A labor data matching system, method and storage medium thereof

Abstract

The present invention provides a method, system and storage medium for matching employment data. By means of feature matrix modeling, feature-based digital modeling is performed on resigned personnel and unemployed personnel, thereby solving the cold start problem of data mining for newly added unemployed personnel in the system. By means of a feature matrix similarity calculation algorithm, unemployed personnel similar to the resigned personnel are found based on the feature matrix of the resigned personnel, thereby improving the efficiency of data mining and the accuracy of job matching. By means of a text comparison method based on sparse vectors, the data matching efficiency is further improved.

Description

Technical Field

[0001] The present invention belongs to the field of big data, and in particular relates to an employment data matching system, method and storage medium. Background Art

[0002] With the development of the economy, a large number of unemployed people need jobs. Due to the limitations of free job search, many people cannot find jobs that match their abilities. Big data in modern technology can help these unemployed people through data matching mechanisms.

[0003] With the development of big data, the task matching mechanism has also been developed. Big data needs to capture, manage and process a large number of data sets within a certain time frame. It is a massive, high-growth and diversified information asset that requires new processing models to have stronger decision-making power, insight discovery and process optimization capabilities. The traditional task matching method is time-consuming and resource-intensive. As the task volume increases, matching information errors and omissions may occur, making it difficult to meet existing needs. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method for matching employment data, and the method steps are as follows:

[0005] Step S101: Establishing a feature matrix of on-call personnel, establishing a feature matrix for the staff currently on-call, and initializing the system;

[0006] Step S102: Establishing a resigned staff feature matrix, establishing a staff feature matrix for the currently resigned staff, and initializing the system;

[0007] Step S103: polling and comparing feature similarity, comparing the similarity between the two feature matrices through an algorithm, performing circular comparison on the work matrices of the employed personnel, and selecting the person with the highest similarity as the recommended person to be matched;

[0008] Step S104: comparing the similarity of the job description text information, comparing the job description text information of the recommended personnel to be matched with the resigned personnel, calculating the similarity, and adding the waiting personnel to the recommendation information list according to the similarity;

[0009] Step S105: Two-way confirmation and feedback of recommendation results. The recommendation results are sent to the enterprise and the individual for two-way confirmation, and the confirmation results are fed back.

[0010] In addition, the present invention also includes a labor data matching system, which includes a feature matrix modeling unit for on-call personnel, a feature matrix modeling unit for off-call personnel, a feature matrix comparison unit, a text similarity comparison unit, a confirmation feedback unit and a system database.

[0011] The on-call staff feature matrix modeling unit establishes a on-call staff feature matrix, establishes a staff feature matrix for the staff currently on-call, and performs system initialization.

[0012] The off-duty personnel feature matrix modeling unit establishes a off-duty personnel feature matrix, establishes a personnel feature matrix for the currently off-duty staff, and performs system initialization.

[0013] The feature matrix comparison unit calculates the similarity of the two matrices, and selects the person with the highest similarity as the recommended person to be matched by performing a circular comparison on the work matrices of the persons to be employed.

[0014] The text similarity comparison unit calculates the similarity of the job description text information, compares the job description text information of the recommended personnel to be matched with the resigned personnel, calculates the similarity, and adds the waiting personnel to the recommendation information list according to the similarity.

[0015] The confirmation feedback unit bidirectionally confirms and feeds back the recommendation result, and the recommendation result is sent to the enterprise and the individual for bidirectional confirmation, and the confirmation result is fed back.

[0016] The database receives and stores the above-mentioned data results and personnel information, etc.

[0017] In addition, the present invention also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the employment data matching method as described above are implemented.

[0018] In summary, the present invention provides a method, system and storage medium for employment data matching. Through feature matrix modeling, feature-based digital modeling is performed on resigned employees and unemployed employees, which solves the cold start problem of data mining for newly added unemployed employees in the system. Through a feature matrix similarity calculation algorithm, based on the feature matrix of the resigned employees, unemployed employees who are similar to the resigned employees are found, thereby improving the efficiency of data mining and the accuracy of job matching. The data matching efficiency is further improved through a text comparison method based on sparse vectors. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of the steps of the employment data matching method;

[0020] Figure 2 This is a system block diagram of the employment data matching method. DETAILED DESCRIPTION

[0021] 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 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.

[0022] The purpose of the present invention is to provide a method, system and storage medium for employment data matching. Through feature matrix modeling, feature-based digital modeling is performed on resigned personnel and unemployed personnel, thereby solving the cold start problem of data mining for newly added unemployed personnel in the system. Through a feature matrix similarity calculation algorithm, based on the feature matrix of the resigned personnel, unemployed personnel similar to the resigned personnel are found, thereby improving the efficiency of data mining and the accuracy of job matching. Through a sparse vector-based text comparison method, the data matching efficiency is further improved.

[0023] like Figure 1 The steps of the employment data matching method of the present invention are as follows:

[0024] Step S101: Establishing a feature matrix of on-call personnel, establishing a feature matrix for the staff currently on-call, and initializing the system;

[0025] Step S102: Establishing a resigned staff feature matrix, establishing a staff feature matrix for the currently resigned staff, and initializing the system;

[0026] Step S103: polling and comparing feature similarity, comparing the similarity between the two feature matrices through an algorithm, performing circular comparison on the work matrices of the employed personnel, and selecting the person with the highest similarity as the recommended person to be matched;

[0027] Step S104: comparing the similarity of the job description text information, comparing the job description text information of the recommended personnel to be matched with the resigned personnel, calculating the similarity, and adding the waiting personnel to the recommendation information list according to the similarity;

[0028] Step S105: Two-way confirmation and feedback of recommendation results. The recommendation results are sent to the enterprise and the individual for two-way confirmation, and the confirmation results are fed back.

[0029] Among them, in step S101 and step S102, the characteristics of the waiting personnel and the resigned personnel are the same, and the characteristic matrix of the waiting personnel is expressed as G N Y (A, B, C, D, E, F), the characteristic matrix of the resigned personnel is represented by D N Y(A, B, C, D, E, F), where N represents the personnel number; Y represents the personnel job description text information; A represents age information, which is divided into 8 categories, represented by 0 to 7, 0 represents the age of 0 to 17 years old, 1 represents the age of 18 to 23 years old, 2 represents the age of 24 to 34 years old, 3 represents the age of 35 to 40 years old, 4 represents the age of 41 to 50 years old, 5 represents the age of 51 to 55 years old, 6 represents the age of 56 to 60 years old, and 6 represents the age of more than 60 years old. B represents gender information, which is divided into two categories, 1 represents male and 0 represents female. C represents the major category information of the type of work, which is divided into 8 categories, represented by 0 to 7. D represents the medium category information of the type of work, which is divided into 66 medium categories of work, represented by 0 to 65. F represents the small category information of the type of work, which is divided into 413 small categories of work, represented by 0 to 412. F is the job level information, which is divided into 4 categories, with 0 to 3 representing elementary, intermediate, senior and expert respectively.

[0030] Furthermore, in step S103, the feature matrix similarity calculation method is as follows:

[0031] Since the existing comparison calculations are heavy and time-consuming, the present invention adopts polling calculation, that is, before calculating the similarity of the feature matrix, a distributed calculation pre-step is added to perform concurrent calculations. During concurrent calculations, s intelligent processors process simultaneously to quickly find the matching value with the highest similarity.

[0032] Initialize s intelligent processors, each of which is allocated with storage space, CPU and other computing resources, and can independently perform computing tasks. Each processor is set with a monitoring task to count the historical computing workload dat and average computing time cot of the processor.

[0033] Initialize the text value list TEXT N Y , the feature matrix D of the waiting personnel N Y The parameters in (A, B, C, D, E, F) are removed in turn and saved to the corresponding text value list, and the text value list TEXT is calculated. N Y Hash value, get the hash value list HASH N Y . HASH the hash value list N Y Each value in is converted to the corresponding decimal number and saved in the decimal hash list TEN N Y .According to the decimal hash list TEN N Y Sort the decimal values ​​from large to small to get the sorted decimal hash list ASS NY The purpose of converting to decimal in this step is to facilitate storage in the database for later identification.

[0034] By formula Calculate the computing power of s intelligent processors, where dat v represents the historical calculation times of the vth intelligent processor, cot v Represents the average computing time of the vth intelligent processor.

[0035] Select the current h v The maximum value, and the idle intelligent processor is used as the processor C to receive the computing task. The feature matrix G of the currently compared off-duty personnel N Y The parameters in (A, B, C, D, E, F) are converted into text and saved in the text LT, LT is converted into a hash value and saved in the hash text HT, HT is converted into a decimal number and saved in TT.

[0036] In decimal hash list ASS N Y Find the person whose decimal hash value TT is closest to the currently compared off-duty person, and use its feature matrix as the calculation object of the intelligent processor C. Each time a calculation task is selected, the decimal hash list ASS is selected first. N Y Among them, other unemployed persons who are closest to the position of the currently compared resigned persons.

[0037] Specifically, after finding other unemployed persons who are closest to the position of the currently compared resigned person, the similarity of the feature matrices of the two is calculated. The calculation formula is as follows:

[0038]

[0039] Where P is the total number of parameters in the feature matrix, P = 6; where X represents the similarity coefficient, and D(n) represents the feature matrix G of off-duty personnel N Y The nth parameter value in (A, B, C, D, E, F); G(n) represents the characteristic matrix D of the waiting personnel N Y The nth parameter value in (A,B,C,D,E,F).

[0040] The similarity coefficient calculation method is as follows:

[0041]

[0042] Where X represents the similarity coefficient; M represents the number of identical A, B, C, D, E, and F in the feature matrix of waiting employees and the feature matrix of resigned employees; P is the total number of parameters in the feature matrix, P = 6.

[0043] Furthermore, in step S104, the method for calculating the similarity of the job description text information is as follows:

[0044] Among them, the jieba word segmentation tool is used to segment the texts Y1 and Y2, and two segmented texts F1 and F2 are obtained;

[0045] Use the defaultdict method to calculate the word frequency information of the two segmented texts F1 and F2 to obtain the word frequency texts C1 and C2;

[0046] The top 10 words with the highest frequency are selected to form the word frequency selected texts J1 and J2;

[0047] Use the doc2bow tool to convert the word frequency selected texts J1 and J2 into sparse vectors S1 and S2;

[0048] Process the sparse vectors S1 and S2 to obtain new lexicons N1 and N2;

[0049] Process the new vocabulary N1 and N2 through the TF-IDF model to obtain the TF-IDF value;

[0050] Get the feature values ​​T1 and T2 corresponding to the TF-IDF value through token2id;

[0051] The sparse matrix similarity of the feature values ​​T1 and T2 is calculated through the sparse matrix similarity method of similarities. The sparse matrix similarity is the similarity of the job description text information.

[0052] It should be noted that the unemployed personnel whose sparse matrix similarity is greater than 40% are added to the recommended information list, and if the sparse matrix similarity is less than 40%, the comparison is repeated for the next unemployed personnel's work matrix.

[0053] Furthermore, in step S105, the recommendation result is mutually confirmed and fed back, and the recommendation result is sent to the enterprise and the individual for mutual confirmation, and the confirmation result is fed back.

[0054] In order to implement the above employment data matching method, the present invention also includes an employment data matching system.

[0055] like Figure 2 As shown in the block diagram of the employment data matching system, the system includes a feature matrix modeling unit for on-call personnel, a feature matrix modeling unit for off-call personnel, a feature matrix comparison unit, a text similarity comparison unit, a confirmation feedback unit and a system database.

[0056] The on-call staff feature matrix modeling unit establishes a on-call staff feature matrix, establishes a staff feature matrix for the staff currently on-call, and performs system initialization.

[0057] The off-duty personnel feature matrix modeling unit establishes a off-duty personnel feature matrix, establishes a personnel feature matrix for the currently off-duty staff, and performs system initialization.

[0058] The feature matrix comparison unit calculates the similarity of the two matrices, and selects the person with the highest similarity as the recommended person to be matched by performing a circular comparison on the work matrices of the persons to be employed.

[0059] The text similarity comparison unit calculates the similarity of the job description text information, compares the job description text information of the recommended personnel to be matched with the resigned personnel, calculates the similarity, and adds the waiting personnel to the recommendation information list according to the similarity.

[0060] The confirmation feedback unit bidirectionally confirms and feeds back the recommendation result, and the recommendation result is sent to the enterprise and the individual for bidirectional confirmation, and the confirmation result is fed back.

[0061] The database receives and stores the above-mentioned data results and personnel information, etc.

[0062] In addition, the present invention also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the employment data matching method as described above are implemented.

[0063] By using the employment data matching method, system and storage medium proposed in this embodiment, unemployed persons similar to resigned persons are found, the efficiency of data mining is improved, the accuracy of job matching is improved, and the data matching efficiency is further improved through a sparse vector-based text comparison method.

[0064] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0065] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A method for matching employment data, characterized in that: Here are the steps: Step S101: Establishing a feature matrix of on-call personnel, establishing a feature matrix for the staff currently on-call, and initializing the system; Step S102: Establishing a resigned staff feature matrix, establishing a staff feature matrix for the currently resigned staff, and initializing the system; The characteristics of the waiting personnel are the same as those of the resigned personnel, including the personnel number N, personnel job description text information Y, age information A, gender information B, major job category information C, medium job category information D, minor job category information E and job level information F; Step S103: polling and comparing feature similarity, comparing the similarity between the above two feature matrices through an algorithm, performing cyclic comparison on the work matrices of the personnel to be posted, and selecting the personnel with the highest similarity as the recommended personnel to be matched; include, Initialize s intelligent processors, allocate computing resources to each intelligent processor, the computing resources include storage space and CPU computing resources, and each intelligent processor can independently perform computing tasks; set a monitoring task for each processor to count the historical computing times dat and the average computing time cot of the processor; Initialize a list of text values , the feature matrix of waiting personnel The parameters are taken out and saved in the corresponding text value list in turn, and the text value list is calculated. Hash value of the hash value list ; List the hash values Convert each value in to the corresponding decimal number and save it in a decimal hash list , according to the decimal hash list Sort the decimal values ​​from large to small to get the sorted decimal hash list ; By formula Calculate the computing power of s intelligent processors, where represents the historical calculation times of the vth intelligent processor, represents the average computing time of the vth intelligent processor; Select Current The maximum value and idle intelligent processor is used as the processor C to receive the computing task; the characteristic matrix of the currently compared resigned personnel The parameters are converted into text and saved into text LT, LT is converted into hash value and saved into hash text HT, HT is converted into decimal number and saved into TT; In decimal hash list Find the person whose decimal hash value TT is closest to the current compared resigned person, and use its feature matrix as the calculation object of the intelligent processor C. Each time a calculation task is selected, the decimal hash list is selected first. Other waiting employees who are closest to the position of the currently compared resigned employee; Specifically, after finding other waiting employees who are closest to the position of the currently compared resigned employee, the similarity of the feature matrices of the two is calculated. The calculation formula is as follows: Where P is the total number of parameters in the feature matrix, P=6; where X represents the similarity coefficient, and D(n) represents the feature matrix of the resigned personnel The nth parameter value in; G(n) represents the characteristic matrix of waiting personnel The nth parameter value in ; The similarity coefficient calculation method is as follows: ； Where X represents the similarity coefficient; M represents the number of identical A, B, C, D, E, and F in the feature matrix of waiting employees and the feature matrix of resigned employees; P represents the total number of parameters in the feature matrix, P=6; Step S104: comparing the similarity of the job description text information, comparing the job description text information of the recommended personnel to be matched with the resigned personnel, calculating the similarity of the job description text information, and adding the waiting personnel to the recommended information list according to the similarity; Step S105: Two-way confirmation and feedback of recommendation results. The recommendation results are sent to the enterprise and the individual for two-way confirmation, and the confirmation results are fed back.

2. The employment data matching method according to claim 1, characterized in that: The method for calculating the similarity of the job description text information is specifically as follows: Segment the first text and the second text to obtain a first segmented text and a second segmented text; Calculate word frequency information of the first word segmentation text and the second word segmentation text to obtain a first word frequency text and a second word frequency text; Select the first n words with the highest frequency in the first word frequency text and the second word frequency text to form a first word frequency selected text and a second word frequency selected text; Convert the first word frequency selected text and the second word frequency selected text into a first sparse vector and a second sparse vector; Processing the first sparse vector and the second sparse vector to obtain a first new vocabulary and a second new vocabulary; Process the first new vocabulary and the second new vocabulary through the TF-IDF model to obtain corresponding TF-IDF values; Further calculate and obtain the first eigenvalue and the second eigenvalue corresponding to the TF-IDF value; The sparse matrix similarity of the first eigenvalue and the second eigenvalue is calculated, and the sparse matrix similarity is recorded as the job description text information similarity.

3. The employment data matching method according to claim 2, characterized in that: The n is 10.

4. The employment data matching method according to claim 2, characterized in that: If the sparse matrix similarity is greater than the preset sparse threshold, the recommended information list is added. If the sparse matrix similarity is less than the preset sparse threshold, the comparison is repeated for the next work matrix of the on-call personnel.

5. The employment data matching method according to claim 4, characterized in that: The preset sparse threshold is 0.

4.

6. A labor data matching system, characterized in that: It includes a feature matrix modeling unit for waiting personnel, a feature matrix modeling unit for resigned personnel, a feature matrix comparison unit, a text similarity comparison unit, and a confirmation feedback unit; The standby staff feature matrix modeling unit is used to establish a standby staff feature matrix, establish a staff feature matrix for the staff currently on standby, and perform system initialization; The resigned personnel feature matrix modeling unit is used to establish a resigned personnel feature matrix, establish a personnel feature matrix for the currently resigned staff, and initialize the system; The characteristics of the waiting personnel are the same as those of the resigned personnel, including the personnel number N, personnel job description text information Y, age information A, gender information B, major job category information C, medium job category information D, minor job category information E and job level information F; The feature matrix comparison unit is used to poll and compare feature similarities, compare the similarities between the two feature matrices through an algorithm, perform cyclic comparisons on the work matrices of the personnel to be on duty, and select the personnel with the highest similarity as the recommended personnel to be matched; include, Initialize s intelligent processors, allocate computing resources to each intelligent processor, the computing resources include storage space and CPU computing resources, and each intelligent processor can independently perform computing tasks; set a monitoring task for each processor to count the historical computing times dat and the average computing time cot of the processor; Initialize a list of text values , the feature matrix of waiting personnel The parameters are taken out and saved in the corresponding text value list in turn, and the text value list is calculated. Hash value of the hash value list ; List the hash values Convert each value in to the corresponding decimal number and save it in a decimal hash list , according to the decimal hash list Sort the decimal values ​​from large to small to get the sorted decimal hash list ; By formula Calculate the computing power of s intelligent processors, where represents the historical calculation times of the vth intelligent processor, represents the average computing time of the vth intelligent processor; Select Current The maximum value and idle intelligent processor is used as the processor C to receive the computing task; the characteristic matrix of the currently compared resigned personnel The parameters are converted into text and saved into text LT, LT is converted into hash value and saved into hash text HT, HT is converted into decimal number and saved into TT; In decimal hash list Find the person whose decimal hash value TT is closest to the currently compared off-duty person, and use its feature matrix as the calculation object of the intelligent processor C. Each time a calculation task is selected, the decimal hash list is selected first. Other waiting employees who are closest to the position of the currently compared resigned employee; Specifically, after finding other waiting employees who are closest to the position of the currently compared resigned employee, the similarity of the feature matrices of the two is calculated. The calculation formula is as follows: Where P is the total number of parameters in the feature matrix, P=6; where X represents the similarity coefficient, and D(n) represents the feature matrix of the resigned personnel The nth parameter value in; G(n) represents the characteristic matrix of waiting personnel The nth parameter value in ; The similarity coefficient calculation method is as follows: ； Where X represents the similarity coefficient; M represents the number of identical A, B, C, D, E, and F in the feature matrix of waiting employees and the feature matrix of resigned employees; P represents the total number of parameters in the feature matrix, P=6; The text similarity comparison unit is used to compare the similarity of job description text information, compare the job description text information of the recommended personnel to be matched and the resigned personnel, calculate the similarity of the job description text information, and add the waiting personnel to the recommended information list according to the similarity; The confirmation feedback unit is used for two-way confirmation and feedback of recommendation results. The recommendation results are sent to enterprises and individuals for two-way confirmation and feedback of confirmation results.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images