Human-post matching algorithm based on V-KSAM + LSTM model

Abstract

The invention relates to the field of talent recruitment and management, in particular to a person and post matching algorithm based on a V-KSAM + LSTM model, and the algorithm comprises the steps: matching a talent portrait based on the V-KSAM + LSTM model with a person and post based on a Sudoku model; the talent portrait based on the V-KSAM + LSTM model comprises the following steps: step S11: data acquisition and preprocessing; step S12, designing a V-KSAM model, and constructing a talent pool; step S13, LSTM model training is carried out, and talent portrait labels are generated; step S14, constructing a knowledge graph; the sudoku model-based person and post matching method comprises the following steps of S21, collecting and preprocessing recruitment information; step S22, performing Sudoku model construction and calculation evaluation; step S23, performing man-post matching; according to the method, staff data is collected through multiple channels, NLP and OCR technologies are used for processing, a V-KSAM model is constructed to analyze the ability and value of the staff data, talents and posts are intelligently and accurately matched based on a Sudoku model, the competency and the job qualification are integrated, and the recruitment efficiency and quality are improved.

Description

Technical Field

[0001] The present invention relates to the field of talent recruitment and management, and particularly to a person-job matching algorithm based on the V-KSAM + LSTM model. Background Art

[0002] In today's digital age, the field of talent recruitment and management is undergoing profound changes. With the rapid development of information technology, the demand for talents by enterprises is becoming increasingly diverse and precise, and the requirements for the intelligence level of talent recruitment platforms are also getting higher and higher. However, most of the current talent recruitment platforms on the market have obvious deficiencies in talent information display and analysis. They can often only provide limited professional and skill information, staying only at the superficial levels such as education background and work experience, and lacking in-depth exploration and analysis of the deep-level capabilities and qualities of talents, especially the comprehensive capabilities, personal traits, career motivations, etc. hidden under the iceberg.

[0003] As a talent group, the staff have accumulated rich professional skills, cultivated comprehensive capabilities and good professional qualities. However, due to the differences in local vocational systems, they face many challenges in the process of entering the local employment market. The traditional recruitment model is difficult to accurately identify and connect their capabilities with enterprise needs. Therefore, a more scientific, efficient and accurate talent recruitment solution is needed to give full play to the talent value of the staff and realize the optimal allocation of talent resources. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above problems and provide a person-job matching algorithm based on the V-KSAM + LSTM model. To achieve the above purpose, the present invention adopts the following technical solutions:

[0005] The person-job matching algorithm based on the V-KSAM + LSTM model includes a talent portrait based on the V-KSAM + LSTM model and a person-job matching based on the nine-square grid model;

[0006] The talent portrait based on the V-KSAM + LSTM model includes the following steps:

[0007] Step S11: Data acquisition and preprocessing;

[0008] Step S12: Design the V-KSAM model and construct a talent pool;

[0009] Step S13: Train the LSTM model to generate talent portrait labels;

[0010] Step S14: Construct a knowledge graph; use the talent portrait labels output by the LSTM model to construct a knowledge graph of the staff talent pool and store the talent information;

[0011] Further, in step S11, data acquisition and preprocessing include the following steps:

[0012] Step S111: Collect raw data from the personal information filled in by the staff through the mobile application, the text information in the personal file, and image materials such as preferential treatment certificates and vocational skill certificates.

[0013] The acquisition of the staff's talent portrait data usually includes the following channels. One is to obtain it by filling in personal information and employment intentions through the mobile APP. The second is to supplement it by extracting the text information in the personal file. The third is to obtain it through certificate materials such as preferential treatment certificates and vocational skill certificates. Different data sources have different data forms and different preprocessing methods.

[0014] Step S112: Use natural language processing (NLP) technology to preprocess the text data from the personal file, and apply optical character recognition (OCR) technology to convert the picture data of awards and certificates into text form, extract keywords and identify professional skill information.

[0015] For the descriptive text data in the personal file, use NLP (Natural Language Processing) technology for field extraction, including Chinese word segmentation, stop word removal, keyword extraction, etc. In this project, the jieba library in Python is used for word segmentation, and a relevant professional term supplement dictionary is added to improve the word segmentation accuracy.

[0016] For picture data such as awards and certificates, use OCR (Optical Character Recognition) character recognition technology to extract the text in the image and convert it into text form for subsequent NLP tasks. By extracting keywords, information such as professional skills is identified.

[0017] In step S12, the V-KSAM model includes the staff's basic information (V), professional knowledge (K), professional skills (S), comprehensive ability (A), and motivation (M).

[0018] Based on the actual situation of the staff group, the present invention proposes a V-KSAM (Veterans-KSAM) model to create a talent portrait of the staff, and constructs a staff talent pool for recruitment and job hunting in the staff talent market. V (Veterans Information) refers to the basic information of the staff, and the description fields include name, gender, age, date of birth, ID number, political status, ethnicity, detailed address, etc. K (Knowledge) refers to the professional information possessed during work, such as professional knowledge in disciplines like aircraft maintenance, electronic communication, navigation, financial management, human resource management, etc. The description fields include graduation school, academic degree, the major actually engaged in, the number of years engaged in the major, etc.; S (Skill) refers to the professional skills mastered during work, such as mechanical operation skills, computer operation skills, electrical engineering skills, financial analysis capabilities, and other professional skills for each position. The description fields include actual work experience, possessed professional skills, operation skills of equipment, office software skills, obtained vocational skill certificates, etc.; A (Ability) refers to the comprehensive ability, including three dimensions: general ability, management ability, and professional quality. Among them, the general ability refers to the traits that are naturally possessed or not easily changed in different environments, such as interpersonal coordination ability, problem analysis ability, judgment and reasoning ability, etc. The management ability refers to the organizational coordination ability during the tenure of an administrative position, such as leadership and influence, communication and coordination ability, planning and execution ability, decision-making ability, stress management and self-adjustment ability, etc. The professional quality refers to the ideological morality, awareness, and behavior habits possessed during the engagement in a specific position or post, such as initiative, responsibility, achievement desire, loyalty, integrity awareness, team awareness, etc. The description fields of the comprehensive ability include moral and talent performance, rewards and punishments, rank, position, etc.; M (Motive) refers to the motivation, and the description fields include whether the job seeker requires a professional counterpart, expected salary level, requirements for the region, requirements for the company culture, expected employment duration, etc.

[0019] Further, in step S13, the training of the LSTM model to generate talent portrait labels includes the following steps:

[0020] Step S131: Perform Text Embedding vectorization processing on the input variables;

[0021] Step S132: Continuously train and adjust the parameters hku, hsv, haw to optimize the outputs F(I), F(K), F(S), F(A), and F(M) as the final talent portrait labels.

[0022] Use PyTorch to build an LSTM (Long Short-Term Memory) model to obtain the label data of the talent portrait from the text fields, and then construct a staff talent knowledge graph and store it in the talent pool.

[0023] In step S132, the talent portrait labels include gender, age, professional knowledge level, skill proficiency, comprehensive ability level, intended occupation, and intended city.

[0024] The core algorithm of the V-KSAM model is to input the 27 fields of information obtained, and use the RNN deep learning model to generate talent portrait labels. The labels used for talent portraits in this project include gender, age, professional knowledge level, skill proficiency, comprehensive ability level, intended occupation, intended city, etc. Among them, the professional knowledge level is evaluated by giving different weights according to academic qualifications, graduation schools, years of professional experience, and meritorious service and awards during the period of professional experience. Skill proficiency is evaluated based on the professional skill certificates obtained, years of professional experience, and awards in professional field competitions. The comprehensive ability level consists of three dimensions: general ability, management ability, and professional quality, and is comprehensively evaluated based on professional knowledge level, skill proficiency, moral and talent performance, rewards and punishments, rank, and position.

[0025] The person-job matching based on the Nine-square Grid model includes the following steps:

[0026] Step S21: Recruitment information collection and preprocessing; the company / enterprise publishes recruitment requirements by logging into the APP, including company information, number of recruits, gender requirements, age limit, job category, educational requirements, marital status requirements, job responsibilities, qualifications, salary situation, job benefits, etc.

[0027] Step S22: Nine-grid model construction and calculation evaluation;

[0028] The nine-square grid model consists of two dimensions: competence and qualification. Talents are differentiated through the nine-square grid to identify the best recommended talents, competent talents, qualified talents and talents to be paid attention to.

[0029] Qualification assessment method: weighted calculation and assessment based on eligibility criteria such as age, gender, professional and academic qualifications, corresponding skills, and work location.

[0030] Furthermore, the qualification calculation function is:

[0031] x=a1A+a2G+a3M+a4E+a5S+a6C

[0032] Among them, A, G, M, E, S and C represent age, gender, major, education, skills and city respectively, and a1, a2, a3, a4, a5 and a6 are weight coefficients. After calculating the qualification coefficient x of all talents in the talent pool, the first third are defined as having high qualifications, the middle third as having medium qualifications, and the last third as having low qualifications.

[0033] The competency assessment method uses four aspects, namely professional skills, general abilities, management abilities and professional qualities, to comprehensively assess the competency of talents for the new positions.

[0034] The calculation function of competency is:

[0035] y=b1Pr+b2Ge+b3Ma+b4Oc

[0036] Among them, Pr, Ge, Ma and Oc represent professional skills, general abilities, management abilities and professional qualities respectively, and b1, b2, b3 and b4 are weight coefficients. After calculating the competency coefficients y of all talents in the talent pool, the first third are defined as having high competency, the middle third as having basic competency, and the last third as having low competency.

[0037] Step S23: Implement person-job matching; select the best recommended talents with high qualifications and competence.

[0038] For companies / enterprises, the Nine-square Grid Model is used to match people with positions, and the best recommended talents with high qualifications and high competence are screened out for the company / enterprise. If there are no best recommended talents, qualified or competent talents can be selected based on the actual talent needs of the company / enterprise.

[0039] For job seekers, the nine-square grid model is used to match the intended positions to job seekers, and the actual position of the job seekers in the nine-square grid is simulated and calculated for the job seekers, so that the job seekers can improve their abilities in a targeted manner or take exams and study to improve their qualifications for the position. The advantages of the present invention are:

[0040] 1. The present invention widely collects staff data through multiple channels, uses NLP and OCR technology to accurately process different types of data, constructs a comprehensive and detailed V-KSAM model, deeply analyzes the various capabilities and qualities of staff, and fully taps their potential value.

[0041] 2. The present invention is based on the design of the nine-square grid model, integrates the two key dimensions of competence and job qualifications, and uses scientific and quantitative calculation functions and evaluation standards to achieve intelligent and accurate matching of talents and positions. It overcomes the one-sidedness of traditional recruitment that only matches professional skills, greatly improves recruitment efficiency and quality, reduces corporate talent screening costs, and provides staff with employment opportunities that are more in line with their own abilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings constituting a part of this application are used to provide a further understanding of this application, so that other features, purposes and advantages of this application become more obvious. The illustrative embodiment drawings and their descriptions of this application are used to explain this application and do not constitute an improper limitation on this application.

[0043] In the accompanying drawings:

[0044] Figure 1 It is a flowchart of the person-job matching algorithm based on the V-KSAM+LSTM model in Embodiment 1.

[0045] Figure 2 It is a design diagram of the V-KSAM model in Embodiment 1.

[0046] Figure 3 It is a design diagram of the LSTM model algorithm in Embodiment 1.

[0047] Figure 4 It is a design diagram of the nine-square grid model in Embodiment 2.

[0048] Figure 5 It is an example diagram of job qualification assessment in Embodiment 2.

[0049] Figure 6 It is an example diagram of competency assessment in Embodiment 2. Detailed implementation manners

[0050] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations.

[0051] The present invention will be described in detail and specifically below through specific embodiments to better understand the present invention. However, the following embodiments do not limit the protection scope of the present invention.

[0052] Embodiment 1

[0053] The person-job matching algorithm based on the V-KSAM+LSTM model includes talent profiling based on the V-KSAM+LSTM model and person-job matching based on the nine-square grid model;

[0054] The talent profiling based on the V-KSAM+LSTM model includes the following steps:

[0055] Step S11: Data acquisition and preprocessing;

[0056] Step S12: Design the V-KSAM model and build a talent pool;

[0057] Step S13: Train the LSTM model to generate talent profiling labels;

[0058] Step S14: Knowledge graph construction; Use the talent portrait tags output by the LSTM model to construct the knowledge graph of the staff talent pool and store the talent information.

[0059] Further, in step S11, data acquisition and preprocessing include the following steps:

[0060] Step S111: Collect raw data from the personal information filled in by the staff through the mobile application, the text information in the personal file, and image materials such as preferential treatment certificates and professional skill certificates.

[0061] The acquisition of staff talent portrait data usually includes the following channels. One is to obtain it by filling in personal information and employment intentions through the mobile APP. The second is to supplement it by extracting the text information in the personal file. The third is to obtain it through certificate materials such as preferential treatment certificates and professional skill certificates. For different data sources, their data forms are different, and the preprocessing methods are also different.

[0062] Step S112: Use natural language processing (NLP) technology to preprocess the text data from the personal file, apply optical character recognition (OCR) technology to convert the picture data of awards and certificates into text form, extract keywords and identify professional skill information.

[0063] For the descriptive text data in the personal file, use NLP (Natural Language Processing) technology for field extraction, including Chinese word segmentation, stop word removal, keyword extraction, etc. In this project, the jieba library in Python is used for word segmentation, and a relevant professional term supplementary dictionary is added to improve the word segmentation accuracy.

[0064] For picture data such as awards and certificates, use OCR (Optical Character Recognition) text recognition technology to extract the text in the image and convert it into text form for subsequent NLP tasks. By extracting keywords, information such as professional skills is identified.

[0065] As Figure 1 shown, in step S12, the V-KSAM model includes the staff's basic information (V), professional knowledge (K), professional skills (S), comprehensive ability (A), and motivation (M).

[0066] Based on the KSAM model, according to the actual situation of the special group of staff, the V-KSAM (Veterans-KSAM) model is proposed to create a talent portrait of the staff and construct a staff talent pool for recruitment and job hunting in the staff talent market. V (Veterans Information) refers to the basic information of the staff, and the description fields include name, gender, age, date of birth, ID number, political status, ethnicity, detailed address, etc. K (Knowledge) refers to the professional information possessed during work, such as professional knowledge in disciplines like aircraft maintenance, electronic communication, navigation, financial management, human resource management, etc. The description fields include graduation school, educational background, the profession actually engaged in, the number of years in the engaged profession, etc.; S (Skill) refers to the professional skills mastered during work, such as mechanical operation skills, computer operation skills, electrical skills, financial analysis capabilities, and other professional skills for each position. The description fields include actual work experience, possessed professional skills, operation skills of equipment, office software skills, obtained vocational skill certificates, etc.; A (Ability) refers to the comprehensive ability, including three dimensions: general ability, management ability, and professional quality. Among them, general ability refers to the traits that are naturally possessed or not easily changed in different environments, such as interpersonal coordination ability, problem analysis ability, judgment and reasoning ability, etc. Management ability refers to the organizational coordination ability during the tenure in the administrative position, such as leadership and influence, communication and coordination ability, planning and execution ability, decision-making ability, stress management and self-adjustment ability, etc. Professional quality refers to the ideological morality, awareness, and behavior habits possessed during the engagement in a specific position or job, such as initiative, responsibility, achievement desire, loyalty, integrity awareness, team awareness, etc. The description fields of the comprehensive ability include moral and talent performance, rewards and punishments, rank, position, etc.; M (Motive) refers to motivation, and the description fields include whether the job seeker requires a professional counterpart, expected salary level, requirements for the region, requirements for the company culture, expected employment duration, etc.

[0067] In step S13, the LSTM model is trained to generate talent portrait tags, including the following steps:

[0068] Step S131: Perform Text Embedding vectorization processing on the input variables;

[0069] Step S132: Continuously train and adjust the parameters hku, hsv, haw to optimize the outputs F(I), F(K), F(S), F(A), and F(M) as the final talent portrait tags.

[0070] Use PyTorch to build an LSTM (Long Short-Term Memory) model to obtain the tag data of the talent portrait from the text fields, and then construct a staff talent knowledge graph and store it in the talent pool.

[0071] In step S132, the talent portrait labels include gender, age, professional knowledge level, skill proficiency, comprehensive ability level, intended occupation, and intended city.

[0072] The core algorithm of the V-KSAM model is to input the 27 fields of information obtained, and use the RNN deep learning model to generate talent portrait labels. The labels used for talent portraits in this project include gender, age, professional knowledge level, skill proficiency, comprehensive ability level, intended occupation, intended city, etc. Among them, the professional knowledge level is evaluated by giving different weights according to academic qualifications, graduation schools, years of professional experience, and meritorious service and awards during the period of professional experience. Skill proficiency is evaluated based on the professional skill certificates obtained, years of professional experience, and awards in professional field competitions. The comprehensive ability level consists of three dimensions: general ability, management ability, and professional quality, and is comprehensively evaluated based on professional knowledge level, skill proficiency, moral and talent performance, rewards and punishments, rank, and position.

[0073] like Figure 2 As shown in the figure, I, Xk, Xs, A and M are algorithm input variables, which represent personal information, professional knowledge, professional skills, comprehensive ability and motivation in the V-KSAM model, respectively. Among them, the comprehensive ability variable A is divided into general ability variable Xa1, management ability variable Xa2 and professional quality variable Xa3. After the above variables are vectorized by Text Embedding, the parameters such as hku, hsv, haw are adjusted through continuous training to output F(I), F(K), F(S), F(A) and F(M) for talent portrait labels. F(I) includes gender and age, F(K) describes the level of professional knowledge, F(S) describes the proficiency of skills, F(A) describes the comprehensive ability level, and F(M) includes the intended occupation and intended city.

[0074] Example 2

[0075] The person-job matching based on the Nine-square Grid model includes the following steps:

[0076] Step S21: Recruitment information collection and preprocessing; the company / enterprise publishes recruitment requirements by logging into the APP, including company information, number of recruits, gender requirements, age limit, job category, educational requirements, marital status requirements, job responsibilities, qualifications, salary situation, job benefits, etc.

[0077] Step S22: Nine-grid model construction and calculation evaluation;

[0078] like Figure 3As shown in the figure, the nine-square grid model consists of two dimensions: competency and qualification. Talents are differentiated through the nine-square grid to identify the best-recommended talents, competent talents, qualified talents, and talents worthy of attention.

[0079] Qualification assessment method: Weighted calculation and assessment are carried out based on qualification conditions such as whether the age and gender match, whether the major and education level meet the requirements, whether the corresponding skills are possessed, and whether the work location matches.

[0080] As Figure 4 shown, for age, gender, and city, a scoring standard of compliance / non-compliance is adopted. Compliance is counted as 1, and non-compliance is counted as 0; for the requirements of major, education level, and skills, those that fully meet the requirements are counted as A (2 points), those that meet part of the requirements or are relevant are counted as B (1 point), and those that do not meet the requirements at all are counted as C (0 points).

[0081] Furthermore, the qualification calculation function is:

[0082] x = a1A + a2G + a3M + a4E + a5S + a6C

[0083] Among them, A, G, M, E, S, and C respectively represent age, gender, major, education level, skills, and city, and a1, a2, a3, a4, a5, a6 are weight coefficients. After calculating the qualification coefficients x of all talents in the talent pool, those in the top one-third are defined as having relatively high qualifications, the middle one-third are defined as having medium qualifications, and the last one-third are defined as having low qualifications.

[0084] The competency assessment method comprehensively assesses a talent's competency for the position based on four aspects: professional skills, general ability, management ability, and professional quality.

[0085] As Figure 5 shown, for the requirements of professional skills, general ability, management ability, and professional quality, those that fully meet the requirements are counted as A (2 points), those that meet part of the requirements or are relevant are counted as B (1 point), and those that do not meet the requirements at all are counted as C (0 points).

[0086] The calculation function of competency is:

[0087] y = b1Pr + b2Ge + b3Ma + b4Oc

[0088] Among them, Pr, Ge, Ma, and Oc respectively represent professional skills, general ability, management ability, and professional quality, and b1, b2, b3, b4 are weight coefficients. After calculating the competency coefficients y of all talents in the talent pool, those in the top one-third are defined as having relatively high competency, the middle one-third are defined as basically having competency, and the last one-third are defined as having low competency.

[0089] Step S23: Implement person-job matching; select the best recommended talents with high qualifications and competence.

[0090] For companies / enterprises, the Nine-square Grid Model is used to match people with positions, and the best recommended talents with high qualifications and high competence are screened out for the company / enterprise. If there are no best recommended talents, qualified or competent talents can be selected based on the actual talent needs of the company / enterprise.

[0091] For job seekers, the nine-square grid model is used to match people with intended positions, and the job seekers' actual positions in the nine-square grid are simulated and calculated, so that they can improve their various abilities in a targeted manner or take exams and study to improve their qualifications for the position.

[0092] The specific embodiments of the present invention are described in detail above, but they are only examples, and the present invention is not equivalent to the specific embodiments described above. For those skilled in the art, any equivalent modifications and substitutions made to the present invention are also within the scope of the present invention. Therefore, the equalization changes and modifications made without departing from the spirit and scope of the present invention should be included in the scope of the present invention.

Claims

1. A person-job matching algorithm based on the V-KSAM + LSTM model, characterized in that Including talent profiling based on the V-KSAM+LSTM model and job matching based on the Nine-squares model; Talent profiling based on the V-KSAM+LSTM model includes the following steps: Step S11: data acquisition and preprocessing; Step S12: designing a V-KSAM model and building a talent pool; Step S13: LSTM model training to generate talent portrait labels; Step S14: knowledge graph construction: using the talent portrait labels output by the LSTM model, construct a knowledge graph of the staff talent pool and store the talent information; The person-job matching based on the Nine-square Grid model includes the following steps: Step S21: recruitment information collection and preprocessing; Step S22: Nine-grid model construction and calculation evaluation; Step S23: Implement person-job matching; select the best recommended talents with high qualifications and competence.

2. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 1, characterized in that: In step S11, the data acquisition and preprocessing includes the following steps: Step S111: Collect original data from personal information filled in by staff through mobile applications, text information in personal files, and image materials such as preferential cards and professional skill certificates. Step S112: Use natural language processing (NLP) technology to pre-process the text data from the personal profile, apply optical character recognition (OCR) technology to convert the award and certificate image data into text form, extract keywords and identify professional skill information.

3. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 1, wherein: In step S12, the V-KSAM model includes basic information (V), professional knowledge (K), professional skills (S), comprehensive ability (A) and motivation (M) of the staff.

4. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 1, wherein: In step S13, LSTM model training and talent portrait label generation include the following steps: Step S131: Perform Text Embedding vectorization processing on the input variable; Step S132: Parameters hku, hsv, and haw are adjusted through continuous training to optimize the output of F(I), F(K), F(S), F(A), and F(M) as the final talent portrait labels.

5. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 4, wherein: In step S132, the talent portrait labels include gender, age, professional knowledge level, skill proficiency, comprehensive ability level, intended occupation, and intended city.

6. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 5, characterized in that: The nine-square grid model consists of two dimensions: competence and qualification. Talents are differentiated through the nine-square grid to identify the best recommended talents, competent talents, qualified talents and talents to be paid attention to.

7. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 6, wherein: In step S22, the qualification calculation function is: x=a1A+a2G+a3M+a4E+a5S+a6C Among them, A, G, M, E, S and C represent age, gender, major, education, skills and city respectively, and a1, a2, a3, a4, a5 and a6 are weight coefficients.

8. The person-job matching algorithm based on the V-KSAM+LSTM model according to claim 6, wherein: In step S22, the calculation function of the competency is: y=b1Pr+b2Ge+b3Ma+b4Oc Among them, Pr, Ge, Ma and Oc represent professional skills, general abilities, management abilities and professional qualities respectively, and b1, b2, b3 and b4 are weight coefficients.

Images