A talent information intelligent matching method

Abstract

The application provides a talent information intelligent matching method, and belongs to the intelligent matching field; solves the problem of difficult talent information searching; and specifically comprises the following steps: S1: obtaining talent information and enterprise employment demand information; S2: obtaining employment demand standards, which are divided into rigid requirements and flexible requirements; obtaining necessary items according to the rigid requirements; obtaining bonus items according to the flexible requirements; S3: screening the talent information through the necessary items to obtain optimized talent information, and performing similarity matching with the flexible requirements to obtain a similarity value; S4: setting a weight for the flexible requirements; constructing a matching score model according to the similarity value and the weight to score the talents; and S5: sorting the score results, and recommending talents to the enterprise in combination with the enterprise employment demand information; the application provides data reference for the enterprise through intelligent matching of the talent information, reduces the recruitment cost, and improves the recruitment efficiency.

Description

TECHNICAL FIELD

[0001] The application is a talent information intelligent matching method based on talent information, and relates to the field of intelligent matching. BACKGROUND

[0002] The existing talent information intelligent matching method has the following disadvantages:

[0003] Insufficient user demand individualization: Different enterprises (such as start-ups and large enterprises) have significant differences in talent requirements, but the system mostly matches talents through existing matching rules, which is difficult to meet the needs of different enterprises;

[0004] Insufficient matching accuracy: The existing talent information intelligent matching method matches talents through a single dimension (such as work experience and skill technology), and the matching result is one-sided. Most matching methods directly output the matching result by matching the text semantics, resulting in poor matching result accuracy;

[0005] High interaction complexity: The talent information intelligent matching method capable of multi-dimensional matching usually requires users to set the weight parameters of different dimensions independently, which has a high operation threshold and is complex to use, making it difficult to be widely used. SUMMARY

[0006] In view of the deficiencies of the prior art, the application aims to provide a talent information intelligent matching method based on talent information, which aims to solve the problem of difficult talent information search.

[0007] In order to achieve the above-mentioned purpose, the application is implemented by the following technical solution: a talent information intelligent matching method based on talent information, the matching method comprising:

[0008] Step S1: obtaining talent information and enterprise's human resource demand information;

[0009] Step S2: processing the enterprise's human resource demand information to obtain human resource demand standards, dividing the human resource demand standards into hard requirements and flexible requirements; obtaining necessary items according to the hard requirements; obtaining bonus items according to the flexible requirements;

[0010] Step S3: processing the talent information, filtering the talent information through the necessary items to obtain optimized talent information; performing similarity matching between the optimized talent information and the bonus items to obtain a similarity value; sorting the similarity value and performing initial scoring according to the sorting result;

[0011] Step S4: setting a weight according to the numerical distribution of the similarity value; constructing a matching score model from the similarity value, the initial score and the weight to score the talents;

[0012] Step S5: sorting the scoring results, and recommending talents to the enterprise in combination with the enterprise's human resource demand information.

[0013] Further, the specific steps of the step S2 are as follows:

[0014] Step S21: obtaining the enterprise's human demand information; performing structural field conversion on the enterprise's human demand information to obtain a human demand standard, the human demand standard being represented by a standard dimension and a standard description;

[0015] Step S22: extracting a hard requirement in the human demand standard according to the human demand standard, obtaining a standard dimension and a standard description corresponding to the hard requirement, integrating the standard dimension and the standard description, recording the integrated standard dimension as a necessary dimension, recording the integrated standard description as a necessary description, and obtaining a necessary item; the necessary item is {necessary dimension: necessary description};

[0016] extracting an elastic requirement in the human demand standard, obtaining a standard dimension and a standard description corresponding to the elastic requirement, integrating the standard dimension and the standard description, recording the integrated standard dimension as a scoring dimension, recording the integrated standard description as a scoring description, and obtaining a scoring item; the scoring item is {scoring dimension: scoring description}.

[0017] Further, the specific steps of the step S21 are as follows:

[0018] Step S211: performing text preprocessing on the enterprise's human demand information, extracting a text of the human demand information, and performing loop judgment on the text; when the text content is irrelevant characters, format markers, or stop words, deleting the text content to obtain a cleaned text;

[0019] Step S212: performing word segmentation processing on the cleaned text, identifying key entities in the text, classifying the key entities to obtain a classification dimension, recording key entities under the same classification dimension as dimension content, and obtaining a human demand standard {standard dimension: standard description} according to the classification dimension and the dimension content.

[0020] Further, the specific steps of the step S211 are as follows:

[0021] Removing emoticons: obtaining the Unicode encoding bq1 of the first emoticon in the computer, obtaining the Unicode encoding bq2 of the last emoticon in the computer, obtaining the encoding interval BQJ of the emoticons according to the encoding range [bq1, bq2] of the emoticons, converting the human demand information into character encodings, counting the number bm of the character encodings, recording the character encodings as ZF(1), ZF(2), …, ZF(bm), and replacing the character encodings with null if ZF(1) to ZF(bm) ∈ BQJ;

[0022] Removing format marks: obtaining the types of format marks z, marking the format marks as gs1, gs2, …, gsz; constructing a format list, taking the format marks gs1 to gsz as list elements, populating the format list, matching the human demand information with gs1 to gsz according to the format list, deleting the corresponding content of gs1 to gsz in the human demand information, and completing the cleaning of the format list;

[0023] Specifically as follows:

[0024] Binary encoding the format marks, obtaining the length r of the binary encoding, marking the encoding bit as x, and marking the value on the encoding bit as y, constructing a rectangular coordinate system, and mapping the binary encoding of the format marks in the rectangular coordinate system; obtaining the point set {(x1, y1), (x2, y2), …, (xr, yr)} of the format encoding, connecting the points with y value of 0 and y value of 1 according to the point set {(x1, y1) to (xr, yr)}, and calculating the connection length thereof:

[0025]

[0026] Wherein: cd0 represents the connection length when y value is 0; x w and y w represent the wth x value and y value;

[0027]

[0028] Wherein: cd1 represents the connection length when y value is 1;

[0029] According to cd0 and cd1, the human demand information is matched and deleted, and the cleaning of the format list is completed;

[0030] Removing stop words: obtaining the types of stop words t, marking the stop words as TY1, TY2, …, TYt; constructing a stop word list by taking the stop words TY1 to TYt as elements, representing the stop word list by a regular expression, searching for the stop words in the human demand information, and deleting the searched stop words.

[0031] Further, the specific steps of the step S3 are as follows:

[0032] Step S31: obtaining talent information, obtaining necessary dimensions and necessary descriptions according to necessary items, matching and positioning the talent information according to the necessary dimensions, matching the necessary descriptions with the talent information, extracting the talent information that meets the necessary items, and deleting the talent information that does not meet the necessary items; obtaining optimized talent information;

[0033] Step S32: According to the bonus item, the bonus dimension and the bonus description are obtained; the bonus dimension is matched with the optimized talent information, the similarity value is obtained by similarity calculation combined with the bonus description, the similar values of the same dimension are sorted according to the bonus dimension, and the initial score is assigned according to the sorting result.

[0034] Further, the specific steps of step S31 are as follows:

[0035] Step S311: The number of dimensions a of the necessary dimensions is obtained, and the necessary dimensions are denoted as bwd1, bwd2, …, bwda; the necessary descriptions corresponding to the necessary dimension items are denoted as bms1, bms2, …, bmsa; the talent information is matched and positioned according to the necessary dimensions in turn; and the necessary descriptions are matched with the talent information according to the matching results;

[0036] The necessary dimensions are taken as the judgment conditions, the talent information is traversed, and the necessary dimensions are matched with the talent information by character comparison; if there is no description of the necessary dimension in the talent information, the talent information is deleted;

[0037] If there is a description of the necessary dimension in the talent information, the corresponding content is extracted and matched with the necessary description, and the talent information is deleted according to the matching result;

[0038] The necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa are matched;

[0039] Step S312: If the necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa can all obtain corresponding matching content in the talent information, the talent information is reserved; the reserved talent information is integrated to obtain the optimized talent information.

[0040] Further, the specific steps of step S312 are as follows:

[0041] The optimized talent information list is constructed, the necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa can all obtain corresponding matching talent information in the talent information, and the necessary dimensions and the necessary descriptions in the talent information are deleted;

[0042] The necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa are XORed with the talent information that can be matched, the talent information is simplified, and the simplified talent information is used as a list element to fill the optimized talent information list;

[0043] The elements in the optimized talent information list are sorted; the data storage value of the simplified talent information is obtained, and the talent information is sorted according to the size of the data storage value;

[0044] Obtaining the number of elements ysl in the optimized talent information list, and recording the elements in the optimized talent information list as ylb[ys]; taking the first talent information in the optimized talent information list as a comparison value bj; and performing a loop comparison on the optimized talent information list according to the comparison value;

[0045] If bj> ylb[ys], then replacing ylb[ys] with bj, and comparing ylb[ys+1] until ys=ysl;

[0046] If bj≤ ylb[ys], then comparing ylb[ys+1] until ys=ysl;

[0047] When the loop is completed, the sorting of the optimized talent information is completed, and the sorting is in descending order of the talent information data storage value; an optimized talent information list after sorting is obtained, and the optimized talent information is called according to the optimized talent information list after sorting.

[0048] Further, the specific steps of the step S32 are as follows:

[0049] Step S321: obtaining the number b of bonus dimensions, recording the bonus dimensions as jwd1, jwd2, …, jwdb; recording the bonus descriptions corresponding to the bonus dimensions as jms1, jms2, …, jmsb; taking the bonus dimensions as a retrieval target, retrieving the optimized talent information; obtaining retrieval content, matching the retrieval content with the bonus descriptions, and obtaining a similarity value;

[0050] Step S322: obtaining the similarity values of the b bonus dimensions, obtaining a similarity value list xsl, xsl=[xsz1, xsz2, …, xszb]; obtaining the number f of optimized talent information; performing descending order sorting on the similarity values of the b bonus dimensions according to the number f of optimized talent information and the similarity value list xsl; and counting the sorting positions px to obtain a sorting list bxl, bxl=[px1, px2, …, pxb];

[0051] Step S323: according to the sorting list, the initial score of the optimized talent information is assigned; the sorting position is mapped to the interval [1, 2] to assign the score, and the specific score assignment is as follows:

[0052] Obtaining the sorting position px, the number f of optimized talent information, and the mapping interval [1, 2]; according to the mapping interval [1, 2], the sorting position px is mapped and calculated to obtain the initial score cff;

[0053]

[0054] Further, the specific steps of step S321 are as follows:

[0055] Obtain the search content, perform word segmentation processing on the search content to obtain a search word list, obtain the number c of search words, and mark the search words as js1, js2, …, andjsc; perform word segmentation processing on the bonus description to obtain a description word list, obtain the number d of description words, and mark the description words as ms1, ms2, …, andmsd;

[0056] Match the search words js1 with the description words ms1 to msd respectively to obtain a semantic matching value, obtain the maximum semantic matching value as the matching value xfz of the search words; judge the matching value of the search words,

[0057] If xfz>0.5, it is judged that the search words and the description words can be matched, and the matching value of the search words is saved;

[0058] If xfz≤0.5, it is judged that the search words and the description words cannot be matched, and the search value is deleted;

[0059] Similarly, match js2 tojscwith the description words ms1 to msd to obtain the matching value of the search words; count the matching values of the search words and the description words that can be matched to obtain a matching value list;

[0060] Obtain the number e of matching values in the matching value list; accumulate the matching values to obtain a similarity value xsz:

[0061]

[0062] xfz i represents the ith matching value.

[0063] Further, the specific steps of step S4 are as follows:

[0064] Step S41: Obtain the similarity values of the b bonus dimensions, calculate the variance of the similarity values of the b bonus dimensions, and set the weight of the score calculation of the bonus dimensions according to the variance of the similarity values, which is as follows:

[0065] Obtain the number f of optimized talent information, and mark the similarity values of the b bonus dimensions as A(U, V) according to the number f of optimized talent information;

[0066] Calculate the average of the similarity values under the same dimension to obtain the average similarity value xjz;

[0067]

[0068] Wherein: xjz(U) represents the average value of the similarity value in the Uth dimension, A(U, V) represents the similarity value of the Uth bonus dimension and the Vth optimization talent information;

[0069] According to the average value of the similarity value xjz(U) and the similarity value A(U, V), the variance fc is calculated:

[0070]

[0071] Wherein: fc(U) represents the variance in the Uth dimension;

[0072] The variance is set as the calculation weight qz of the bonus dimension, that is, fc(U) = qz(U);

[0073] Step S42: Obtain the initial score cff(U), and combine the similarity value xszU and the weight qz(U) to construct a matching score model pmx:

[0074]

[0075] Wherein: b is the number of bonus dimensions, and xszU represents the similarity value in the Uth dimension;

[0076] Step S43: The similarity value of the optimization talent information, the initial score and the weight are substituted into the matching score model for calculation to obtain the score of the optimization talent information, and the optimization talent information is recommended according to the score.

[0077] Compared with the prior art, the beneficial effects of the present application are:

[0078] Improve the individualization of user demand: the demand of the user is obtained, the text is processed according to the user demand, the matching rule is formulated according to the user demand, the individualization of the user demand is improved, different matching rules are provided for different users, and the matching demand of different users is met;

[0079] Refine the matching result: according to the individualized demand of the user, the demand is analyzed to obtain a specific matching dimension, the matching is performed according to the matching dimension, the matching result is more accurate, the matching result is scored in multiple dimensions, and high-quality talent information is provided for the user;

[0080] Reduce the complexity of interaction: the matching result of the talent information is counted, the data distribution of the talent information in different matching dimensions is analyzed, the weight of the matching dimension is set according to the data distribution of the different matching dimensions, the user operation is reduced, and the operation threshold is reduced; according to the weight, the detailed matching process is matched, and the matching result is optimized. BRIEF DESCRIPTION OF DRAWINGS

[0081] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments, made with reference to the accompanying drawings.

[0082] Figure 1 This is a schematic diagram of the method of the present invention;

[0083] Figure 2 This is a schematic diagram of the data processing flow of the present invention;

[0084] Figure 3 This is a diagram illustrating the talent information of the present invention. Detailed Implementation

[0085] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0086] Example 1

[0087] Please see Figure 1 A talent information-based intelligent matching method includes:

[0088] Step S1: Obtain talent information and the company's hiring needs;

[0089] Please see Figure 2 Step S2: Process the company's staffing needs information, obtain staffing needs standards, and divide the staffing needs standards into hard requirements and flexible requirements; based on the hard requirements, obtain the necessary items; based on the flexible requirements, obtain the bonus items.

[0090] Step S21: Obtain the company's staffing needs information; perform structured field transformation on the company's staffing needs information to obtain staffing needs standards, which are represented by standard dimensions and standard descriptions;

[0091] Step S211: Perform text preprocessing on the company's personnel demand information, extract the text of the personnel demand information, perform iterative judgment on the text, and delete the text when the text content contains irrelevant characters, format marks, or stop words to obtain clean text;

[0092] Step S2111: Remove emojis: Obtain the Unicode encoding bq1 of the first emoji in the computer, obtain the Unicode encoding bq2 of the last emoji in the computer, and obtain the encoding interval BQJ of the emoji according to the encoding range [bq1, bq2]; convert the user demand information into character encoding, count the number of character encodings bm, and denote the character encodings as ZF(1), ZF(2), ..., ZF(bm). If the character encodings ZF(1), ZF(2), ..., ZF(bm) ∈ BQJ, then replace the character encodings with null.

[0093] Specifically as follows:

[0094] An array cleaned_text is created according to the character encoding ZF(1), ZF(2), …, ZF(bm), and cleaned_text = [ZF(1), ZF(2), …, ZF(bm)]

[0095] The array is traversed:

[0096] for char in text:

[0097] When it is judged that the traversed element char belongs to an emoji, it is set to null;

[0098] if is_emoji(char):

[0099] cleaned_text.append("null")

[0100] else:

[0101] cleaned_text.append(char)

[0102] It should be noted that the emoji symbol has a specific range in Unicode encoding (such as U+1F600 to U+1F64F), and the encoding in this range is deleted to complete the deletion of the emoji symbol.

[0103] Step S2112: removing format markers: obtaining the type z of format markers, and setting the format markers as gs1, gs2, …, gsz; constructing a format list, filling the format list with the format markers gs1, gs2, …, gsz as list elements, and matching the human demand information with gs1, gs2, …, gsz according to the format list, deleting the content corresponding to gs1, gs2, …, gsz in the human demand information, and completing the cleaning of the format list;

[0104] Specifically as follows:

[0105] The format markers are binary coded, and the length r of the binary coding is obtained; the bit position of the coding is denoted as x, and the value at the coding bit position is denoted as y, a rectangular coordinate system is constructed, and the binary coding of the format markers is mapped in the rectangular coordinate system; a point set {(x1, y1), (x2, y2), …, (xr, yr)} of the format coding is obtained, and according to the point set {(x1, y1), (x2, y2), …, (xr, yr)}, the cases of y = 0 and y = 1 are connected, and the length is calculated:

[0106]

[0107] Wherein: cd0 represents the length when y = 0; y w , x w represents the wth x value and y value; the coordinate length of y value 0 is reserved by y w -1, and the length of y value 1 is discarded;

[0108]

[0109] Wherein: cd1 represents the length when y = 0, and the coordinate length of y value 1 is reserved by multiplying y w -1, and the length of y value 0 is discarded.

[0110] According to cd0 and cd1, the demand information is matched and deleted, and the format list is cleaned.

[0111] Step S2113: removing stop words: obtaining stop word category t, and recording stop words as TY1, TY2, …, TYt; taking stop words TY1, TY2, …, TYt as elements, constructing a stop word list, representing the stop word list through a regular expression, searching for stop words in the demand information, and deleting the searched stop words;

[0112] It should be noted that: stop words refer to words with no actual semantic value, high frequency of occurrence, but no help for analysis in natural language processing (NLP) or text analysis. In the tasks of keyword extraction, text classification, and information retrieval, deleting stop words can reduce noise and improve efficiency.

[0113] Step S212: cleaning the text, performing word segmentation processing through the jieba word segmentation tool, identifying key entities in the text by combining natural language processing technology, classifying the key entities, obtaining classification dimensions, recording the key entities under the same classification dimension as dimension content, and obtaining the demand standard according to the classification dimension and the dimension content: {standard dimension: standard description};

[0114] For example, the demand information of the enterprise is: “We need a software engineer with more than 3 years of Java development experience, familiar with Spring framework, good communication skills and team cooperation spirit.”

[0115] After structured conversion, the demand standard is obtained: {post: “software engineer”, skill: [“Java”, “Spring framework”], experience: “more than 3 years”, soft skill: [“communication skills”, “team cooperation spirit”]}.

[0116] Step S22: According to the human demand standard, the hard requirement in the human demand standard is extracted, the standard dimension and standard description corresponding to the hard requirement are obtained, they are integrated, the integrated standard dimension is recorded as necessary dimension, the integrated standard description is recorded as necessary description, and the necessary item is obtained; the necessary item is {necessary dimension: necessary description};

[0117] The flexible requirement in the human demand standard is extracted, the standard dimension and standard description corresponding to the flexible requirement are obtained, they are integrated, the integrated standard dimension is recorded as scoring dimension, the integrated standard description is recorded as scoring description, and the scoring item is obtained; the scoring item is {scoring dimension: scoring description};

[0118] It should be noted that: the hard requirement refers to the condition that must be strictly met and cannot be changed, and the flexible requirement refers to the requirement with certain flexibility and adjustability on the premise of meeting the basic demand or target, which allows appropriate changes and optimization according to actual situation;

[0119] Step S3: The talent information is processed, the talent information is filtered through the necessary item, and the optimized talent information is obtained; the optimized talent information is matched with the scoring item according to similarity, and the similarity value is obtained; the similarity value is sorted, and the initial score is assigned according to the sorting result;

[0120] Please refer to Figure 3 Step S31: The talent information is obtained, the necessary dimension and necessary description are obtained according to the necessary item, the talent information is matched and positioned according to the necessary dimension, the necessary description is matched with the talent information, the talent information meeting the necessary item is extracted, and the talent information not meeting the necessary item is deleted; the optimized talent information is obtained;

[0121] Step S311: The dimension number a of the necessary dimension is obtained, the necessary dimension is recorded as bwd1, bwd2, …, bwda; the necessary description corresponding to the necessary dimension item is recorded as bms1, bms2, …, bmsa; the talent information is matched and positioned according to the necessary dimension in turn; the necessary description is matched with the talent information according to the matching result;

[0122] Step S3111: The necessary dimension is taken as the judgment condition, the talent information is traversed, and the necessary dimension and the talent information are matched by character comparison; if the talent information does not exist for the description of the necessary dimension, the talent information is deleted;

[0123] Step S3112: If the talent information exists for the description of the necessary dimension, the corresponding content is extracted and matched with the necessary description, and the talent information is deleted according to the matching result;

[0124] Step S3113: According to steps S3111-S3112, the necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa are matched;

[0125] Step S312: If the necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa can obtain corresponding matching content in the talent information, the talent information is reserved; the reserved talent information is integrated to obtain optimized talent information;

[0126] Step S3121: An optimized talent information list is constructed, the necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa can obtain corresponding matching talent information in the talent information, and the necessary dimensions and the necessary descriptions in the talent information are deleted;

[0127] Step S3122: The necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa are XORed with the talent information that can be matched, the talent information is simplified, the data storage cost is reduced, and the simplified talent information is filled as a list element to the optimized talent information list;

[0128] Step S3123: The elements in the optimized talent information list are sorted; the data storage value of the simplified talent information is obtained, and the talent information is sorted according to the size of the data storage value;

[0129] It should be noted that the data storage value reflects the content richness of the talent information, and the talent information with rich content can better reflect whether it is suitable for the enterprise demand;

[0130] The number of elements ysl in the optimized talent information list is obtained, the elements in the optimized talent information list are recorded as ylb[ys], the first talent information in the optimized talent information list is taken as a comparison value bj, and the optimized talent information list is compared according to the comparison value;

[0131] If bj> ylb[ys], ylb[ys] is replaced with bj, ylb[ys+1] is compared, and ys=ysl;

[0132] If bj≤ ylb[ys], ylb[ys+1] is compared, and ys=ysl;

[0133] When the cycle is completed, the sorting of the optimization talent information is completed, and the sorting is in descending order of the talent information data storage value; an optimized talent information list is obtained after sorting, and the optimized talent information is called according to the optimized talent information list after sorting;

[0134] Step S32: According to the bonus item, the bonus dimension and the bonus description are obtained; the bonus dimension is matched with the optimization talent information, the similarity value is obtained by combining the bonus description, the similarity value of the same dimension is sorted according to the bonus dimension, and the initial score is assigned according to the sorting result;

[0135] Step S321: The number b of bonus dimensions is obtained, and the bonus dimensions are denoted as jwd1, jwd2, …, jwdb; the bonus description corresponding to the bonus dimension is denoted as jms1, jms2, …, jmsb; the bonus dimension is taken as a retrieval target, and the optimization talent information is retrieved; the retrieval content is obtained, and the retrieval content is matched with the bonus description to obtain a similarity value; the details are as follows:

[0136] Step S3211: The retrieval content is obtained, and the retrieval content is processed by word segmentation to obtain a retrieval word list; the number c of retrieval words is obtained, and the retrieval words are denoted as js1, js2, …, jsc; the bonus description is processed by word segmentation to obtain a description word list; the number d of description words is obtained, and the description words are denoted as ms1, ms2, …, msd;

[0137] Step S3212: The retrieval word js1 is matched with the description words ms1, ms2, …, msd respectively to obtain a semantic matching value, the maximum semantic matching value is obtained as the matching value xfz of the retrieval word; the matching value of the retrieval word is judged,

[0138] If xfz>0.5, it is judged that the retrieval word and the description word can be matched, and the matching value of the retrieval word is saved;

[0139] If xfz≤0.5, it is judged that the retrieval word and the description word cannot be matched, and the retrieval value is deleted;

[0140] Step S3213: According to step S3213, js2, …, jsc are matched with the description words ms1, ms2, …, msd, and the matching value of the retrieval word is obtained; the matching values of the retrieval words and the description words that can be matched are counted to obtain a matching value list;

[0141] Step S3214: The number e of matching values in the matching value list is obtained; the matching values are accumulated to obtain a similarity value xsz:

[0142]

[0143] wherein: xfz i represents the i-th matching value;

[0144] Step S322: According to steps S3211-S3214, the similarity values of the b bonus dimensions are obtained to obtain a similarity value list xsl, xsl=[xsz1, xsz2, …, xszb]; the number of optimization talent information f is obtained; and the similarity values of the b bonus dimensions are sorted in descending order according to the number of optimization talent information f and the similarity value list xsl; the sorting position px is counted to obtain a sorting list bxl, bxl=[px1, px2, …, pxb];

[0145] Step S323: According to the sorting list, the optimization talent information is initially scored; the sorting position is mapped to the interval [1, 2] for scoring by mapping, and the specific scoring is as follows:

[0146] The sorting position px, the number of optimization talent information f, and the mapping interval [1, 2] are obtained; the sorting position px is mapped and calculated according to the mapping interval [1, 2] to obtain the initial score cff;

[0147]

[0148] It should be noted that the initial score is limited by mapping the sorting position px to the interval [1, 2] to prevent the data from being too large;

[0149] For example, for the sorting position with a sorting of 1, the similarity value is the largest, and the initial score approaches 2; for the sorting position with a sorting of f, the similarity value is the smallest, and the initial score is 1; the initial score is limited to between 1 and 2 by mapping.

[0150] Step S4: According to the numerical distribution of the similarity values, the weight is set; the matching score model is constructed by the similarity values, the initial score, and the weight to score the talents;

[0151] Step S41: The similarity values of the b bonus dimensions are obtained, the variance of the similarity values of the b bonus dimensions is calculated, and the weight of the score calculation of the bonus dimension is set according to the variance of the similarity values, and the specific setting is as follows:

[0152] The number of optimization talent information f is obtained, and the similarity values of the b bonus dimensions are denoted as A(U, V) according to the number of optimization talent information f; wherein: A(U, V) represents the similarity value of the U-th bonus dimension and the V-th optimization talent information;

[0153] The similarity value mean xjz is calculated under the same dimension to obtain the similarity value mean xjz;

[0154]

[0155] wherein: xsz(U) represents the similarity value mean in the Uth dimension, A(U, V) represents the similarity value of the Uth bonus dimension and the Vth optimization talent information;

[0156] According to the similarity value mean xsz(U) and the similarity value A(U, V), the variance fc is calculated:

[0157]

[0158] wherein: fc(U) represents the variance in the Uth dimension;

[0159] The variance is set as the calculation weight qz of the bonus dimension, i.e. fc(U) = qz(U);

[0160] It should be noted that: variance is an important indicator in statistics for measuring the dispersion degree of a group of data, representing the average squared distance between data points and the mean. By calculating the variance, the size of the data fluctuation can be understood, thereby providing a basis for data analysis and model construction.

[0161] Step S42: Obtain the initial score cff(U), and combine the similarity value xszU and the weight qz(U) to construct a matching score model pmx:

[0162]

[0163] wherein: b is the number of bonus dimensions;

[0164] It should be noted that: the data is limited by the initial score, to prevent the data from being inaccurate due to some similarity values being 0, and to better express the difference by summing the similarity values xszU and scaling the score by the weight; the weight is obtained according to the variance, and a large variance indicates that the difference between the data is large and more representative; the weight is enlarged;

[0165] Step S43: Substitute the similarity value of the optimization talent information, the initial score and the weight into the matching score model to calculate the score of the optimization talent information, recommend the optimization talent information according to the score, and enter step S5;

[0166] Step S5: Sort the score results, and recommend talents to the enterprise according to the enterprise's demand for talents;

[0167] Step S51: Obtain the score results, sort the scores in descending order to obtain the score sorting; according to the enterprise's demand for talents, obtain the actual demand talent information quantity s of the enterprise and the recruitment expansion ratio h of the enterprise; calculate the recommended quantity tjs of the enterprise according to the actual demand talent information quantity s of the enterprise and the recruitment expansion ratio h of the enterprise;

[0168] tjs=s*(1+h);

[0169] It should be noted that: the demand ratio of the enterprise refers to that the enterprise needs 10 people, but in the actual interview or notification process, the personnel will be increased by a certain percentage, and this percentage is defined as the expansion ratio in the application.

[0170] Step S52: Obtain the number f of optimization talent information, compare the number f of optimization talent information with the recommended number tjs:

[0171] If f≤tjs, all optimization talent information is recommended;

[0172] If f>tjs, obtain the score ranking, and recommend the optimization talent information ranked from 1 to tjs.

[0173] The above formulas are dimensionless to calculate the numerical value, the formula is obtained by collecting a large amount of data to simulate the recent real situation, and the preset parameters in the formula are set by the person skilled in the art according to the actual situation, such as the weight coefficient and the proportion coefficient, the size of which is to obtain a specific numerical value by quantifying each parameter, which is convenient for subsequent comparison, and the size of the weight coefficient and the proportion coefficient only needs to not affect the proportional relationship between the parameters and the quantized numerical value.

[0174] Finally, it should be noted that: the above-described embodiments are only specific implementations of the present application, which are used to illustrate the technical solutions of the present application, but not to limit it, the protection scope of the present application is not limited to this, although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art in the technical range disclosed by the present application can still modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications, changes or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and all should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A talent information-based intelligent matching method, characterized in that, The matching method comprises: Step S1: obtaining talent information and enterprise's employment demand information; Step S2: processing the enterprise's employment demand information, obtaining employment demand standards, and dividing the employment demand standards into rigid requirements and flexible requirements; obtaining necessary items according to the rigid requirements, and obtaining bonus items according to the flexible requirements; Step S3: processing the talent information, screening the talent information through the necessary items to obtain optimized talent information, performing similarity matching on the optimized talent information and the bonus items to obtain similarity values, sorting the similarity values, and performing initial scoring according to the sorting results; Step S4: setting weights according to the numerical distribution of the similarity values; constructing a matching score model from the similarity values, the initial scores and the weights to score the talents; The specific steps of step S4 are as follows: Step S41: obtaining similarity values of b bonus dimensions, calculating the variance of the similarity values of the b bonus dimensions, and setting weights for the score calculation of the bonus dimensions according to the variance of the similarity values, specifically as follows: Obtaining the number f of optimized talent information, and recording the similarity values of the b bonus dimensions as A(U, V) according to the number f of the optimized talent information; Calculating the mean of the similarity values under the same dimension to obtain the mean xjz of the similarity values; ； Wherein: xjz(U) represents the mean of the similarity values under the Uth dimension, A(U, V) represents the similarity value of the Uth bonus dimension and the Vth optimized talent information; According to the mean xjz(U) of the similarity values and the similarity value A(U, V), the variance fc is calculated: ； Wherein: fc(U) represents the variance under the Uth dimension; Set the variance as the calculation weight qz of the bonus dimension, that is, fc(U) = qz(U); Step S42: obtaining the initial score cff(U), and constructing a matching score model pmx combining the similarity value xszU and the weight qz(U): ； Wherein: b is the number of bonus dimensions, and xszU represents the similarity value under the Uth dimension; Step S43: substituting the similarity value of the optimized talent information, the initial score and the weight into the matching score model to calculate the score of the optimized talent information, and recommending the optimized talent information according to the score; Step S5: sorting the score results, and recommending talents to the enterprise in combination with the enterprise's employment demand information.

2. The talent information intelligent matching method according to claim 1, characterized in that, The specific steps of step S2 are as follows: Step S21: obtaining the enterprise's employment demand information; converting the enterprise's employment demand information into a structural field to obtain employment demand standards, and representing the employment demand standards through standard dimensions and standard descriptions; Step S22: extracting the rigid requirements in the employment demand standards according to the employment demand standards, obtaining the standard dimensions and standard descriptions corresponding to the rigid requirements, integrating them, recording the integrated standard dimensions as necessary dimensions, recording the integrated standard descriptions as necessary descriptions, and obtaining necessary items; the necessary items are {necessary dimensions: necessary descriptions}; Extract the flexible requirements in the human demand standard, obtain the standard dimension and standard description corresponding to the flexible requirements, integrate them, mark the integrated standard dimension as a bonus dimension, mark the integrated standard description as a bonus description, and obtain a bonus item; the bonus item is {bonus dimension: bonus description}.

3. The talent information intelligent matching method according to claim 2, characterized in that, The specific steps of the step S21 are as follows: Step S211: text preprocessing is performed on the human demand information of the enterprise, text of the human demand information is extracted, and the text is cyclically judged; when the text content is irrelevant characters, format markers, and stop words, the text content is deleted to obtain cleaned text; Step S212: the cleaned text is subjected to word segmentation processing, key entities in the text are identified, the key entities are classified to obtain classification dimensions, key entities under the same classification dimension are marked as dimension content, and human demand standards are obtained according to the classification dimensions and the dimension content: {standard dimension: standard description}.

4. The talent information intelligent matching method according to claim 3, characterized in that, The specific steps of the step S211 are as follows: Remove the emoji: obtain the Unicode encoding bq1 of the first emoji in the computer, obtain the Unicode encoding bq2 of the last emoji in the computer, obtain the encoding interval BQJ of the emoji according to the encoding range [bq1, bq2] of the emoji; the human demand information is converted into character codes, the number of character codes bm is counted, the character codes are marked as ZF(1), ZF(2), …, ZF(bm), and if the character codes ZF(1) to ZF(bm) are in BQJ, the character codes are replaced with null; Remove the format markers: the types of the format markers z are obtained, the format markers are marked as gs1, gs2, …, gsz, a format list is constructed, the format markers gs1 to gsz are used as list elements, the format list is filled, the human demand information is matched with gs1 to gsz according to the format list, the contents corresponding to gs1 to gsz in the human demand information are deleted, and the cleaning of the format list is completed. Specifically as follows: The format markers are binary coded, the length r of the binary code is obtained, the bit position of the code is marked as x, and the value at the bit position of the code is marked as y, a rectangular coordinate system is constructed, and the binary code of the format markers is mapped in the rectangular coordinate system; a point set { (x1, y1), (x2, y2), …, (xr, yr)} of the format code is obtained, and the points with y value of 0 and y value of 1 are connected according to the point set { (x1, y1) to (xr, yr)}, and the connection length is calculated: ； wherein: cd0represents the length of the connection when the value of y is 0; x w and y w represents the wthx value and y value; ； Wherein: cd1 represents the connection length when y value is 1; According to cd0 and cd1, the human demand information is matched and deleted, and the cleaning of the format list is completed. Remove the stop words: obtain the types of the stop words t, mark the stop words as TY1, TY2, …, TYt, construct a stop word list by taking the stop words TY1 to TYt as elements, express the stop word list through a regular expression, search for the stop words in the human demand information, and delete the searched stop words.

5. The talent information intelligent matching method according to claim 1, characterized in that, The specific steps of the step S3 are as follows: Step S31: acquiring talent information, acquiring necessary dimensions and necessary descriptions according to necessary items, matching and positioning the talent information according to the necessary dimensions, matching the necessary descriptions with the talent information, extracting the talent information meeting the necessary items, and deleting the talent information not meeting the necessary items; obtaining optimized talent information; Step S32: acquiring bonus dimensions and bonus descriptions according to bonus items; matching the bonus dimensions with the optimized talent information, combining the bonus descriptions to calculate the similarity, obtaining the similarity value, sorting the similarity values of the same dimensions according to the bonus dimensions, and initially assigning scores according to the sorting results.

6. The talent information intelligent matching method according to claim 5, characterized in that, The specific steps of the step S31 are as follows: Step S311: acquiring the dimension number a of the necessary dimensions, recording the necessary dimensions as bwd1, bwd2, …, bwda, recording the necessary descriptions corresponding to the necessary item as bms1, bms2, …, bmsa, and sequentially matching and positioning the talent information according to the necessary dimensions; and matching the necessary descriptions with the talent information according to the matching results; Taking the necessary dimensions as the judgment condition, the talent information is traversed, and the necessary dimensions and the talent information are matched by character comparison; if the talent information does not exist for the description of the necessary dimension, the talent information is deleted; If the talent information exists for the description of the necessary dimension, the corresponding content is extracted and matched with the necessary description, and the talent information is deleted according to the matching result; The necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa are matched; Step S312: if the necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa can all obtain corresponding matching content in the talent information, the talent information is reserved; The reserved talent information is integrated to obtain the optimized talent information.

7. The talent information intelligent matching method according to claim 6, characterized in that, The specific steps of the step S312 are as follows: An optimized talent information list is constructed, the necessary dimensions bwd1, bwd2, …, bwda and the necessary descriptions bms1, bms2, …, bmsa are obtained, and the talent information that can be matched is obtained, the necessary dimensions and the necessary descriptions in the talent information are deleted; The necessary dimensions bwd1 to bwda and the necessary descriptions bms1 to bmsa are XORed with the talent information that can be matched, the talent information is simplified, and the talent information after simplification is used as a list element to fill the optimized talent information list; The elements in the optimized talent information list are sorted; The data storage value of the talent information after simplification is obtained, and the talent information is sorted according to the size of the data storage value; The number of elements ysl in the optimized talent information list is obtained, the elements ylb[ys] in the optimized talent information list are recorded, the first talent information in the optimized talent information list is taken as a comparison value bj, and the optimized talent information list is compared according to the comparison value; If bj> ylb[ys], ylb[ys] is replaced with bj, ylb[ys+1] is compared, and ys=ysl. If bj≤ ylb[ys], compare ylb[ys+1] until ys= ysl; When the loop is completed, the sorting of the optimization talent information is completed, and the sorting is a descending order sorting of the talent information data storage value; an optimization talent information list after sorting is obtained, and the optimization talent information is called according to the optimization talent information list after sorting.

8. The talent information intelligent matching method according to claim 5, characterized in that, The specific steps of the step S32 are as follows: Step S321: obtaining the number b of bonus dimensions, and recording the bonus dimensions as jwd1, jwd2,..., jwdb; recording the bonus descriptions corresponding to the bonus dimensions as jms1, jms2,..., jmsb; taking the bonus dimensions as a retrieval target, retrieving the optimization talent information; obtaining retrieval content, matching the retrieval content with the bonus descriptions, and obtaining a similarity value; Step S322: obtaining the similarity values of the b bonus dimensions to obtain a similarity value list xsl, xsl=[xsz1, xsz2,..., xszb]; obtaining the number f of optimization talent information; according to the number f of optimization talent information and the similarity value list xsl, the similarity values of the b bonus dimensions are sorted in descending order; the sorting position px is counted to obtain a sorting list bxl, bxl=[px1, px2,..., pxb]; Step S323: according to the sorting list, the optimization talent information is initially scored; by mapping the sorting position to the interval [1, 2] for scoring, the specific scoring is as follows: Obtaining the sorting position px, the number f of optimization talent information and the mapping interval [1, 2]; according to the mapping interval [1, 2], the sorting position px is mapped and calculated to obtain the initial score cff; 。 9. The talent information intelligent matching method according to claim 8, characterized in that, The specific steps of the step S321 are as follows: Obtaining the retrieval content, performing word segmentation processing on the retrieval content to obtain a retrieval word list; obtaining the number c of retrieval words, and recording the retrieval words as js1, js2,..., jsc; performing word segmentation processing on the bonus description to obtain a description word list; obtaining the number d of description words, and recording the description words as ms1, ms2,..., msd; Matching the retrieval word js1 with the description words ms1 to msd respectively to obtain a semantic matching value, obtaining the maximum semantic matching value as the coincidence value xfz of the retrieval word; judging the coincidence value of the retrieval word, If xfz>0.5, it is judged that the retrieval word and the description word can be matched, and the coincidence value of the retrieval word is saved; If xfz≤0.5, it is judged that the retrieval word and the description word cannot be matched, and the retrieval value is deleted; Similarly, js2 to jsc are matched with the description words ms1 to msd to obtain the coincidence values of the retrieval words; the coincidence values of the retrieval words and the description words that can be matched are counted to obtain a coincidence value list; Obtaining the number e of coincidence values in the coincidence value list; the coincidence values are accumulated to obtain the similarity value xsz: ； where: xfz i represents the ith matching value.

Images