Personnel matching method and system based on artificial intelligence
By analyzing the keywords in the communication records between job seekers and recruiters, false attributes in the recruitment position are identified, and the matching inaccuracy caused by false attributes in the existing technology is solved, and the efficiency and accuracy of the recruitment and job hunting process are improved. sex.
Patent Information
- Application Number
- CN202411000443.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology cannot identify false job attributes in recruitment positions, resulting in an increase in communication time cost between job seekers and recruiters, and the accuracy of personnel matching is not high.
By collecting keywords in the communication records between job seekers and recruiters, using NLP technology and clustering algorithms to convert keywords into vectors, input them into the recognition model for processing, identifying intentions and recruitment keyword tags, and comparing recruitment attribute tags with Job attributes, replace false attributes, and enter the pre-trained personnel matching model to match.
It realizes accurate identification of recruitment positions, improves the accuracy of personnel matching, reduces the communication time cost between job seekers and recruiters, and improves the efficiency of the recruitment and job hunting process.
Smart Images

Figure CN119025620A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of personnel matching technology, and more specifically, to an artificial intelligence-based personnel matching method and system. Background Art
[0002] In the current recruitment environment, companies are often faced with thousands of resumes, and human resources departments need to spend a lot of time and energy to screen and match suitable personnel.
[0003] In order to improve the efficiency and accuracy of personnel screening and matching, the existing application publication number CN1 16578933A proposes a personnel resume matching method and system based on artificial intelligence big data, the method comprising: S100, obtaining resume information and recruitment demand information, and collecting a first data set in the resume information and a second data set in the recruitment demand information through a preset intelligent model; S200, performing big data collection based on the first data set and the second data set, respectively, to obtain a first associated data set and a second associated data set; S300, inputting the first associated data set and the second associated data set into the training model, respectively, to obtain a first label tree and a second label tree; evaluating the matching degree of the resume information and the recruitment demand information based on the comparison of the first label tree and the second label tree; the method can realize the matching degree evaluation process of the resume information and the professionalism of the recruitment demand, and assist the efficient screening process of recruiters and applicants.
[0004] Although the above method can meet most scenarios, research and practical application of the above method and existing technology have found that the above method and existing technology have at least the following defects:
[0005] In the actual job search process, existing technologies are unable to identify false job attributes in recruitment positions. For example, the recruitment position attribute is a technical position, but the actual recruitment is for a sales position; the recruitment position attribute is a two-day weekend off, but the actual recruitment is a half-day weekend shift. False job attributes of positions mislead job seekers, increase the communication time cost between job seekers and recruiters, and result in low accuracy in personnel matching.
[0006] In view of this, the present invention proposes an artificial intelligence-based personnel matching method and system to solve the above-mentioned problems. Summary of the invention
[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a personnel matching method based on artificial intelligence, comprising:
[0008] Collect resume content of job seekers;
[0009] Obtaining a keyword set in the communication record between the job seeker and the recruiter; the keyword set includes an intended keyword set and a recruitment keyword set;
[0010] The intention keyword set and the recruitment keyword set are processed separately to obtain the intention attribute label set and the recruitment attribute label set;
[0011] Collecting basic job information, including job requirements and job attributes;
[0012] Compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts;
[0013] The personnel matching data set is input into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job seeker end; the personnel matching data set includes basic position information, resume content and intention attribute label set.
[0014] Furthermore, the method for obtaining the keyword set is:
[0015] S131: Anonymize communication records;
[0016] S132: pre-define a stop word list, and remove corresponding words in the communication record according to the stop word list;
[0017] S133: using NLP technology to remove punctuation marks and special characters from the communication records, and segmenting the communication records to obtain a keyword set;
[0018] S134: Divide the keyword set into an intention keyword set and a recruitment keyword set according to the sending subject; the sending subject corresponding to the intention keyword set is the job seeker, and the sending subject corresponding to the recruitment keyword set is the recruitment end.
[0019] Furthermore, the method for obtaining the intention attribute tag set includes:
[0020] S141: using Word2Vec to convert keywords in the intended keyword set into intended keyword vectors;
[0021] S142: using a clustering algorithm to classify similar intention keyword vectors to obtain an intention keyword vector set;
[0022] S143: inputting the classified intention keyword vector into the intention recognition model to obtain the intention attribute label corresponding to the intention keyword vector;
[0023] The training method of the intention recognition model includes:
[0024] Pre-collect an intention label data set, wherein the intention label data includes an intention keyword vector and an intention attribute label corresponding to the intention keyword vector, divide the intention label data set into a training set and a test set, construct a classifier, use the intention keyword vector in the training set as the input of the intention recognition model, use the intention attribute label in the training set as the output of the intention recognition model, train the classifier to obtain an initial classifier, test the initial classifier using the test set, and output a classifier that meets a preset accuracy as the intention recognition model, wherein the classifier is a naive Bayes model or a support vector machine model;
[0025] S144: Add the intent attribute label to the intent attribute label set.
[0026] Furthermore, the method for obtaining the recruitment attribute tag set includes:
[0027] S151: Use Word2Vec to convert keywords in the recruitment keyword set into recruitment keyword vectors;
[0028] S152: using a clustering algorithm to classify similar recruitment keyword vectors to obtain a recruitment keyword vector set;
[0029] S153: inputting the classified recruitment keyword vector into a recruitment recognition model to obtain a recruitment keyword label corresponding to the recruitment keyword vector;
[0030] The training method of the recruitment recognition model includes:
[0031] Pre-collecting a recruitment label data set, wherein the recruitment label data includes a recruitment keyword vector and a recruitment keyword label corresponding to the recruitment keyword vector, dividing the recruitment label data set into a training set and a test set, constructing a classifier, using the recruitment keyword vector in the training set as the input of a recruitment recognition model, using the recruitment keyword label in the training set as the output of the recruitment recognition model, training the classifier to obtain an initial classifier, testing the initial classifier using the test set, and outputting a classifier that meets a preset accuracy as a recruitment recognition model, wherein the recruitment recognition model is a naive Bayes model or a support vector machine model;
[0032] S154: Add the recruitment keyword tag to the recruitment attribute tag set.
[0033] Furthermore, the training method of the personnel matching model includes:
[0034] Pre-collecting a job matching data set, wherein the job matching data set includes a personnel matching data set and matching label values corresponding to the personnel matching data set, dividing the job matching data set into a training set and a test set, constructing a classifier, using the personnel matching data set in the training set as the input of the personnel matching model, using the matching label values in the training set as the output of the personnel matching model, and minimizing the sum of the prediction accuracies of all predicted matching label values as the training goal; stopping the training when the sum of the prediction accuracies reaches convergence; the personnel matching model is an RNN model;
[0035] The calculation formula for prediction accuracy is: L z =(Y z -S z ) 2 , where L z is the prediction accuracy, Y z is the predicted matching label value corresponding to the matching data set of the zth group of people, S z The actual matching label value corresponding to the matching data set for the zth group of people.
[0036] Furthermore, the personnel matching data set also includes a set of intended positions, and the method for obtaining the set of intended positions includes:
[0037] Collect job seekers’ job browsing records and resume submission records;
[0038] Process the job browsing records and resume delivery records to obtain a job browsing record set and a resume delivery record set;
[0039] The job browsing record set and the resume submission record set are processed to obtain the intended job set.
[0040] Furthermore, the method for obtaining the post browsing record set is:
[0041] S101: Obtaining job browsing records;
[0042] S102: Grouping the browsed job titles in the acquired job browsing records, that is, grouping the same browsed job titles into one group;
[0043] S103: Count the number of views corresponding to the browsed job titles in each group;
[0044] S104: sorting the browsing times of all groups;
[0045] S105: Add the browsed job titles and their corresponding browsed times to a job browsed record set in descending order of the browsed times; the job browsed record set includes m browsed record subsets, which are expressed as follows:
[0046] {{L1:M1}, {L2:M2}...{L m :M m}};
[0047] L m is the browsing position name in the mth browsing record subset; M m is the number of views of the job title in the mth browsing record subset; M1>M2>M m .
[0048] Furthermore, the method for obtaining the resume delivery record set is:
[0049] S111: Obtain resume delivery records;
[0050] S112: Grouping the job titles in the obtained resume delivery records, that is, grouping the same job titles into one group;
[0051] S113: Count the number of submissions corresponding to the submission job names in each group;
[0052] S114: sorting the delivery times of all groups;
[0053] S115: Add the job titles and their corresponding submission times to a resume submission record set in descending order of submission times; the resume submission record set includes n resume submission record subsets, which are expressed as follows:
[0054] {{T1:N1},{T2:N2}...{T n :N n}};
[0055] T n The name of the position submitted in the nth resume submission record subset; N n is the number of submissions of the post name in the nth resume submission record subset; N1>N2>N n .
[0056] Furthermore, the method for obtaining the intended job set is:
[0057] S121: The initial value of i is preset to be 1, i∈m; the weight of the number of views is preset to be The weight of the preset delivery times is
[0058] S122: Obtain the i-th browsing record sub-set in the job browsing record set, and obtain the browsing job L in the browsing record sub-set i and M i ; L i is the browsing position name in the i-th browsing record subset, Mi is the number of views of the job title in the i-th browsing record subset;
[0059] S123: Determine whether there is a match with L in the resume delivery record set i Same job title as L; if it exists, get the same job title as L from the resume delivery record subset i The number of delivery times corresponding to the same delivery position name is recorded as TD; if it does not exist, TD is set to 0;
[0060] S124: M i Substitute TD into the job intention score formula to get the intention score of browsing job i;
[0061] The job intention scoring formula is as follows:
[0062]
[0063] For L i Job intention score;
[0064] S125: L i and L i The intention score is added to the temporary data table; let i=i+1, if i is less than or equal to m, then continue to execute S122 to S124; if i is equal to m, then execute S126;
[0065] S126: sorting the temporary data table from high to low according to the intention scores of the browsed positions, and adding the browsed position names with the top r position intention scores to the intended position set;
[0066] S127: Determine whether the names of the k most viewed positions in the position browsing record set are in the intended position set. If not, add the names of the viewed positions that are not in the intended position set to the intended position set; determine whether the names of the k most submitted positions in the resume submission record set are in the intended position set. If not, add the submitted positions that are not in the intended position set to the intended position set.
[0067] The personnel matching system based on artificial intelligence implements the personnel matching method based on artificial intelligence, including:
[0068] The first collection module is used to collect the resume content of job seekers;
[0069] The second acquisition module is used to obtain a keyword set in the communication records between job seekers and recruiters; the keyword set includes an intention keyword set and a recruitment keyword set;
[0070] The first processing module is used to process the intention keyword set and the recruitment keyword set respectively to obtain the intention attribute tag set and the recruitment attribute tag set;
[0071] The third collection module is used to collect basic information of the position, and the basic information of the position includes the requirements and attributes of the recruitment position;
[0072] The second processing module is used to compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts;
[0073] The personnel matching module inputs the personnel matching data set into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job application end; the personnel matching data set includes basic position information, resume content and intention attribute label set;
[0074] The fourth collection module is used to collect job seekers' job browsing records and resume submission records;
[0075] The third processing module is used to process the job browsing records and resume submission records to obtain a job browsing record set and a resume submission record set;
[0076] The fourth processing module is used to process the job browsing record set and the resume submission record set to obtain the intended job set; the personnel matching data set also includes the intended job set.
[0077] Technical effects and advantages of the personnel matching method and system based on artificial intelligence of the present invention:
[0078] By analyzing the keywords in the communication records of job seekers, we can more accurately infer the job attributes that job seekers value; by analyzing the keywords in the communication records of recruiters, we can identify the false job attributes in the recruitment positions, and match personnel with the real recruitment position attributes, so that the recommended resumes and recruitment positions are highly correlated, which improves the accuracy of personnel matching, reduces the communication time cost of both job seekers and recruiters, improves the user experience of recruiters and job seekers, reduces repetitive work, and makes the recruitment process and job search process smoother and more efficient. BRIEF DESCRIPTION OF THE DRAWINGS
[0079] Figure 1 This is a schematic diagram of an artificial intelligence-based personnel matching system according to Embodiment 1 of the present invention;
[0080] Figure 2 This is a schematic diagram of an artificial intelligence-based personnel matching system according to Embodiment 2 of the present invention;
[0081] Figure 3This is a flow chart of a personnel matching method based on artificial intelligence according to Embodiment 3 of the present invention;
[0082] Figure 4 It is a schematic diagram of the electronic device of the present invention;
[0083] Figure 5 Schematic diagram of the storage medium of the present invention. DETAILED DESCRIPTION
[0084] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0085] Example 1
[0086] See also Figure 1 As shown, the artificial intelligence-based personnel matching method and system described in this embodiment include a first acquisition module, a second acquisition module, a first processing module, a third acquisition module, a second processing module and a personnel matching module, and each module realizes data transmission through wired and / or wireless connections.
[0087] The first collection module is used to collect the resume content of the job seeker; the resume content includes expected salary, education background, graduation school, work experience, and work skills.
[0088] The intention data is obtained from the background database of the recruitment platform.
[0089] The second collection module is used to obtain a set of keywords in the communication records between job seekers and recruiters; the keywords are collected on the premise that job seekers and recruiters agree to the privacy policy and terms of use of the job search platform; the keyword set includes an intention keyword set and a recruitment keyword set; the intention keyword set is a set of keywords reflected when job seekers communicate, and the recruitment keyword set is a set of keywords reflected when recruiters communicate.
[0090] The method for obtaining the keyword set is:
[0091] S131: Anonymize the communication records; for example, replace the real name, contact information and address in the communication records with "[name]", "[contact information]" and "[address]".
[0092] An example of a communication record between a recruiter and a job seeker is as follows:
[0093] Recruiter: "Hello, please provide your name and contact information, we will contact you later."
[0094] Job seeker: "Xiao Ming, 12345678910."
[0095] An example of anonymized communication records is as follows:
[0096] Recruiter: "Hello, please provide your name and contact information, we will contact you later."
[0097] Job Applicant: “[Name], [Contact Information]”.
[0098] It should be noted that anonymizing communication records can avoid collecting private information, and only the anonymized information will be analyzed subsequently.
[0099] S132: pre-define a stop word list, and remove corresponding words in the communication record according to the stop word list; the stop word list includes "I", "of", "is", "this", etc.
[0100] S133: Use NLP technology (such as jieba, NLTK, spaCy, etc.) to remove punctuation marks and special characters from the communication records, and segment the communication records to obtain a keyword set; for example, after segmenting "I like this job with two days off on weekends", we get {{like}, {weekend}, {two days off}, {job}}; the special characters include "()", ")" and "——", etc.
[0101] S134: Divide the keyword set into an intention keyword set and a recruitment keyword set according to the sending subject; the sending subject corresponding to the intention keyword set is the job seeker, and the sending subject corresponding to the recruitment keyword set is the recruitment end.
[0102] The first processing module is used to process the intention keyword set and the recruitment keyword set respectively to obtain the intention attribute tag set and the recruitment attribute tag set.
[0103] The method for obtaining the intention attribute tag set includes:
[0104] S141: Use Word2Vec to convert keywords in the intended keyword set into intended keyword vectors; Word2Vec is a word embedding technology.
[0105] S142: Use the K-means clustering algorithm to classify similar intention keyword vectors to obtain an intention keyword vector set.
[0106] S143: Input the classified intention keyword vector into the intention recognition model to obtain the intention attribute label corresponding to the intention keyword vector; for example, the intention attribute label corresponding to "no work on weekends", "two days off", and "two days off on weekends" is two days off.
[0107] The training method of the intention recognition model includes:
[0108] An intention label data set is collected in advance, where the intention label data includes an intention keyword vector and an intention attribute label corresponding to the intention keyword vector. The intention label data set is divided into a training set and a test set, and a classifier is constructed. The intention keyword vector in the training set is used as the input of the intention recognition model, and the intention attribute label in the training set is used as the output of the intention recognition model. The classifier is trained to obtain an initial classifier, and the initial classifier is tested using the test set. A classifier that meets a preset accuracy is output as an intention recognition model, and the classifier is a naive Bayes model or a support vector machine model.
[0109] S144: Add the intent attribute label to the intent attribute label set.
[0110] The method for obtaining the recruitment attribute tag set includes:
[0111] S151: Use Word2Vec to convert keywords in the recruitment keyword set into recruitment keyword vectors.
[0112] S152: Use the clustering algorithm K-means to classify similar recruitment keyword vectors to obtain a recruitment keyword vector set.
[0113] S153: Input the classified recruitment keyword vector into the recruitment recognition model to obtain the recruitment keyword label corresponding to the recruitment keyword vector; for example, the recruitment keyword label corresponding to "scheduling", "half-day weekend shift", and "big and small week" is no two days off.
[0114] The training method of the recruitment recognition model includes:
[0115] A recruitment label data set is collected in advance, wherein the recruitment label data includes a recruitment keyword vector and a recruitment keyword label corresponding to the recruitment keyword vector. The recruitment label data set is divided into a training set and a test set, and a classifier is constructed. The recruitment keyword vector in the training set is used as the input of a recruitment recognition model, and the recruitment keyword label in the training set is used as the output of the recruitment recognition model. The classifier is trained to obtain an initial classifier, and the initial classifier is tested using a test set. The classifier that meets the preset accuracy is output as a recruitment recognition model, and the recruitment recognition model is a naive Bayes model or a support vector machine model.
[0116] S154: Add the recruitment keyword tag to the recruitment attribute tag set.
[0117] The third collection module is used to collect basic information of the position, which includes the requirements and attributes of the recruitment position; the requirements include the title of the recruitment position, salary, education background, graduation school, work experience, and work skills; the attributes of the recruitment position include salary, education background, working hours, whether there are two days off a week, work experience, work location, work skills, and the nature of the recruitment; the nature of the recruitment includes full-time recruitment, part-time recruitment, internship recruitment, and outsourced recruitment.
[0118] The basic information of the position is set by the recruiter and obtained through the settings of the recruiter.
[0119] The second processing module is used to compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts, so as to obtain the updated recruitment position attributes.
[0120] It should be noted that the recruitment attribute label set is compared with the recruitment position attributes. If there are inconsistencies, they are false position attributes in the recruitment position. Therefore, the recruitment keyword labels in the recruitment attribute label set are used to replace the recruitment position attributes, so as to obtain the real recruitment position attributes and then perform personnel matching, so as to achieve a high correlation between the recommended resumes and recruitment positions.
[0121] For example: if the recruitment position attribute set by the recruiter is "two days off a week", but "no two days off a week" exists in the recruitment attribute tag set, then "two days off a week" in the recruitment position attribute will be replaced with "no two days off a week"; if the recruitment position attribute set by the recruiter is "full-time recruitment", but "outsourcing" exists in the recruitment attribute tag set, then "full-time recruitment" in the recruitment position attribute will be replaced with "outsourced recruitment".
[0122] The personnel matching module inputs the personnel matching data set into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job seeker end; the personnel matching data set includes basic position information, resume content and intention attribute label set; the matching label value includes 0 and 1, 0 indicates no match, and 1 indicates a match.
[0123] The training method of the personnel matching model includes:
[0124] A job matching data set is collected in advance, wherein the job matching data set includes a personnel matching data set and matching label values corresponding to the personnel matching data set. The job matching data set is divided into a training set and a test set, a classifier is constructed, the personnel matching data set in the training set is used as the input of the personnel matching model, and the matching label values in the training set are used as the output of the personnel matching model, and the training goal is to minimize the sum of the prediction accuracies of all predicted matching label values; the training is stopped when the sum of the prediction accuracies converges; the personnel matching model is an RNN model.
[0125] The calculation formula for prediction accuracy is: L z =(Y z -S z ) 2 , where L z is the prediction accuracy, Y z is the predicted matching label value corresponding to the matching data set of the zth group of people, S z The actual matching label value corresponding to the matching data set for the zth group of people.
[0126] The method for obtaining the matching tag value is:
[0127] S161: Count the number of recruitment job requirements, recorded as H; match the recruitment job requirements with the intended job set and resume content in sequence, and count the number of successful matches, recorded as G.
[0128] S162: Count the number of recruitment position attributes, recorded as U; match the recruitment position attributes with the intended attribute tag set in sequence, and count the number of successful matches, recorded as Y.
[0129] S163: Calculate the matching rate. The calculation formula of the matching rate is:
[0130]
[0131] The sum of the number of recruitment job requirements H and the number of recruitment job attributes U is used as the denominator, reflecting the number of overall recruitment job requirements. The number of successful matches between recruitment job requirements G and the number of successful matches between recruitment job attributes Y are used as the numerator, reflecting the number of people who meet the overall recruitment job requirements. The higher the sum of G and Y and the larger the numerator, the higher the matching rate and the more matched the person is to the job.
[0132] S164: Preset a matching rate threshold β. If PP is greater than or equal to β, the matching label value is 1; if PP is less than β, the matching label value is 0.
[0133] Example 2
[0134] See also Figure 2As shown, this embodiment provides a personnel matching system based on artificial intelligence, and also includes:
[0135] The fourth collection module is used to collect job seekers' job browsing records and resume submission records.
[0136] The job browsing records and resume submission records are obtained from the background database of the recruitment platform.
[0137] The third processing module is used to process the job browsing records and resume submission records to obtain a job browsing record set and a resume submission record set.
[0138] The method for obtaining the post browsing record set is:
[0139] S101: Obtain job browsing records.
[0140] S102: Grouping the browsed job titles in the acquired job browsing records, that is, grouping the same browsed job titles into one group.
[0141] S103: Count the number of views corresponding to the browsed job titles in each group.
[0142] S104: Sort the browsing times of all groups.
[0143] S105: Add the browsed job titles and their corresponding browsed times to a job browsed record set in descending order of the browsed times; the job browsed record set includes m browsed record subsets, which are expressed as follows:
[0144] {{L1:M1}, {L2:M2}...{L m :M m}}.
[0145] It should be noted that L m is the browsing position name in the mth browsing record subset; M m is the number of views of the job title in the mth browsing record subset; M1>M2>M m ; The collection of job browsing records can reflect the job seekers' attention to different jobs; the more browsing times, the higher the attention to the job.
[0146] The method for obtaining the resume delivery record set is:
[0147] S111: Obtain resume submission records.
[0148] S112: Grouping the job titles in the obtained resume delivery records, that is, grouping the same job titles into one group.
[0149] S113: Count the number of submissions corresponding to the job title in each group.
[0150] S114: Sort the delivery times of all groups.
[0151] S115: Add the job titles and their corresponding submission times to a resume submission record set in descending order of submission times; the resume submission record set includes n resume submission record subsets, which are expressed as follows:
[0152] {{T1:N1},{T2:N2}...{T n :N n}}.
[0153] It should be noted that T n The name of the position submitted in the nth resume submission record subset; N n is the number of submissions of the post name in the nth resume submission record subset; N1>N2>N n ; The resume submission record collection can reflect the job seekers' attention to different positions; the more submission times, the higher the attention to the position.
[0154] The fourth processing module is used to process the job browsing record set and the resume submission record set to obtain the intended job set. The personnel matching data set also includes the intended job set.
[0155] The intended job set is expressed as follows:
[0156] {YX1, YX2...YX x}.
[0157] It should be noted that YX x It is the name of the xth intended position in the intended position set.
[0158] The method for obtaining the intended job set is:
[0159] S121: The initial value of i is preset to be 1, i∈m; the weight of the number of views is preset to be The weight of the preset delivery times is.
[0160] S122: Obtain the i-th browsing record sub-set in the job browsing record set, and obtain the browsing job L in the browsing record sub-set i and M i ; L i is the browsing position name in the i-th browsing record subset, M i is the number of views of the job title in the i-th browsing record subset.
[0161] S123: Determine whether there is a match with L in the resume delivery record set i Same job title as L; if it exists, get the same job title as L from the resume delivery record subset i The number of delivery times corresponding to the same delivery position name is recorded as TD; if it does not exist, TD is set to 0.
[0162] S124: M i Substitute TD into the job intention score formula to get the intention score of browsing job i;
[0163] The job intention scoring formula is as follows:
[0164]
[0165] in, For L i Job intention score.
[0166] It should be noted that the weight factor and weight factor All are greater than 0, the number of views is M i The higher the score, the higher the job seeker's intention for the position, and vice versa. The higher the number of submissions TD, the higher the job seeker's intention for the position. The weighted sum of the comprehensive browsing times and the submission times can more accurately reflect the job seeker's intention. The higher the job intention score, the higher the job seeker's intention for the position.
[0167] S125: L i and L i The intention score is added to the temporary data table; let i=i+1, if i is less than or equal to m, continue to execute S122 to S124; if i is equal to m, execute S126.
[0168] S126: Sort the temporary data table from high to low according to the intention scores of the browsed positions, and add the browsed position names with the top r position intention scores to the intended position set.
[0169] S127: Determine whether the names of the k most viewed positions in the position browsing record set are in the intended position set. If not, add the names of the viewed positions that are not in the intended position set to the intended position set; determine whether the names of the k most submitted positions in the resume submission record set are in the intended position set. If not, add the submitted positions that are not in the intended position set to the intended position set.
[0170] It should be noted that there is a position whose browsing count ranks in the top k in the job browsing record set, but because the attributes of the position do not meet the job seeker's expectations, such as expected salary, education, graduation school, work experience or work skills, that is, no resume has been submitted or the number of submissions is small, resulting in the position intention score not being in the top r. However, this scenario can infer that the job seeker has a high intention for the position, so the position is added to the intended position set.
[0171] Similarly, there is a job with the number of submissions in the top k, but the number of jobs pushed to job seekers is small, and the number of views is not high, resulting in the score of the intended job not being in the top r. However, this scenario can infer that the job seeker has a high degree of intention for the job, so the job is added to the set of intended jobs.
[0172] By analyzing job seekers' job browsing records and resume submission records, we can more accurately infer the job seekers' real job intentions, thereby avoiding mismatches caused by relying solely on the intended positions in the resume.
[0173] Example 3
[0174] See also Figure 3 As shown, this embodiment provides a personnel matching method based on artificial intelligence, including:
[0175] Collect resume content of job seekers;
[0176] Obtaining a keyword set in the communication record between the job seeker and the recruiter; the keyword set includes an intended keyword set and a recruitment keyword set;
[0177] The intention keyword set and the recruitment keyword set are processed separately to obtain the intention attribute label set and the recruitment attribute label set;
[0178] Collecting basic job information, including job requirements and job attributes;
[0179] Compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts;
[0180] The personnel matching data set is input into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job seeker end; the personnel matching data set includes basic position information, resume content and intention attribute label set.
[0181] Example 4
[0182] See also Figure 4As shown, according to another aspect of the present application, an electronic device 500 is also provided. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the above-mentioned personnel matching method based on artificial intelligence can be executed.
[0183] The method or system according to the embodiment of the present application can also be used by Figure 4 The electronic device architecture shown in FIG. Figure 4 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, ROM 503, RAM 504, a communication port 505 connected to a network, an input / output 506, a hard disk 507, etc. The storage device in the electronic device 500, such as ROM 503 or hard disk 507, may store the personnel matching method based on artificial intelligence provided by the present application. Furthermore, the electronic device 500 may also include a user interface 508. Of course, Figure 4 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 3 One or more components of an electronic device are shown.
[0184] Example 5
[0185] See also Figure 5 As shown, a computer-readable storage medium 600 according to an embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 600. When the computer-readable instructions are executed by the processor, the personnel matching method based on artificial intelligence according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium 600 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.
[0186] In addition, according to the embodiments of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided by the present application, such as: a personnel matching method based on artificial intelligence. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.
[0187] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
[0188] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. The personnel matching method based on artificial intelligence is characterized by: include: Collect resume content of job seekers; Obtain a set of keywords from the communication records between job seekers and recruiters; The keyword set includes an intention keyword set and a recruitment keyword set; The intention keyword set and the recruitment keyword set are processed separately to obtain the intention attribute label set and the recruitment attribute label set; Collecting basic job information, including job requirements and job attributes; Compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts; The personnel matching data set is input into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job seeker end; the personnel matching data set includes basic position information, resume content and intention attribute label set.
2. The personnel matching method based on artificial intelligence according to claim 1, characterized in that: The method for obtaining the keyword set is: S131: Anonymize communication records; S132: pre-define a stop word list, and remove corresponding words in the communication record according to the stop word list; S133: using NLP technology to remove punctuation marks and special characters from the communication records, and segmenting the communication records to obtain a keyword set; S134: Divide the keyword set into an intention keyword set and a recruitment keyword set according to the sending subject; the sending subject corresponding to the intention keyword set is the job seeker, and the sending subject corresponding to the recruitment keyword set is the recruitment end.
3. The personnel matching method based on artificial intelligence according to claim 2 is characterized in that: The method for obtaining the intention attribute tag set includes: S141: using Word2Vec to convert keywords in the intended keyword set into intended keyword vectors; S142: using a clustering algorithm to classify similar intention keyword vectors to obtain an intention keyword vector set; S143: inputting the classified intention keyword vector into the intention recognition model to obtain the intention attribute label corresponding to the intention keyword vector; The training method of the intention recognition model includes: Pre-collect an intention label data set, wherein the intention label data includes an intention keyword vector and an intention attribute label corresponding to the intention keyword vector, divide the intention label data set into a training set and a test set, construct a classifier, use the intention keyword vector in the training set as the input of the intention recognition model, use the intention attribute label in the training set as the output of the intention recognition model, train the classifier to obtain an initial classifier, test the initial classifier using the test set, and output a classifier that meets a preset accuracy as the intention recognition model, wherein the classifier is a naive Bayes model or a support vector machine model; S144: Add the intent attribute label to the intent attribute label set.
4. The personnel matching method based on artificial intelligence according to claim 3 is characterized in that: The method for obtaining the recruitment attribute tag set includes: S151: Use Word2Vec to convert keywords in the recruitment keyword set into recruitment keyword vectors; S152: using a clustering algorithm to classify similar recruitment keyword vectors to obtain a recruitment keyword vector set; S153: inputting the classified recruitment keyword vector into a recruitment recognition model to obtain a recruitment keyword label corresponding to the recruitment keyword vector; The training method of the recruitment recognition model includes: Pre-collecting a recruitment label data set, wherein the recruitment label data includes a recruitment keyword vector and a recruitment keyword label corresponding to the recruitment keyword vector, dividing the recruitment label data set into a training set and a test set, constructing a classifier, using the recruitment keyword vector in the training set as the input of a recruitment recognition model, using the recruitment keyword label in the training set as the output of the recruitment recognition model, training the classifier to obtain an initial classifier, testing the initial classifier using the test set, and outputting a classifier that meets a preset accuracy as a recruitment recognition model, wherein the recruitment recognition model is a naive Bayes model or a support vector machine model; S154: Add the recruitment keyword tag to the recruitment attribute tag set.
5. The personnel matching method based on artificial intelligence according to claim 4 is characterized in that: The training method of the personnel matching model includes: A job matching data set is collected in advance, wherein the job matching data set includes a personnel matching data set and matching label values corresponding to the personnel matching data set. The job matching data set is divided into a training set and a test set, a classifier is constructed, the personnel matching data set in the training set is used as the input of the personnel matching model, and the matching label values in the training set are used as the output of the personnel matching model, and the training goal is to minimize the sum of the prediction accuracies of all predicted matching label values; the training is stopped when the sum of the prediction accuracies converges; the personnel matching model is an RNN model.
6. The personnel matching method based on artificial intelligence according to claim 1, characterized in that: The personnel matching data set also includes a set of intended positions, and a method for obtaining the set of intended positions includes: Collect job seekers’ job browsing records and resume submission records; Process the job browsing records and resume delivery records to obtain a job browsing record set and a resume delivery record set; The job browsing record set and the resume submission record set are processed to obtain the intended job set.
7. The personnel matching method based on artificial intelligence according to claim 6, characterized in that: The method for obtaining the post browsing record set is: S101: Obtaining job browsing records; S102: Grouping the browsed job titles in the acquired job browsing records, that is, grouping the same browsed job titles into one group; S103: Count the number of views corresponding to the browsed job titles in each group; S104: sorting the browsing times of all groups; S105: Add the browsed job titles and their corresponding browsed times to a job browsed record set in descending order of the browsed times; the job browsed record set includes m browsed record subsets, which are expressed as follows: {{L1:M1},{L2:M2}...{L m :M m }}; L m is the browsing position name in the mth browsing record subset; M m is the number of views of the job title in the mth browsing record subset; M1>M2>M m .
8. The personnel matching method based on artificial intelligence according to claim 7, characterized in that: The method for obtaining the resume delivery record set is: S111: Obtain resume delivery records; S112: Grouping the job titles in the obtained resume delivery records, that is, grouping the same job titles into one group; S113: Count the number of submissions corresponding to the submission job names in each group; S114: sorting the delivery times of all groups; S115: Add the job titles and their corresponding submission times to a resume submission record set in descending order of submission times; the resume submission record set includes n resume submission record subsets, which are expressed as follows: {{T1:N1],{T2:N2}...{T n :N n }}; T n The name of the position submitted in the nth resume submission record subset; N n is the number of submissions of the post name in the nth resume submission record subset; N1>N2>N n .
9. The personnel matching method based on artificial intelligence according to claim 8, characterized in that: The method for obtaining the intended job set is: S121: The initial value of i is preset to be 1, i∈m; the weight of the number of views is preset to be The weight of the preset delivery times is S122: Obtain the i-th browsing record sub-set in the job browsing record set, and obtain the browsing job L in the browsing record sub-set i and M i ; L i is the browsing position name in the i-th browsing record subset, M i is the number of views of the job title in the i-th browsing record subset; S123: Determine whether there is a match with L in the resume delivery record set i Same job title as L; if it exists, get the same job title as L from the resume delivery record subset i The number of delivery times corresponding to the same delivery position name is recorded as TD; if it does not exist, TD is set to 0; S124: M i Substitute TD into the job intention score formula to get the intention score of browsing job i; The job intention scoring formula is as follows: For L i Job intention score; S125: L i and L i The intention score is added to the temporary data table; Let i=i+1, if i is less than or equal to m, continue to execute S122 to S124; If i is equal to m, execute S126; S126: sorting the temporary data table from high to low according to the intention scores of the browsed positions, and adding the browsed position names with the top r position intention scores to the intended position set; S127: Determine whether the names of the k most viewed positions in the position browsing record set are in the intended position set. If not, add the names of the viewed positions that are not in the intended position set to the intended position set; determine whether the names of the k most submitted positions in the resume submission record set are in the intended position set. If not, add the submitted positions that are not in the intended position set to the intended position set.
10. An artificial intelligence-based personnel matching system, implementing the artificial intelligence-based personnel matching method according to any one of claims 1 to 9, characterized in that: include: The first collection module is used to collect the resume content of job seekers; The second acquisition module is used to obtain a set of keywords in the communication records between job seekers and recruiters; The keyword set includes an intention keyword set and a recruitment keyword set; The first processing module is used to process the intention keyword set and the recruitment keyword set respectively to obtain the intention attribute tag set and the recruitment attribute tag set; The third collection module is used to collect basic information of the position, and the basic information of the position includes the requirements and attributes of the recruitment position; The second processing module is used to compare the recruitment attribute tag set with the recruitment position attributes, and replace the recruitment position attributes with the recruitment keyword tags in the recruitment attribute tag set for the inconsistent parts; The personnel matching module inputs the personnel matching data set into the pre-trained personnel matching model to obtain the matching label value. If there is a match, the job application resume is pushed to the recruitment end, and the recruitment position is pushed to the job application end; the personnel matching data set includes basic position information, resume content and intention attribute label set; The fourth collection module is used to collect job seekers' job browsing records and resume submission records; The third processing module is used to process the job browsing records and resume submission records to obtain a job browsing record set and a resume submission record set; The fourth processing module is used to process the job browsing record set and the resume submission record set to obtain the intended job set; the personnel matching data set also includes the intended job set.
Citation Information
Cited By
Intelligent screening method, device and system for intelligent personnel platform
CN121032454A
