Recruitment resource intelligent distribution system based on logistics talent evaluation

Abstract

The invention discloses a recruitment resource intelligent distribution system based on logistics talent evaluation, and the system comprises a comprehensive evaluation score obtaining unit which is used for obtaining multi-dimensional evaluation items; the standard candidate screening unit is used for performing standard screening on the plurality of basic candidates by using the comprehensive evaluation score to obtain standard candidates; the fusion vector unit is used for executing weight matching feature fusion on the multi-dimensional evaluation items of the standard candidates to generate fusion vectors of the standard candidates; the resource allocation unit is used for inputting the fusion vector of the standard candidate into the fine-tuned recruitment resource allocation model and outputting a recruitment resource allocation strategy ID; the resource package matching unit is used for matching a recruitment resource package in a training space of a recruitment resource allocation model according to the recruitment resource allocation strategy ID; by comprehensively considering the multi-dimensional evaluation items of the candidates, optimization can be performed in real time according to the actual performance of the candidates, the maximum utilization of recruitment resources is ensured, and the talent matching precision is improved.

Description

Technical Field

[0001] The present invention relates to a recruitment resource allocation system, in particular to an intelligent recruitment resource allocation system based on logistics talent evaluation. Background Art

[0002] As the employment needs of logistics companies become increasingly diversified, traditional recruitment processes can no longer effectively meet the requirements of accurate recruitment and optimized resource allocation. Especially in the logistics industry, multi-dimensional evaluation information such as candidates' professional capabilities, personality traits, interview performance, etc. is often difficult to be effectively integrated, resulting in low efficiency in the allocation of recruitment resources and failure to accurately match the most suitable candidates.

[0003] Traditional recruitment methods rely on resume screening and interview assessment, but these methods often fail to fully consider the comprehensive performance of candidates in multiple dimensions and their adaptability to the needs of logistics companies. For example, the patent document CN202310689731.0 discloses a human resource allocation system based on personnel demand analysis. The adaptation effect determination module records the performance assessment of each job seeker in the assessment plant, providing a reference for the subsequent matching of the adaptation plant. Finally, the ranking module combines the performance assessment of each job seeker in the assessment plant to rank each job seeker, thereby improving the two-way selection and adaptation effect of personnel and plant.

[0004] However, the allocation of recruitment resources in existing technologies often relies on experience and judgment, which is highly subjective and uncertain. In order to improve the accuracy and efficiency of the allocation of recruitment resources, many logistics companies have begun to try to adopt data-driven recruitment decision support systems. These systems usually predict the most suitable candidates by analyzing candidate historical data, comprehensive evaluation results, and post-employment performance. However, existing recruitment resource allocation systems mostly use static or simple rules to allocate resources, lacking intelligent adjustment mechanisms based on data and models, resulting in low resource allocation efficiency. In addition, the allocation of recruitment resources mostly relies on initial evaluation scores as the main basis for recruitment decisions, and fails to fully consider post-employment performance and changes, resulting in the inability to accurately match the most suitable candidates and recruitment resources. Summary of the invention

[0005] In view of the deficiencies of the prior art, the present invention provides an intelligent recruitment resource allocation system based on logistics talent evaluation, which solves the technical problems raised in the background technology by fine-tuning through a posteriori scoring.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] An intelligent recruitment resource allocation system based on logistics talent evaluation, including:

[0008] A comprehensive evaluation score acquisition unit, used to acquire multi-dimensional evaluation items of a number of basic candidates and generate a comprehensive evaluation score based on the multi-dimensional evaluation items;

[0009] A standard candidate screening unit, used for performing standard screening on a plurality of basic candidates using a comprehensive evaluation score to obtain a standard candidate;

[0010] A fusion vector unit, used to perform weighted feature fusion on the multi-dimensional evaluation items of the standard candidate to generate a fusion vector of the standard candidate;

[0011] A resource allocation unit, used for inputting the fusion vector of the standard candidates into the fine-tuned recruitment resource allocation model, and outputting the recruitment resource allocation strategy ID;

[0012] The resource package matching unit is used to match the recruitment resource package in the training space of the recruitment resource allocation model according to the recruitment resource allocation strategy ID.

[0013] In some embodiments, obtaining multi-dimensional evaluation items of a plurality of basic candidates and generating a comprehensive evaluation score based on the multi-dimensional evaluation items include:

[0014] S1-1. Screen a number of basic candidates who meet the basic requirements from the submitted electronic resumes;

[0015] S1-2, label the basic candidates with independent candidate IDs;

[0016] S1-3, perform online logistics skill assessment on the basic candidates with the candidate ID marked, and generate several multi-dimensional assessment items for each basic candidate;

[0017] S1-4, encode several multi-dimensional evaluation items to generate several single-item evaluation scores;

[0018] S1-5. Perform weighted summation on several individual assessment scores to generate a comprehensive assessment score for each basic candidate.

[0019] In some of the embodiments, a comprehensive evaluation score is used to perform standard screening on a number of basic candidates to obtain standard candidates, including:

[0020] S2-1, presetting the minimum threshold, middle threshold and maximum threshold of the comprehensive evaluation score;

[0021] S2-2, the comprehensive evaluation score below the minimum threshold is defined as the first scoring range; the comprehensive evaluation score between the minimum threshold and the middle threshold is defined as the second scoring range; the comprehensive evaluation score between the middle threshold and the maximum threshold is defined as the third scoring range; the comprehensive evaluation score exceeding the maximum threshold is defined as the fourth scoring range;

[0022] S2-3. Define the basic candidates falling within the second score range, the third score range and the fourth score range as standard candidates, and assign independent candidate grades; the candidate grades include: primary candidate grade, senior candidate grade and excellent candidate grade.

[0023] In some of the embodiments, an online logistics skill assessment is performed on the basic candidates with the candidate IDs marked, and a number of multi-dimensional assessment items are generated for each basic candidate, including:

[0024] S1-3-1. Execute online assessment of the logistics talents, and classify the assessment results of the online assessment into professional assessment items, personality assessment items, and interview assessment items;

[0025] Among them, the professional assessment items include: logistics skills score and logistics experience score; the personality assessment items include: personality type label and behavioral tendency label; the interview assessment items include: oral expression quality score and cultural adaptation score;

[0026] S1-3-2 defines the combination of the logistics skill score, logistics experience score, personality type label, behavior tendency label, oral expression quality score and cultural adaptation score as the multi-dimensional evaluation item.

[0027] In some embodiments, encoding is performed on a plurality of multi-dimensional evaluation items to generate a plurality of single-item evaluation scores. The encoding is performed on a plurality of multi-dimensional evaluation items to generate a plurality of single-item evaluation scores, including:

[0028] S1-4-1. Standardize the logistics skill score, logistics experience score, oral expression quality score and cultural adaptation score to obtain the professional score, experience score, expression score and adaptation score.

[0029] S1-4-2, assigning target coding scores to personality type labels and behavioral tendency labels to obtain personality item scores and behavioral item scores;

[0030] S1-4-3. Define the professional item score, experience item score, expression item score, adaptation item score, personality item score and behavior item score as the single item assessment score.

[0031] In some embodiments, performing weighted feature fusion on the multi-dimensional evaluation items of the standard candidate to generate a fusion vector of the standard candidate includes:

[0032] S3-1, perform triple pre-definition on professional evaluation items, personality evaluation items and interview evaluation items to obtain professional item triples, personality item triples and interview item triples;

[0033] S3-2, pre-define the contribution pairs for the professional item triples, the personality item triples and the interview item triples to obtain corresponding professional contribution pairs, personality contribution pairs and interview contribution pairs; wherein the professional contribution pair includes the first professional contribution weight and the second professional contribution weight; the personality contribution pair includes the first personality contribution weight and the second personality contribution weight; the interview contribution pair includes the first interview contribution weight and the second interview contribution weight;

[0034] S3-3 calculates the professional contribution vector of the professional item triple according to the first professional contribution weight and the second professional contribution weight;

[0035] The expression for calculating the professional contribution vector of the professional term triple is:

[0036] P contribution =w P1 ·P1+w P2 P2;

[0037] P contribution represents the professional contribution vector, P1 and P2 represent the professional score and experience score respectively, and w P1 and w P2 They represent the contribution weight of the first profession and the contribution weight of the second profession respectively;

[0038] Calculate the character contribution vector of the character item triple according to the first character contribution weight and the second character contribution weight;

[0039] The expression for calculating the character contribution vector of the character item triple is:

[0040] C contribution =w C1 C1+w C2 C2;

[0041] C contribution represents the personality contribution vector, C1 and C2 represent the personality item score and behavior item score respectively, and w C1 and w C2 Represent the first character contribution weight and the second character contribution weight respectively;

[0042] Calculate the interview contribution vector of the interview item triple according to the first interview contribution weight and the second interview contribution weight;

[0043] The expression for calculating the interview contribution vector of the interview item triple is:

[0044] I contribution =w I1 I1+w I2 I2;

[0045] I contributionrepresents the interview contribution vector, I1 and I2 represent the expression score and adaptation score respectively, and w I1 and w I2 They represent the contribution weight of the first interview and the contribution weight of the second interview respectively;

[0046] S3-4. Perform feature fusion on the professional contribution vector, the personality contribution vector and the interview contribution vector to obtain a fusion vector of the standard candidate.

[0047] In some of the embodiments, matching the recruitment resource package in the training space of the recruitment resource allocation model according to the recruitment resource allocation strategy ID includes:

[0048] S5-1. Reversely decompose the recruitment resource allocation strategy ID to obtain several recruitment resource classification numbers;

[0049] S5-2. According to the mapping of several recruitment resource classification numbers, the classified recruitment resources are matched in the pre-constructed training space to form a recruitment resource package for each standard candidate.

[0050] In some embodiments, the step of fine-tuning the recruitment resource allocation model includes:

[0051] A1. Obtain the prior comprehensive assessment scores and the posterior comprehensive assessment scores of historical employees;

[0052] A2. Calculate the first mean square error of the priori comprehensive measurement score and the posterior comprehensive measurement score;

[0053] The expression of the first mean square error is:

[0054] MSE represents the first mean square error between the prior comprehensive test score and the posterior comprehensive test score, N represents the total number of historical employees, S prior,i represents the prior comprehensive assessment score of the i-th historical employee, S posterior,i represents the posterior comprehensive assessment score of the i-th historical employee;

[0055] A3. define the first mean square error exceeding the error threshold as the ideal error, and associate several ideal historical employees with the ideal error based on the historical ID;

[0056] A4. Obtain recruitment resource training samples of ideal historical entrants to construct a recruitment resource sample set;

[0057] A5. Iteratively train the neural network model using the recruitment resource training sample set and output the predicted value of the historical recruitment resource strategy ID.

[0058] A6. When the iterative training reaches model convergence, the neural network model is exported as the recruitment resource allocation model.

[0059] In some embodiments, obtaining a priori comprehensive assessment scores and a posteriori comprehensive assessment scores of historical job applicants includes:

[0060] A1-1. Obtain multi-dimensional historical evaluation items of several historical candidates and mark them with independent historical IDs;

[0061] A1-2. Select historical candidates from a number of historical candidates;

[0062] A1-3. Generate a historical comprehensive assessment score based on the multi-dimensional historical assessment items of the historical employees;

[0063] A1-4, defining the historical comprehensive assessment score of the historical employee as a priori comprehensive assessment score;

[0064] A1-5. Determine the a posteriori comprehensive assessment score of historical entrants based on their post-employment work performance.

[0065] In some of the embodiments, obtaining a recruitment resource training sample of an ideal historical employee to construct a recruitment resource sample set includes:

[0066] A4-1. Obtain some historical recruitment resources of ideal historical candidates during recruitment;

[0067] A4-2. Assign a preset resource classification ID to each historical recruitment resource, and generate several resource classification IDs of ideal historical recruits during recruitment;

[0068] A4-3, concatenating the resource classification IDs to generate a historical recruitment resource strategy ID for an ideal historical employee;

[0069] A4-4. Obtain multi-dimensional historical evaluation items of the ideal historical employee based on the historical ID of the ideal historical employee;

[0070] A4-5. Perform the weighted feature fusion on the multi-dimensional historical evaluation items of the ideal historical employee to generate a fusion vector of the ideal historical employee;

[0071] A4-5. Use the fusion vector of the ideal historical entrants as input features and the historical recruitment resource strategy ID as the target label to construct a recruitment resource sample set, and define it as the training space.

[0072] The present invention provides a recruitment resource intelligent allocation system based on logistics talent evaluation, which has the following beneficial effects:

[0073] The present invention comprehensively considers the multi-dimensional evaluation items of candidates (such as logistics skills, personality traits and interview performance, etc.), and uses the posterior scores in historical data to fine-tune the model, so that the recruitment resource classification strategy ID output by the model is based on the photo resources allocated to the real ideal historical employees in history during recruitment, so that the model can give a dynamically adjusted recruitment resource allocation strategy. This recruitment strategy ID based on the candidate's historical and future performance prediction makes the recruitment resource allocation more accurate. Compared with the traditional static screening and allocation method, it can be optimized in real time according to the candidate's actual performance, ensuring the maximum utilization of recruitment resources and improving the accuracy of talent matching.

[0074] On the other hand, by combining the comprehensive assessment scores of historical entrants with their a posteriori performance, we can more accurately predict the candidates’ performance in actual work. By optimizing the recruitment resource allocation strategy with this a posteriori score, we can provide candidates with more suitable job matches and reduce the turnover rate caused by the mismatch between the position and the candidate’s abilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 This is a structural block diagram of a recruitment resource intelligent allocation system based on logistics talent evaluation in the present invention;

[0076] Figure 2 This is a schematic diagram of the execution steps of a recruitment resource intelligent allocation system based on logistics talent evaluation according to the present invention;

[0077] Figure 3 It is a schematic diagram of the fine-tuning steps of the recruitment resource allocation model in the recruitment resource intelligent allocation system based on logistics talent evaluation of the present invention;

[0078] Figure 4 It is a schematic diagram of the steps for obtaining the a priori comprehensive evaluation score and the a posteriori comprehensive evaluation score described in the present invention. DETAILED DESCRIPTION

[0079] 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.

[0080] Example 1: Please refer to Figure 1 to Figure 2 The present invention provides a recruitment resource intelligent allocation system based on logistics talent evaluation, comprising:

[0081] A comprehensive evaluation score acquisition unit, used to acquire multi-dimensional evaluation items of a number of basic candidates and generate a comprehensive evaluation score based on the multi-dimensional evaluation items;

[0082] A standard candidate screening unit, used for performing standard screening on a plurality of basic candidates using a comprehensive evaluation score to obtain a standard candidate;

[0083] A fusion vector unit, used to perform weighted feature fusion on the multi-dimensional evaluation items of the standard candidate to generate a fusion vector of the standard candidate;

[0084] A resource allocation unit, used for inputting the fusion vector of the standard candidates into the fine-tuned recruitment resource allocation model, and outputting the recruitment resource allocation strategy ID;

[0085] The resource package matching unit is used to match the recruitment resource package in the training space of the recruitment resource allocation model according to the recruitment resource allocation strategy ID.

[0086] Specifically, this method first obtains the multi-dimensional evaluation items of the candidates, including professional ability, personality traits and interview performance, and then calculates the comprehensive evaluation score of each candidate. Through the comprehensive evaluation score, the candidates can be screened by standards to determine whether they meet the basic requirements. Next, the evaluation items of the standard candidates that have passed the screening are weighted and feature fused to generate a fusion vector for each candidate. Finally, these fusion vectors are input into the fine-tuned recruitment resource allocation model, and the model outputs the corresponding recruitment resource allocation strategy, so as to match the appropriate recruitment resource package in the training space. Among them, the training space is the training space obtained by combining a large number of historical recruitment resource strategy IDs when the recruitment resource allocation model is training historical data.

[0087] Furthermore, in this embodiment, a recruitment resource package refers to a set of resources matched from the training space according to the recruitment resource allocation strategy ID. These resource packages usually include various resources required in the recruitment process, such as job requirements, job descriptions, recruitment channels, interview processes, compensation and other specific contents. Specifically, the recruitment resource package is matched through the following steps:

[0088] First, the recruitment resource allocation strategy ID is generated by the fine-tuned recruitment resource allocation model, which represents an optimal resource allocation plan. This strategy ID contains the optimal recruitment resource allocation strategy calculated based on the multi-dimensional evaluation data of the candidate (such as professional ability, personality characteristics, interview performance, etc.).

[0089] The training space is a space obtained by combining a large number of historical recruitment resource strategy IDs during the historical data training process of the recruitment resource allocation model. Its essence is the representation of the recruitment resource sample set in the non-training stage. Each strategy ID has a corresponding representation in the historical data. The training space stores these historical recruitment resource strategies and their corresponding effects for subsequent matching and prediction.

[0090] After obtaining the historical recruitment resource strategy ID through the fine-tuned recruitment resource allocation model, the model will search for the recruitment resource package that best matches the strategy ID in the historical training space. This usually includes the specific skill requirements of the job, the selection of recruitment channels, the formulation of recruitment advertisements, the allocation of interviewers, etc.

[0091] The acquisition step of the comprehensive evaluation score acquisition unit includes:

[0092] S1-1. Screen a number of basic candidates who meet the basic requirements from the submitted electronic resumes;

[0093] S1-2, label the basic candidates with independent candidate IDs;

[0094] S1-3, perform online logistics skill assessment on the basic candidates with the candidate ID marked, and generate several multi-dimensional assessment items for each basic candidate;

[0095] S1-4, encode several multi-dimensional evaluation items to generate several single-item evaluation scores;

[0096] S1-5. Perform weighted summation on several individual assessment scores to generate a comprehensive assessment score for each basic candidate.

[0097] Specifically, candidates who meet the basic requirements are screened from the submitted electronic resumes, and each candidate is marked with an independent candidate ID. Next, an online assessment of logistics skills is performed for each candidate to generate multiple dimensions of assessment items, which cover professional skills, experience, and other relevant abilities. Then, these assessment items are encoded and converted into quantifiable individual assessment scores. Finally, the comprehensive assessment score of each basic candidate is obtained by weighted summation of all individual assessment scores.

[0098] The screening step of the standard candidate screening unit includes:

[0099] S2-1, presetting the minimum threshold, middle threshold and maximum threshold of the comprehensive evaluation score;

[0100] S2-2, the comprehensive evaluation score below the minimum threshold is defined as the first scoring range; the comprehensive evaluation score between the minimum threshold and the middle threshold is defined as the second scoring range; the comprehensive evaluation score between the middle threshold and the maximum threshold is defined as the third scoring range; the comprehensive evaluation score exceeding the maximum threshold is defined as the fourth scoring range;

[0101] S2-3. Define the basic candidates falling within the second score range, the third score range and the fourth score range as standard candidates, and assign independent candidate grades; the candidate grades include: primary candidate grade, senior candidate grade and excellent candidate grade.

[0102] Specifically, in order to perform standard screening, four scoring ranges of the comprehensive assessment score are preset: the minimum threshold, the middle threshold, and the maximum threshold. According to these thresholds, the comprehensive assessment score is divided: candidates below the minimum threshold enter the first scoring range; candidates with comprehensive assessment scores between the minimum threshold and the middle threshold enter the second scoring range; candidates with comprehensive assessment scores between the middle threshold and the maximum threshold enter the third scoring range; candidates exceeding the maximum threshold enter the fourth scoring range. Finally, according to the scoring range, candidates with comprehensive assessment scores in the second, third, or fourth scoring ranges are defined as standard candidates, and corresponding candidate grades are assigned according to their comprehensive assessment scores, including primary candidate grades, senior candidate grades, and excellent candidate grades.

[0103] Furthermore, the step of generating a plurality of multi-dimensional evaluation items includes:

[0104] S1-3-1. Execute online assessment of the logistics talents, and classify the assessment results of the online assessment into professional assessment items, personality assessment items, and interview assessment items;

[0105] Among them, the professional assessment items include: logistics skills score and logistics experience score; the personality assessment items include: personality type label and behavioral tendency label; the interview assessment items include: oral expression quality score and cultural adaptation score;

[0106] For example, the logistics skill score can be obtained through an online skill assessment platform, a special skill test by the recruiting company, a certificate of completion of an online training course, etc. Candidates provide relevant certificates when submitting their electronic resumes or pass online skill tests, and their logistics skills are scored through online skill tests.

[0107] The logistics experience score is obtained from the candidate's past work experience, personal resume, and interview records registered in the online resume. The employer queries the candidate's work experience through a structured questionnaire or professional resume platform, evaluates his or her actual operational experience in the logistics field, and gives a score.

[0108] The personality type label can be obtained through personality testing tools, psychological questionnaires (such as MBTI test, DISC personality assessment, five major personality tests, etc.). Candidates participate in online standardized personality tests to generate personality type labels.

[0109] Behavioral tendency labels are obtained by evaluating the candidate's behavioral tendencies through online interview questions and then deriving labels.

[0110] Oral expression quality scores can be obtained by recording interview videos and using speech analysis software (such as Speechmatics, Voxpopme, etc.) to automatically score the candidate's oral expression quality scores, taking into account factors such as pronunciation clarity, fluency, speaking speed and communication effectiveness.

[0111] The cultural fit score is obtained by evaluating whether the candidate matches the company culture through questions about values, work attitudes, etc. in the structured interview. At the same time, the score can be combined with the candidate's ability to adapt to a multicultural environment, such as whether he has experience in cross-cultural communication.

[0112] S1-3-2 defines the combination of the logistics skill score, logistics experience score, personality type label, behavior tendency label, oral expression quality score and cultural adaptation score as the multi-dimensional evaluation item.

[0113] Furthermore, the step of generating a plurality of single-item assessment scores includes:

[0114] S1-4-1. Standardize the logistics skill score, logistics experience score, oral expression quality score and cultural adaptation score to obtain the professional score, experience score, expression score and adaptation score.

[0115] S1-4-2, assigning target coding scores to personality type labels and behavioral tendency labels to obtain personality item scores and behavioral item scores;

[0116] Exemplarily, the categories are encoded according to the target variable (such as the candidate's comprehensive assessment score, recruitment success rate, etc.). The category labels are replaced according to the average target variable value of each category; for example, if the average comprehensive assessment score of the personality type candidates "extrovert" is 85, "introvert" is 80, and "neutral" is 90, then the encoding of these categories can be set to 85, 80, and 90.

[0117] S1-4-3. Define the professional item score, experience item score, expression item score, adaptation item score, personality item score and behavior item score as the single item assessment score.

[0118] Specifically, in this step, the multi-dimensional evaluation items are coded and standardized to ensure the comparability of the scoring results. First, the logistics skills score, logistics experience score, oral expression quality score, and cultural adaptation score are standardized or normalized. The purpose is to unify the scores of different scales to the same scale so that the scores between different dimensions can be directly compared. Subsequently, the personality type labels and behavioral tendency labels are target-coded and converted into numerical scores to reflect the candidate's performance on these psychological characteristics. Through this process, each evaluation score is unified into a standardized numerical form, which helps to accurately evaluate the comprehensive quality of the candidate.

[0119] The step of performing weighted feature fusion includes:

[0120] S3-1, perform triple pre-definition on professional evaluation items, personality evaluation items and interview evaluation items to obtain professional item triples, personality item triples and interview item triples;

[0121] S3-2, pre-define the contribution pairs for the professional item triples, the personality item triples and the interview item triples to obtain corresponding professional contribution pairs, personality contribution pairs and interview contribution pairs; wherein the professional contribution pair includes the first professional contribution weight and the second professional contribution weight; the personality contribution pair includes the first personality contribution weight and the second personality contribution weight; the interview contribution pair includes the first interview contribution weight and the second interview contribution weight;

[0122] S3-3 calculates the professional contribution vector of the professional item triple according to the first professional contribution weight and the second professional contribution weight;

[0123] The expression for calculating the professional contribution vector of the professional term triple is:

[0124] P contribution =w P1 ·P1+w P2 P2;

[0125] P contribution represents the professional contribution vector, P1 and P2 represent the professional score and experience score respectively, and w P1 and w P2 They represent the contribution weight of the first profession and the contribution weight of the second profession respectively;

[0126] Calculate the character contribution vector of the character item triple according to the first character contribution weight and the second character contribution weight;

[0127] The expression for calculating the character contribution vector of the character item triple is:

[0128] C contribution =w C1 C1+wC2 C2;

[0129] C contribution represents the personality contribution vector, C1 and C2 represent the personality item score and behavior item score respectively, and w C1 and w C2 Represent the first character contribution weight and the second character contribution weight respectively;

[0130] Calculate the interview contribution vector of the interview item triple according to the first interview contribution weight and the second interview contribution weight;

[0131] The expression for calculating the interview contribution vector of the interview item triple is:

[0132] I contribution =w I1 I1+w I2 I2;

[0133] I contribution represents the interview contribution vector, I1 and I2 represent the expression score and adaptation score respectively, and w I1 and w I2 They represent the contribution weight of the first interview and the contribution weight of the second interview respectively;

[0134] S3-4. Perform feature fusion on the professional contribution vector, the personality contribution vector and the interview contribution vector to obtain a fusion vector of the standard candidate.

[0135] Specifically, the key to this step is to perform weighted fusion of multi-dimensional evaluation items to generate a fusion vector of standard candidates. First, each evaluation item is encoded through a predefined triple structure, and the professional evaluation items, personality evaluation items, and interview evaluation items are converted into professional item triples, personality item triples, and interview item triples, respectively. Then, these triplets are further weighted to obtain contribution pairs for each evaluation, including professional contribution pairs, personality contribution pairs, and interview contribution pairs. In each contribution pair, two weights are included (for example, the first and second weights), which means that the impact of different dimensions of each evaluation on the final result can be adjusted by weights.

[0136] Subsequently, these weights are used to calculate the corresponding contribution vectors. Through weighted calculation, the contribution vectors of professional, personality, and interview items are determined separately. Finally, these three contribution vectors are feature fused to obtain the fusion vector of the standard candidate, which integrates the candidate's performance in different dimensions.

[0137] Furthermore, the step of predefining the triples includes:

[0138] S3-1-1, constructing a triple structure, wherein the structure includes a first element, a second element and a third element;

[0139] S3-1-2. Define the first element as the type index of the evaluation item, and define the second element and the third element as two evaluation items of the same type.

[0140] Specifically, the core of this step is to construct a triple structure to represent the relationship between each evaluation item. First, by constructing a triple structure, each evaluation item is decomposed into three elements: the first element, the second element, and the third element. Among them, the first element represents the type index of the evaluation item, that is, determines which category the evaluation item belongs to (such as professional evaluation, personality evaluation, or interview evaluation). The second element and the third element correspond to two specific evaluation items under the same category.

[0141] The matching step of the resource package matching unit includes:

[0142] S5-1. Reversely decompose the recruitment resource allocation strategy ID to obtain several recruitment resource classification numbers;

[0143] S5-2. According to the mapping of several recruitment resource classification numbers, the classified recruitment resources are matched in the pre-constructed training space to form a recruitment resource package for each standard candidate.

[0144] Specifically, reverse decomposition refers to extracting specific classification numbers of several recruitment resources from the recruitment resource allocation strategy ID. These classification numbers correspond to specific resource types or conditions, such as job categories, salary levels, work locations, etc. Next, by mapping these classification numbers, the system will find the corresponding recruitment resources in the pre-built training space. These resources are integrated into a recruitment resource package for each standard candidate. In this way, suitable recruitment resources can be accurately matched based on the characteristics and needs of the candidates.

[0145] Example 2: See Figures 1 to 4 The technical solution of this embodiment 2 is different from that of embodiment 1 in that a fine-tuning step of the recruitment resource allocation model is disclosed, and the fine-tuning step includes:

[0146] A1. Obtain the prior comprehensive assessment scores and the posterior comprehensive assessment scores of historical employees;

[0147] A2. Calculate the first mean square error of the priori comprehensive measurement score and the posterior comprehensive measurement score;

[0148] The expression of the first mean square error is:

[0149] MSE represents the first mean square error between the prior comprehensive test score and the posterior comprehensive test score, N represents the total number of historical employees, S prior,i represents the prior comprehensive assessment score of the i-th historical employee, Sposterior,i represents the posterior comprehensive assessment score of the i-th historical employee;

[0150] A3. define the first mean square error exceeding the error threshold as the ideal error, and associate several ideal historical employees with the ideal error based on the historical ID;

[0151] A4. Obtain recruitment resource training samples of ideal historical entrants to construct a recruitment resource sample set;

[0152] A5. Iteratively train the neural network model using the recruitment resource training sample set and output the predicted value of the historical recruitment resource strategy ID.

[0153] A6. When the iterative training reaches model convergence, the neural network model is exported as the recruitment resource allocation model.

[0154] Furthermore, the step A1 comprises:

[0155] A1-1. Obtain multi-dimensional historical evaluation items of several historical candidates and mark them with independent historical IDs;

[0156] A1-2. Select historical candidates from a number of historical candidates;

[0157] A1-3. Generate a historical comprehensive assessment score based on the multi-dimensional historical assessment items of the historical employees;

[0158] A1-4, defining the historical comprehensive assessment score of the historical employee as a priori comprehensive assessment score;

[0159] A1-5. Determine the a posteriori comprehensive assessment score of historical entrants based on their post-employment work performance.

[0160] Furthermore, the step A4 comprises:

[0161] A4-1. Obtain some historical recruitment resources of ideal historical candidates during recruitment;

[0162] A4-2. Assign a preset resource classification ID to each historical recruitment resource, and generate several resource classification IDs of ideal historical recruits during recruitment;

[0163] A4-3, concatenating the resource classification IDs to generate a historical recruitment resource strategy ID for an ideal historical employee;

[0164] A4-4. Obtain multi-dimensional historical evaluation items of the ideal historical employee based on the historical ID of the ideal historical employee;

[0165] A4-5. Perform the weighted feature fusion on the multi-dimensional historical evaluation items of the ideal historical employee to generate a fusion vector of the ideal historical employee;

[0166] A4-5. Use the fusion vector of the ideal historical entrants as input features and the historical recruitment resource strategy ID as the target label to construct a recruitment resource sample set, and define it as the training space.

[0167] Specifically, in step A1, the acquisition of the prior comprehensive assessment score and the posterior comprehensive assessment score is the basis for fine-tuning the recruitment resource allocation model. First, by extracting the multi-dimensional historical assessment items from the historical data of historical entrants, an independent historical ID is assigned to each historical entrant, and these assessment items are used to generate a historical comprehensive assessment score. In this embodiment, the acquisition steps and acquisition logic of the historical comprehensive assessment score and the comprehensive assessment score are the same, except that one is based on historical data and the other is based on real-time data.

[0168] Subsequently, based on the post-employment work performance of historical entrants, their actual performance is converted into a posterior comprehensive assessment score. This score reflects the performance of historical entrants in actual work and can be compared with previous predictions to assess whether the historical entrants have achieved the expected performance level.

[0169] In step A4, you first need to obtain the historical recruitment resource data of the ideal historical entrants. These resource data include the resources allocated to the historical entrants during the actual recruitment process. Assign a preset resource classification ID to each historical recruitment resource, and concatenate these resource classification IDs to generate a historical recruitment resource strategy ID, which will serve as an important feature and label. Next, based on the historical ID of the historical entrant, extract its related multi-dimensional historical evaluation items, and perform weighted feature fusion on these evaluation items to generate a fusion vector for the historical entrant. Of course, the acquisition logic of the multi-dimensional historical evaluation items and the aforementioned multi-dimensional evaluation items is the same. The multi-dimensional evaluation items are based on real-time basic candidates, while the multi-dimensional historical evaluation items are based on historical entrants in history.

[0170] Finally, these fused vectors are used as input features together with the historical recruitment resource strategy ID (target label) to construct a recruitment resource sample set, which will be used to train the recruitment resource allocation model and form the training basis of the model.

[0171] The present invention comprehensively considers the multi-dimensional evaluation items of candidates (such as logistics skills, personality traits and interview performance, etc.), and uses the posterior scores in historical data to fine-tune the model, so that the recruitment resource classification strategy ID output by the model is based on the photo resources allocated to the real ideal historical employees in history during recruitment, so that the model can give a dynamically adjusted recruitment resource allocation strategy. This recruitment strategy ID based on the candidate's historical and future performance prediction makes the recruitment resource allocation more accurate. Compared with the traditional static screening and allocation method, it can be optimized in real time according to the candidate's actual performance, ensuring the maximum utilization of recruitment resources and improving the accuracy of talent matching.

[0172] The above embodiments can be implemented in whole or in part by software, hardware, firmware or other arbitrary combinations. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.

[0173] The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0174] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division of a waterway underwater terrain change analysis system and method. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0175] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. An intelligent recruitment resource allocation system based on logistics talent evaluation, characterized in that: include: A comprehensive evaluation score acquisition unit, used to acquire multi-dimensional evaluation items of a number of basic candidates and generate a comprehensive evaluation score based on the multi-dimensional evaluation items; A standard candidate screening unit, used for performing standard screening on a plurality of basic candidates using a comprehensive evaluation score to obtain a standard candidate; A fusion vector unit, used to perform weighted feature fusion on the multi-dimensional evaluation items of the standard candidate to generate a fusion vector of the standard candidate; A resource allocation unit, used for inputting the fusion vector of the standard candidates into the fine-tuned recruitment resource allocation model, and outputting the recruitment resource allocation strategy ID; The resource package matching unit is used to match the recruitment resource package in the training space of the recruitment resource allocation model according to the recruitment resource allocation strategy ID.

2. According to claim 1, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: Obtain multi-dimensional evaluation items of several basic candidates, and generate a comprehensive evaluation score based on the multi-dimensional evaluation items, including: S1-1. Screen a number of basic candidates who meet the basic requirements from the submitted electronic resumes; S1-2, label the basic candidates with independent candidate IDs; S1-3, perform online logistics skill assessment on the basic candidates with the candidate ID marked, and generate several multi-dimensional assessment items for each basic candidate; S1-4, encode several multi-dimensional evaluation items to generate several single-item evaluation scores; S1-5. Perform weighted summation on several individual assessment scores to generate a comprehensive assessment score for each basic candidate.

3. According to claim 1, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: A standard screening is performed on several basic candidates using a comprehensive evaluation score to obtain standard candidates, including: S2-1, presetting the minimum threshold, middle threshold and maximum threshold of the comprehensive evaluation score; S2-2, the comprehensive evaluation score below the minimum threshold is defined as the first scoring range; the comprehensive evaluation score between the minimum threshold and the middle threshold is defined as the second scoring range; the comprehensive evaluation score between the middle threshold and the maximum threshold is defined as the third scoring range; the comprehensive evaluation score exceeding the maximum threshold is defined as the fourth scoring range; S2-3. Define the basic candidates falling within the second score range, the third score range and the fourth score range as standard candidates, and assign independent candidate grades; the candidate grades include: primary candidate grade, senior candidate grade and excellent candidate grade.

4. According to claim 2, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: Perform online logistics skills assessment on basic candidates with marked candidate IDs, and generate several multi-dimensional assessment items for each basic candidate, including: S1-3-1. Execute online assessment of the logistics talents, and classify the assessment results of the online assessment into professional assessment items, personality assessment items, and interview assessment items; The professional assessment items include: logistics skills score and logistics experience score; the personality assessment items include: personality type label and behavior tendency label; the interview assessment items include: oral expression quality score and cultural adaptation score; S1-3-2 defines the combination of the logistics skill score, logistics experience score, personality type label, behavior tendency label, oral expression quality score and cultural adaptation score as the multi-dimensional evaluation item.

5. According to claim 2, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: Encode several multi-dimensional evaluation items and generate several single-item evaluation scores. Encode several multi-dimensional evaluation items and generate several single-item evaluation scores, including: S1-4-1. Standardize the logistics skill score, logistics experience score, oral expression quality score and cultural adaptation score to obtain the professional score, experience score, expression score and adaptation score. S1-4-2, assigning target coding scores to personality type labels and behavioral tendency labels to obtain personality item scores and behavioral item scores; S1-4-3. Define the professional item score, experience item score, expression item score, adaptation item score, personality item score and behavior item score as the single item assessment score.

6. According to claim 1, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: Performing weighted feature fusion on the multi-dimensional evaluation items of the standard candidate to generate a fusion vector of the standard candidate includes: S3-1, perform triple pre-definition on professional evaluation items, personality evaluation items and interview evaluation items to obtain professional item triples, personality item triples and interview item triples; S3-2, pre-define the contribution pairs for the professional item triples, the personality item triples and the interview item triples to obtain the corresponding professional contribution pairs, personality contribution pairs and interview contribution pairs; wherein the professional contribution pair includes the first professional contribution weight and the second professional contribution weight; the personality contribution pair includes the first personality contribution weight and the second personality contribution weight; the interview contribution pair includes the first interview contribution weight and the second interview contribution weight; S3-3 calculates the professional contribution vector of the professional item triple according to the first professional contribution weight and the second professional contribution weight; The expression for calculating the professional contribution vector of the professional term triple is: P contribution =w P1 ·P1+w P2 ·P2; P contribution represents the professional contribution vector, P1 and P2 represent the professional score and experience score respectively, and w P1 and w P2 They represent the contribution weight of the first profession and the contribution weight of the second profession respectively; Calculate the character contribution vector of the character item triple according to the first character contribution weight and the second character contribution weight; The expression for calculating the character contribution vector of the character item triple is: C contribution =in C1 ·C1+in C2 ·C2; C contribution represents the personality contribution vector, C1 and C2 represent the personality item score and behavior item score respectively, and w C1 and w C2 Represent the first character contribution weight and the second character contribution weight respectively; Calculate the interview contribution vector of the interview item triple according to the first interview contribution weight and the second interview contribution weight; The expression for calculating the interview contribution vector of the interview item triple is: I contribution =w I1 ·I1+w I2 ·I2; I contribution represents the interview contribution vector, I1 and I2 represent the expression score and adaptation score respectively, and w I1 and w I2 They represent the contribution weight of the first interview and the contribution weight of the second interview respectively; S3-4. Perform feature fusion on the professional contribution vector, the personality contribution vector and the interview contribution vector to obtain a fusion vector of the standard candidate.

7. According to claim 1, a recruitment resource intelligent allocation system based on logistics talent evaluation is characterized in that: According to the recruitment resource allocation strategy ID, matching the recruitment resource package in the training space of the recruitment resource allocation model includes: S5-1. Reversely decompose the recruitment resource allocation strategy ID to obtain several recruitment resource classification numbers; S5-2. According to the mapping of several recruitment resource classification numbers, the classified recruitment resources are matched in the pre-constructed training space to form a recruitment resource package for each standard candidate.

8. The intelligent recruitment resource allocation system based on logistics talent evaluation according to claim 1 is characterized in that: The steps for fine-tuning the recruitment resource allocation model include: A1. Obtain the prior comprehensive assessment scores and the posterior comprehensive assessment scores of historical employees; A2. Calculate the first mean square error of the priori comprehensive measurement score and the posterior comprehensive measurement score; The expression of the first mean square error is: MSE represents the first mean square error between the prior comprehensive test score and the posterior comprehensive test score, N represents the total number of historical employees, S prior,i represents the prior comprehensive assessment score of the i-th historical employee, S posterior,i represents the posterior comprehensive assessment score of the i-th historical employee; A3. define the first mean square error exceeding the error threshold as the ideal error, and associate several ideal historical employees with the ideal error based on the historical ID; A4. Obtain recruitment resource training samples of ideal historical entrants to construct a recruitment resource sample set; A5. Iteratively train the neural network model using the recruitment resource training sample set and output the predicted value of the historical recruitment resource strategy ID. A6. When the iterative training reaches model convergence, the neural network model is exported as the recruitment resource allocation model.

9. The intelligent recruitment resource allocation system based on logistics talent evaluation according to claim 1 is characterized in that: Obtain the prior comprehensive assessment scores and the posterior comprehensive assessment scores of historical employees, including: A1-1. Obtain multi-dimensional historical evaluation items of several historical candidates and mark them with independent historical IDs; A1-2. Select historical candidates from a number of historical candidates; A1-3. Generate a historical comprehensive assessment score based on the multi-dimensional historical assessment items of the historical employees; A1-4, defining the historical comprehensive assessment score of the historical employee as a priori comprehensive assessment score; A1-5. Determine the a posteriori comprehensive assessment score of historical entrants based on their post-employment work performance.

10. The intelligent recruitment resource allocation system based on logistics talent evaluation according to claim 1 is characterized in that: Obtain recruitment resource training samples of ideal historical entrants to build a recruitment resource sample set, including: A4-1. Obtain some historical recruitment resources of ideal historical candidates during recruitment; A4-2. Assign a preset resource classification ID to each historical recruitment resource, and generate several resource classification IDs of ideal historical recruits during recruitment; A4-3, concatenating the resource classification IDs to generate a historical recruitment resource strategy ID for an ideal historical employee; A4-4. Obtain multi-dimensional historical evaluation items of the ideal historical employee based on the historical ID of the ideal historical employee; A4-5. Perform the weighted feature fusion on the multi-dimensional historical evaluation items of the ideal historical employee to generate a fusion vector of the ideal historical employee; A4-5. Use the fusion vector of the ideal historical entrants as input features and the historical recruitment resource strategy ID as the target label to construct a recruitment resource sample set, and define it as the training space.

Images