module 102 takes the resume information R and the job requirement information J from the keyword matching module 101 and inputs them into the pre-trained language model (such as BERT, Word2Vec, Sentence-BERT). The pre-trained language model uses the semantic vector generation function Embedding to convert the resume information R into the semantic vector v of the resume information. R , the algorithm formula is: R = Embedding(R), the pre-trained language model uses the semantic vector generation function Embedding to convert the job requirement information J into the semantic vector v of the job requirement information J , the algorithm formula is: J =Embedding(J);

[0032] Finally, by combining the semantic vector algorithm formula of resume information, the semantic vector algorithm formula of job requirement information with the similarity calculation function Similarity, the score S of semantic vector similarity is obtained. semantic Through the score of semantic vector similarity, the deficiency of keyword matching can be made up. When some keywords are missing in the resume information of job seekers, but their actual abilities match the job requirements, semantic analysis can provide a more in-depth evaluation, reduce the misjudgment rate, and avoid missing potential suitable job seekers due to the limitations of keyword matching.

[0033] The specific algorithm formula for scoring semantic vector similarity:

[0034] S semantic = Similarity(Embedding(R), Embedding(J));

[0035] Among them, the score of semantic vector similarity is a dimensionless value. The subscript "semantic" of S semantic has no special physical meaning and is only used for distinction. The final results of Embedding(R) and Embedding(J) are numerical values, representing the positions and characteristics of resume information and job requirement information in the vector space, and do not correspond to physical quantities in the real world, so there is no unit.

[0036] Using the score of semantic vector similarity and the set semantic vector similarity score threshold to judge whether suitable job seekers can be preliminarily screened out by the semantic vectors of job requirement information and the semantic vectors of resume information in the human resource information management database. The specific judgment situations include:

[0037] Situation ①: When the score of semantic vector similarity is greater than the set semantic vector similarity score threshold, it is determined that suitable job seekers are preliminarily screened out from the human resource information management database, and the number of job seekers M is recorded. Then, use the number of job seekers and the candidate threshold set for the job to judge whether to trigger the collaborative filtering algorithm.

[0038] If the number of job seekers is greater than the candidate threshold set for the job, it is determined not to trigger the collaborative filtering algorithm, and the job seekers are taken as candidates for the job.

[0039] If the number of job seekers is less than the candidate threshold set for the job, it is determined to trigger the collaborative filtering algorithm.

[0040] Case ②: When the score of the semantic vector similarity is less than the set threshold of the semantic vector similarity score, it is determined that no suitable job seekers have been preliminarily screened out from the human resource information management database. Then, the resume information of the candidate is less or the resume information is vague, resulting in difficulty for semantic analysis to match and screen out suitable job seekers from the human resource information management database. For example, the candidate's resume only simply lists some skill names and lacks specific project practice descriptions, reducing the reliability of the semantic analysis score.

[0041] The evaluation and filtering unit 200 evaluates the matching degree between the candidate and the position through the keyword matching score in the matching and combining unit 100, then calculates the similarity between the resume information and the position requirement information through the semantic vector of the position requirement information in the matching and combining unit 100, uses the collaborative filtering algorithm to perform collaborative filtering scoring on job seekers according to the similarity between the resume information and the position requirement information and the keyword matching score, screens out candidates for the position, and then uses the weighted scoring method to evaluate the matching degree between the candidate and the position according to the keyword matching score and the collaborative filtering score of the job seeker.

[0042] The evaluation and filtering unit 200 includes an evaluation matching degree module 201 and a collaborative filtering module 202.

[0043] The evaluation matching degree module 201 receives the command that does not trigger the collaborative filtering algorithm in the conversion and combining module 102. The evaluation matching degree module 201 obtains the keyword matching score and the score of the semantic vector similarity from the conversion and combining module 102, inputs the keyword matching score and the score of the semantic vector similarity into the linear model. The linear model learns the input data and outputs the corresponding weight coefficients w1 and w2. Using the weighted scoring method, through the keyword matching score S keyword and the score of the semantic vector similarity S semantic as well as the weight coefficients w1 and w2 to evaluate the matching degree between the candidate and the position, and uses the evaluated matching degree and the set matching degree threshold to determine whether the candidate and the position match. When the evaluated matching degree is greater than the set matching degree threshold, it is determined that the candidate and the position match, and the resume information of the candidate and the number of key information in the candidate's resume are recorded. When the evaluated matching degree is less than the set matching degree threshold, it is determined that the candidate and the position do not match, and the candidates who do not match the position are excluded.

[0044] The implementation principle of using the weighted scoring method to evaluate the matching degree between the candidate and the position:

[0045] Collect the keyword matching score S keyword and the score of the semantic vector similarity S semantic as well as the weight coefficients w1 and w2 to evaluate the matching degree between the candidate and the position, and obtain the evaluated matching degree S final . The specific algorithm formula:

[0046] S final = w1·S keyword + w2·S semantic ;

[0047] Among them, S final , w1, and w2 are scores and weights. The scores and weights themselves have no units because they are all normalized values. The subscript "final" in S final has no specific physical meaning and is used for distinction. This formula is used to evaluate the matching degree between candidates and positions.

[0048] The collaborative filtering module 202 receives the command to trigger the collaborative filtering algorithm in the conversion and combination module 102. The collaborative filtering module 202 takes the number of job seekers M, the semantic vector v of the resume information R , and the semantic vector v of the job requirement information J and the keyword matching score S keyword from the conversion and combination module 102. According to the semantic vector v of the resume information R and the semantic vector v of the job requirement information J , it calculates the similarity between the resume information and the job requirement information. Through the collaborative filtering algorithm, based on the similarity Similarity(R,J) between the resume information and the job requirement information, the keyword matching score S keyword and the number of job seekers M, it conducts collaborative filtering scoring for job seekers. It uses the collaborative filtering score of job seekers and the set collaborative filtering score threshold to screen out candidates for the position. When the collaborative filtering score of a job seeker is greater than the set collaborative filtering score threshold, the job seeker is regarded as a candidate for the position, and the resume information of the candidate and the number of key information in the candidate's resume are recorded;

[0049] The matching degree evaluation module 201 inputs the collaborative filtering score S of job seekers in the collaborative filtering module 202 CF into the linear model. The linear model learns the input data and outputs the corresponding weight coefficient w3. Using the weighted scoring method, through the keyword matching score S keyword , the score S of the semantic vector similarity semantic , the collaborative filtering score S of job seekers CF and the weight coefficients w1, w2, and w3, it evaluates the matching degree between the candidate and the position, and obtains the evaluated matching degree S final . The specific algorithm formula: S final = w1·S keyword + w2·S semantic + w3·S CF, determine whether the candidate matches the position by using the evaluated matching degree and the set matching degree threshold. When the evaluated matching degree is greater than the set matching degree threshold, it is determined that the candidate matches the position. When the evaluated matching degree is less than the set matching degree threshold, it is determined that the candidate does not match the position, and the candidates who do not match the position are eliminated;

[0050] The evaluation and sorting unit 300 uses the weighted scoring method to evaluate the matching degree between the job seeker and the position according to the keyword matching score in the matching combination unit 100, sorts them from high to low, and selects the top five job seekers in the ranking as candidates for the position;

[0051] The evaluation and sorting unit 300 receives the command that no suitable job seekers can be initially screened out from the human resource information management database in the conversion and combination module 102. The evaluation and sorting unit 300 inputs the keyword matching score and the semantic vector similarity score in the conversion and combination module 102 into a linear model. The linear model learns the input data and outputs the corresponding weight coefficients w1 and w2. Using the weighted scoring method, through the keyword matching score S keyword , the scoring S of the semantic vector similarity semantic and the weight coefficients w1 and w2 to evaluate the matching degree between the job seeker and the position, and obtain the evaluated matching degree S final , the specific algorithm formula:

[0052] S final = w1·S keyword + w2·S semantic ;

[0053] Use the evaluated matching degree to sort from high to low, select the top five job seekers in the ranking as candidates for the position, and record the resume information of the candidates and the number of key information in the candidate resumes;

[0054] The verification and credibility unit 400 verifies the resume information of the candidate through the verification mechanism in the human resource information management database, and then calculates the verification score and the resume information credibility score;

[0055] The verification and credibility unit 400 receives the resume information of the candidate from the evaluation matching degree module 201 or the collaborative filtering module 202 or the evaluation and sorting unit 300 and the number of key information N in the candidate resume total , where the subscript "total" of N total has no special physical meaning and is used for distinction. Verify the resume information of the candidate through the verification mechanism in the human resource information management database, and obtain the number of verified information N in the resume verified , where the subscript "verified" of N verified has no special physical meaning and is used for distinction. Use the number of candidates N totaland the number N of verified messages verified Calculate the verification score

[0056] When the verification score is known, input the candidate's resume information into the machine learning model. The machine learning model analyzes the language patterns in the resume information (such as overuse of exaggerated vocabulary), identifies suspicious information in the resume information, and obtains the number N of suspicious information ky , where N ky The subscript "ky" in has no special physical meaning and is used for distinction. Use the number N of key information in the candidate's resume total and the number N of suspicious information ky to perform a credibility score on the resume information where the value range is in [0,1], S kxd The subscript "kxd" in has no special physical meaning and is used for distinction;

[0057] The evaluation filtering unit 200 uses the weighted scoring method to evaluate the authenticity score of the candidate's resume according to the resume information credibility score and the verification score in the verification credibility unit 400, and then determines whether there is authenticity or suspicion in the candidate's resume;

[0058] The evaluation matching degree module 201 inputs the resume information credibility score S kxd and the verification score S verify in the verification credibility unit 400 into the linear model. The linear model learns the input data and outputs the corresponding weight coefficients α1 and α2. Use the weighted scoring method through the resume information credibility score S kxd and the verification score S verify and the weight coefficients α1 and α2 to evaluate the authenticity score of the candidate's resume, and obtain the evaluated authenticity score S zs , specific algorithm formula: S zs =α1·S kxd +α2·S verify , use the evaluated authenticity score and the set authenticity score threshold to judge whether there is authenticity or suspicion in the candidate's resume. When the evaluated authenticity score is greater than the set authenticity score threshold, determine that the candidate's resume is authentic, and use this candidate as the final job candidate, and then recommend the final job candidate and the resume information to the recruiter. When the evaluated authenticity score is less than the set authenticity score threshold, determine that there is suspicion in the candidate's resume, and eliminate this candidate.

[0059] The above has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should be in the field of nuclear physics. The present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. A human resource information management system based on big data, characterized in that: It comprises a matching and combining unit (100), an evaluation and filtering unit (200), an evaluation and sorting unit (300) and a verification credibility unit (400); The matching and combining unit (100) obtains job requirement information and extracts job requirement keywords for keyword matching scoring, converts the job requirement information into a semantic vector of the job requirement information through a pre-trained language model, and determines whether suitable job seekers can be preliminarily screened out; The evaluation and filtering unit (200) evaluates the matching degree between the candidate and the position through the keyword matching score in the matching and combining unit (100), calculates the similarity between the resume information and the position requirement information through the semantic vector of the position requirement information in the matching and combining unit (100), and performs collaborative filtering scoring of the job seekers to screen out the candidates for the position, and evaluates the matching degree between the candidate and the position based on the keyword matching score and the collaborative filtering score of the job seekers; The evaluation and ranking unit (300) is used to evaluate the matching degree between job seekers and positions; The verification credibility unit (400) verifies the candidate's resume information through a verification mechanism in a human resources information management library, and then calculates a verification score and a resume information credibility score.

2. The human resource information management system based on big data according to claim 1, wherein: The matching and combining unit (100) comprises a keyword matching module (101) and a conversion and combining module (102); The keyword matching module (101) collects job requirement information of recruiters in real time through big data, extracts job requirement keywords and performs keyword matching with resume information in the human resources information management database to determine whether the job requirement keywords appear in the resume information; When the job requirement keywords appear in the resume information, it means that the job requirement keywords are matched successfully; When the job requirement keywords do not appear in the resume information, it means that the job requirement keywords have failed to match, and the job is relatively unpopular and lacks keywords.

3. The human resource information management system based on big data according to claim 2, wherein: The conversion and combination module (102) receives a command from the keyword matching module (101) indicating that the job requirement keyword matching fails. Since the job position is relatively unpopular, it is impossible to match and select suitable job seekers from the human resources information management database through keywords. First, the job requirement keywords and resume information in the keyword matching module (101) are used to perform keyword matching scoring, and then determine whether to trigger the semantic vector generation function; Secondly, when a command to trigger a semantic vector generation function is known, the conversion and combination module (102) converts the resume information and the job requirement information in the keyword matching module (101) into a semantic vector of the resume information and a semantic vector of the job requirement information through a pre-trained language model; Finally, by combining the semantic vector of the resume information and the semantic vector of the job requirement information with the similarity calculation function Similarity, a semantic vector similarity score is obtained to determine whether suitable job seekers can be preliminarily screened out from the human resources information management database.

4. The human resource information management system based on big data according to claim 3, wherein: The specific judgment situations in the conversion and combination module (102) include: Case ①: When the score of the semantic vector similarity is greater than the set semantic vector similarity score threshold, it is determined that suitable job seekers can be preliminarily screened out from the human resource information management database, and the number of job seekers is recorded to determine whether to trigger the collaborative filtering algorithm; If the number of job seekers is greater than the candidate threshold set for the position, it is determined not to trigger the collaborative filtering algorithm, and the job seekers are used as candidates for the position; If the number of job seekers is less than the candidate threshold set for the position, it is determined to trigger the collaborative filtering algorithm; Case ②: When the score of the semantic vector similarity is less than the set semantic vector similarity score threshold, it is determined that suitable job seekers cannot be preliminarily screened out from the human resource information management database.

5. The human resource information management system based on big data according to claim 3, wherein: The evaluation and filtering unit (200) includes an evaluation matching degree module (201) and a collaborative filtering module (202); The evaluation matching degree module (201) receives the command not to trigger the collaborative filtering algorithm in the conversion and combination module (102). The evaluation matching degree module (201) evaluates the matching degree between the candidate and the position by the keyword matching score and the score of the semantic vector similarity in the conversion and combination module (102), and uses the evaluated matching degree to judge whether the candidate and the position match. When they match, the resume information of the candidate and the number of key information in the candidate's resume are recorded. When they do not match, the candidates who do not match the position are eliminated.

6. The human resource information management system based on big data according to claim 5, characterized in that: The collaborative filtering module (202) receives the command to trigger the collaborative filtering algorithm in the conversion and combination module (102). The collaborative filtering module (202) calculates the similarity between the resume information and the job requirement information by the semantic vectors of the resume information and the semantic vectors of the job requirement information in the conversion and combination module (102), and then performs collaborative filtering scoring of job seekers according to the similarity between the resume information and the job requirement information and the keyword matching score in the conversion and combination module (102) through the collaborative filtering algorithm. The job seekers are screened as candidates for the position by using the collaborative filtering score of the job seekers. When the collaborative filtering score of the job seekers is greater than the set collaborative filtering score threshold, the job seekers are used as candidates for the position, and the resume information of the candidates and the number of key information in the candidate's resume are recorded.

7. The human resource information management system based on big data according to claim 6, wherein: The evaluation matching degree module (201) uses the weighted scoring method to evaluate the matching degree between the candidate and the position according to the keyword matching score, the score of the semantic vector similarity, and the collaborative filtering score of the job seekers in the collaborative filtering module (202); The evaluated matching degree is used to judge whether the candidate and the position match. When they do not match, the candidates who do not match the position are eliminated.

8. The human resource information management system based on big data according to claim 3, characterized in that: The evaluation and sorting unit (300) uses the weighted scoring method to evaluate the matching degree between the job seeker and the position according to the keyword matching score and the score of the semantic vector similarity in the conversion and combination module (102); The job seekers are sorted from high to low according to the evaluated matching degree, and the top five sorted job seekers are selected as candidates for the position, and the resume information of the candidates and the number of key information in the candidate's resume are recorded.

9. The human resource information management system based on big data according to claim 5, characterized in that: The verification credibility unit (400) receives the resume information of candidates from the evaluation matching degree module (201) or the collaborative filtering module (202) or the evaluation ranking unit (300), and the number of key information in the candidate resume, verifies the resume information of candidates through the verification mechanism in the human resource information management database, and calculates the verification score by obtaining the number of verified information in the resume; When the verification score is known, a machine learning model is used to identify suspicious information in the resume information, and then the credibility score of the resume information is obtained.

10. The human resource information management system based on big data according to claim 9, characterized in that: The evaluation matching degree module (201) uses the weighted scoring method to evaluate the authenticity score of the candidate resume according to the credibility score and verification score of the resume information in the verification credibility unit (400), and uses the evaluated authenticity score to judge whether the candidate resume is true or suspicious. When the resume is true, the candidate is regarded as the final candidate for the position. When the resume is suspicious, the candidate is eliminated.

Images