A method for constructing a comprehensive recruitment difficulty model and a method for evaluating recruitment difficulty

The comprehensive recruitment difficulty model and evaluation method address the challenge of attracting high-quality talents by analyzing recruitment information in multiple dimensions, enhancing recruitment efficiency through targeted advertisement modifications.

CN114331304BActive Publication Date: 2024-12-03SHANGHAI JINYU INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202011056259.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-09-30
Publication Date
2024-12-03
Estimated Expiration
2040-09-30

AI Technical Summary

Technical Problem

Current technologies lack the ability to evaluate the attractiveness of recruitment advertisements and recruitment difficulty, hindering HR's ability to attract high-quality talents and affecting recruitment efficiency.

Method used

A comprehensive recruitment difficulty model and evaluation method that analyzes recruitment information in multiple dimensions, including enterprise scale, position salary, talent saturation, recruitment channel, and advertisement attractiveness, to provide a comprehensive recruitment difficulty score, guiding HR to modify advertisements for improved recruitment efficiency.

Benefits of technology

Enhances the ability of HR to attract high-quality talents by evaluating and modifying recruitment advertisements, thereby improving the efficiency of talent recruitment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114331304B_ABST
    Figure CN114331304B_ABST
Patent Text Reader

Abstract

The present invention relates to a construction method and device for a comprehensive recruitment difficulty model, a recruitment difficulty evaluation method and device, and a computer-readable storage medium. The construction method includes: obtaining recruitment information and candidate information; classifying and arranging the recruitment information by position to obtain multiple types of position information of each recruitment information, and storing the multiple types of position information of each recruitment information in a database by position; selecting multiple types of position information of multiple recruitment information from the database, and determining the analysis results of the selected multiple recruitment information in multiple dimensions respectively according to the multiple types of position information and candidate information of the corresponding positions stored in the database; marking the comprehensive recruitment difficulty of the selected multiple recruitment information respectively according to the analysis results of the multiple dimensions; and using the analysis results of the selected multiple recruitment information in multiple dimensions as input, and using the corresponding comprehensive recruitment difficulty scores as output, so as to train the comprehensive recruitment difficulty model.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to auxiliary technology for talent recruitment, and in particular to a method and device for constructing a comprehensive recruitment difficulty model, and a method and device for evaluating recruitment difficulty. Background Art

[0002] Talent is an important source for companies to gain sustainable competitive advantages, and the "talent war" between companies is becoming more and more fierce. How to attract outstanding talents to apply for job openings has become a major challenge for corporate recruitment. With the popularization of network technology, more and more companies choose to use online recruitment channels such as Zhaopin.com and 51Job to improve their recruitment efficiency and reduce recruitment costs. However, the situation where multiple companies publish recruitment advertisements through online recruitment channels at the same time has further formed a competitive relationship between recruiting companies.

[0003] For the HR (Human Resource) responsible for recruitment, posting recruitment advertisements is a profound science. If the recruitment advertisements published are not attractive, even if they are posted on a large number of recruitment websites at the same time, they will only receive a handful of suitable resumes, and it is often difficult to recruit excellent candidates. Therefore, how to design or modify your recruitment advertisements to attract the attention and resumes of high-quality talents to the greatest extent has become a major problem that plagues recruiting companies.

[0004] At present, there is no technology in this field that can help HR evaluate the attractiveness of recruitment advertisements and the difficulty of recruitment, and guide HR to modify recruitment advertisements, which is not conducive to recruiting companies to attract the attention of high-quality talents and submit resumes, and affects the efficiency of corporate recruitment. This field urgently needs an auxiliary technology for talent recruitment to make up for this lack. Summary of the invention

[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or decisive elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0006] As mentioned above, there is currently no technology in the art that can help human resources (HR) evaluate the attractiveness of recruitment advertisements and the difficulty of recruitment, and guide HR to modify recruitment advertisements, which is not conducive to recruiting companies to attract high-quality talents to pay attention to and submit resumes, and affects the efficiency of companies in recruiting talents. In order to make up for the lack of the above technology, the present invention provides a method and device for constructing a comprehensive recruitment difficulty model, a method and device for evaluating recruitment difficulty, and two corresponding computer-readable storage media, which are used to help HR evaluate the attractiveness of recruitment advertisements and the difficulty of recruitment, and guide HR to modify recruitment advertisements, thereby helping recruiting companies to attract high-quality talents to pay attention to and submit resumes, and improve the efficiency of companies in recruiting talents.

[0007] The method for constructing the above-mentioned comprehensive recruitment difficulty model provided by the present invention includes: obtaining recruitment information for at least one position of multiple recruiting companies from at least one recruitment channel, and candidate information of multiple candidates seeking employment in the at least one recruitment channel; classifying and arranging the recruitment information by position to obtain multiple types of position information of each recruitment information, and storing the multiple types of position information of each recruitment information in a database by position; selecting multiple types of position information of multiple recruitment information from the database, and determining the analysis results of the selected multiple recruitment information in multiple dimensions respectively according to the multiple types of position information of the corresponding position stored in the database and the candidate information; marking the comprehensive recruitment difficulty of the selected multiple recruitment information respectively according to the analysis results of the multiple dimensions; and taking the analysis results of the selected multiple recruitment information in multiple dimensions as input, and taking the corresponding comprehensive recruitment difficulty score as output, so as to train the comprehensive recruitment difficulty model. The comprehensive recruitment difficulty model can give a comprehensive recruitment difficulty score according to the multiple types of position information in the input recruitment information, so as to help HR evaluate the attractiveness and recruitment difficulty of recruitment advertisements, thereby helping recruiting companies to attract the attention of high-quality talents and submit resumes, and improve the efficiency of enterprise recruitment of talents.

[0008] The above-mentioned recruitment difficulty evaluation method provided by the present invention includes: classifying and arranging the recruitment information to be evaluated to obtain multiple types of position information of the recruitment information to be evaluated; determining the analysis results of the recruitment information to be evaluated in multiple dimensions according to the multiple types of position information and candidate information of the multiple recruitment information of the corresponding positions stored in the database; and inputting the analysis results of the recruitment information to be evaluated in multiple dimensions into the comprehensive recruitment difficulty model to obtain the corresponding comprehensive recruitment difficulty score. The evaluation method can use the above-mentioned comprehensive recruitment difficulty model to give the analysis results of the input recruitment advertisement in multiple dimensions and the comprehensive recruitment difficulty score, so as to help HR evaluate the attractiveness of the recruitment advertisement and the difficulty of recruitment, and guide HR to modify the recruitment advertisement according to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score, so as to help the recruiting company attract the attention of high-quality talents and submit resumes, and improve the efficiency of the company's recruitment of talents.

[0009] The construction device of the above-mentioned comprehensive recruitment difficulty model provided by the present invention includes a memory and a processor. The processor is connected to the memory and is configured to execute computer instructions stored in the memory to implement a construction method of a comprehensive recruitment difficulty model. The construction device can construct a comprehensive recruitment difficulty model. The comprehensive recruitment difficulty model can give a comprehensive recruitment difficulty score based on multiple types of job information in the input recruitment information, so as to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty, thereby helping recruiting companies attract the attention of high-quality talents and submit resumes, and improve the efficiency of corporate recruitment of talents.

[0010] The above-mentioned recruitment difficulty evaluation device provided by the present invention includes a memory and a processor. The processor is connected to the memory and is configured to execute computer instructions stored in the memory to implement a recruitment difficulty evaluation method. The evaluation device can use the above-mentioned comprehensive recruitment difficulty model to give the analysis results of the input recruitment advertisement in multiple dimensions and the comprehensive recruitment difficulty score, so as to help HR evaluate the attractiveness and recruitment difficulty of the recruitment advertisement, and guide HR to modify the recruitment advertisement according to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score, thereby helping the recruiting company to attract the attention of high-quality talents and submit resumes, and improve the efficiency of the company's recruitment of talents.

[0011] The present invention provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, a method for constructing a comprehensive recruitment difficulty model can be implemented. The comprehensive recruitment difficulty model constructed by the construction method can give a comprehensive recruitment difficulty score based on multiple types of job information in the input recruitment information, so as to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty, thereby helping recruiting companies attract the attention of high-quality talents and submit resumes, and improving the efficiency of companies in recruiting talents.

[0012] The present invention provides a computer-readable storage medium, on which computer instructions are stored. When the computer instructions are executed by a processor, a recruitment difficulty evaluation method can be implemented. The evaluation method can use the above-mentioned comprehensive recruitment difficulty model to give the analysis results of the input recruitment advertisement in multiple dimensions and the comprehensive recruitment difficulty score, so as to help HR evaluate the attractiveness and recruitment difficulty of the recruitment advertisement, and guide HR to modify the recruitment advertisement according to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score, thereby helping the recruiting company to attract the attention of high-quality talents and submit resumes, and improve the efficiency of the company's recruitment of talents. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0014] Figure 1 A flow chart of a method for constructing a comprehensive recruitment difficulty model provided according to some embodiments of the present invention is shown.

[0015] Figure 2 A schematic flow chart of a recruitment difficulty assessment method provided according to some embodiments of the present invention is shown.

[0016] Figure 3 A schematic diagram of the architecture of a device for constructing a comprehensive recruitment difficulty model provided according to some embodiments of the present invention is shown.

[0017] Figure 4 A schematic diagram of the architecture of a recruitment difficulty assessment device provided according to some embodiments of the present invention is shown. DETAILED DESCRIPTION

[0018] The following specific embodiments illustrate the implementation of the present invention, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Although the description of the present invention will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may be extended based on the claims of the present invention. In order to provide a deep understanding of the present invention, the following description will include many specific details. The present invention can also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present invention, some specific details will be omitted in the description.

[0019] As mentioned above, there is currently no technology in the art that can help human resources (HR) evaluate the attractiveness of recruitment advertisements and the difficulty of recruitment, and guide HR to modify recruitment advertisements, which is not conducive to recruiting companies to attract high-quality talents to pay attention to and submit resumes, and affects the efficiency of companies in recruiting talents. In order to make up for the lack of the above technology, the present invention provides a method and device for constructing a comprehensive recruitment difficulty model, a method and device for evaluating recruitment difficulty, and two corresponding computer-readable storage media, which are used to help HR evaluate the attractiveness of recruitment advertisements and the difficulty of recruitment, and guide HR to modify recruitment advertisements, thereby helping recruiting companies to attract high-quality talents to pay attention to and submit resumes, and improve the efficiency of companies in recruiting talents.

[0020] In some non-limiting embodiments, the method for constructing the comprehensive recruitment difficulty model can be implemented by a device for constructing a comprehensive recruitment difficulty model. Specifically, the device for constructing the comprehensive recruitment difficulty model can include a memory and a processor. The memory can include a computer-readable storage medium on which computer instructions are stored. The processor is connected to the memory and configured to execute the computer instructions stored in the memory to implement the method for constructing the comprehensive recruitment difficulty model.

[0021] The following will be described in conjunction with some embodiments of the construction method of the comprehensive recruitment difficulty model and the construction device of the comprehensive recruitment difficulty model. Those skilled in the art will understand that the embodiments of the construction method of the comprehensive recruitment difficulty model and the construction device of the comprehensive recruitment difficulty model are only some non-limiting embodiments provided by the present invention, which are intended to clearly demonstrate the concept of the present invention and provide some specific solutions that are convenient for the public to implement, rather than to limit the scope of protection of the present invention.

[0022] Please refer to Figure 1 , Figure 1 A flow chart of a method for constructing a comprehensive recruitment difficulty model provided according to some embodiments of the present invention is shown.

[0023] like Figure 1 As shown, in some embodiments of the present invention, the method for constructing the above-mentioned comprehensive recruitment difficulty model may include the steps of: obtaining recruitment information for at least one position from multiple recruiting companies from at least one recruitment channel, and candidate information of multiple candidates seeking jobs in at least one recruitment channel.

[0024] The above-mentioned recruitment channels include but are not limited to online recruitment platforms such as Zhaopin.com and 51job.com. Each recruitment platform can serve as a recruitment channel, on which recruitment advertisements for one or more positions of multiple recruiting companies can be recorded. In some embodiments, the recruitment advertisements can include recruitment information. The recruitment information includes but is not limited to the job title, work location, salary and benefits, job responsibilities, work content, the company size of the recruiting company, and one or more types of job information in the advertisement text.

[0025] In some embodiments, the operator of the construction device can cooperate with one or more of the above-mentioned recruitment platforms, so that the processor of the construction device can obtain the recruitment information of multiple recruiting companies for one or more positions, and the candidate information of multiple candidates seeking employment in various recruitment channels from the database of the one or more recruitment platforms. The candidate information includes but is not limited to the information of the recruitment channel where each candidate is located, and the information of the position applied for by the candidate. In some preferred embodiments, the processor of the construction device can determine the activity of each candidate based on the last login time of each candidate, and only obtain the candidate information of active candidates who have recently logged into the recruitment platform to improve the accuracy of the evaluation results.

[0026] Optionally, in other embodiments, the processor of the construction device may also utilize a web crawler program to automatically crawl programs or scripts of public web pages of various recruitment platforms according to preset rules to obtain recruitment information of various recruiting companies for various positions, as well as candidate information of various candidates seeking employment in various recruitment channels and other public information that can be legally obtained.

[0027] After obtaining the recruitment information of at least one position from multiple recruiting companies, the processor of the construction device can classify and organize the collected recruitment information and candidate information for subsequent statistics and analysis of the recruitment information and candidate information. Specifically, in some embodiments, the processor of the construction device can classify and organize the collected recruitment information according to job information categories such as work location, company size, position, salary, etc., and store multiple categories of position information of each recruitment information in a position information database by position. In some embodiments, the processor of the construction device can store the collected candidate information in a candidate database according to recruitment channel information and delivery position information by recruitment channel. In some preferred embodiments, the processor of the construction device can also process the collected recruitment advertisements to extract the advertising text therein, and store the extracted advertising text in a position advertisement database. The advertising text may further include the company's organizational characteristics (for example: company atmosphere, internal interpersonal relationships, company values), the company's learning and training system (for example: whether the training and learning mechanism is perfect, whether there is a mentor system), the company's compensation system (for example: whether the welfare system is complete), the company's career development path (for example: whether the future career development direction is clear, whether the promotion channels are transparent), the company's work characteristics (for example: work content, work intensity, work environment) and other information not covered in the preset position information categories.

[0028] like Figure 1 As shown, in some embodiments of the present invention, the method for constructing the above-mentioned comprehensive recruitment difficulty model may further include the step of: collecting statistics on multiple types of job information and candidate information stored in the database.

[0029] Specifically, in some embodiments, the processor of the construction device can count the multiple types of job information of each recruitment information stored in the job information database by job position to construct a competitor portrait for each job position. It can be understood that the competitors here refer to other recruiting companies that recruit the same job position on each recruitment platform. The competitor portrait is a data file that records multiple dimensional information of competitors and may include multiple measurement dimensions, wherein each measurement dimension may correspond to a type of job information for the job to be recruited. For a comprehensive recruitment difficulty model that supports analysis functions for multiple jobs, the processor of the construction device can construct multiple competitor portraits by job position, wherein each competitor portrait corresponds to one job position.

[0030] In some embodiments, the processor of the construction device can also count the candidate information of multiple active candidates by recruitment channel to construct candidate portraits for each recruitment channel. It is understandable that the active candidates here refer to job seekers who submit resumes through various recruitment platforms. The candidate portrait is a data file that records the number of people who submit for each position in the corresponding recruitment channel. For a comprehensive recruitment difficulty model that supports the analysis function of multiple recruitment channels, the processor of the construction device can construct multiple candidate portraits by recruitment channel, where each candidate portrait corresponds to a recruitment channel.

[0031] In some preferred embodiments, the processor of the construction device can also collect statistics on the recruitment channel information of each recruitment information stored in the position information database according to the position, so as to construct a recruitment channel portrait of each position. It can be understood that the recruitment channel portrait is a data file that records the recruitment quantity information of the corresponding position in various recruitment channels. In particular, in order to characterize the recent recruitment performance of various recruitment channels and predict their future recruitment performance, the recruitment channel portrait can further retain the recruitment quantity information at different time points for the processor to analyze the performance of different recruitment channels for recruitment positions.

[0032] In some embodiments of the present invention, the method for constructing the above-mentioned comprehensive recruitment difficulty model may also include the steps of: selecting multiple categories of position information of multiple recruitment information from a database, and determining the analysis results of the selected multiple recruitment information in multiple dimensions based on the multiple categories of position information and candidate information of the corresponding positions stored in the database.

[0033] like Figure 1 As shown, in some non-limiting embodiments, the processor of the construction device can first construct a recruitment analysis model based on the multiple categories of position information, competitor portraits and candidate portraits to be input, and then input the multiple categories of position information of the selected multiple recruitment information into the recruitment analysis model in sequence, so as to use the recruitment analysis model to respectively determine the analysis results of the selected multiple recruitment information in multiple dimensions.

[0034] In some embodiments, the above-mentioned recruitment analysis model can analyze the scale competitiveness of the recruiting company based on the work location information, company size information, and competitor portraits of the corresponding positions in the recruitment information to be input, and output the analysis results of the scale competitiveness. Specifically, the processor of the construction device can first select a stored recruitment information from the database as a training sample, and obtain its work location information and company size information from the selected recruitment information. Afterwards, the processor can screen the competitor portrait based on the obtained work location information to determine the total number of companies e recruiting for the corresponding position in the same city. According to the rules of the recruitment analysis model, the processor can compare the company size information of the training sample with the company size information of other recruiting companies recruiting for the corresponding position in the same city, so as to perform a secondary screening of the competitor portrait to determine the number of companies e0 whose scale is smaller than that of the sample company. Afterwards, the recruitment analysis model can calculate the scale percentile of the sample company in the city based on the total number of companies e and the number of companies e0. This is used as the analysis result of the selected recruitment information in the dimension of enterprise size.

[0035] In some embodiments, the above-mentioned recruitment analysis model can also analyze the salary competitiveness of the recruiting company based on the work location information, job salary information, and competitor portraits of the corresponding positions in the recruitment information to be input, and output the analysis results of the salary competitiveness. Specifically, the processor of the construction device can first obtain its work location information and job salary information from the above-mentioned training sample. Afterwards, the processor can filter the competitor portrait based on the obtained work location information to determine the total number of companies e recruiting for the corresponding position in the same city. According to the rules of the recruitment analysis model, the processor can compare the job salary information of the training sample with the job salary information of other recruiting companies recruiting for the corresponding position in the same city, so as to perform a secondary screening of the competitor portrait to determine the number of companies s0 whose job salary is lower than the job salary information. Afterwards, the recruitment analysis model can calculate the salary percentile of the sample company in the city based on the total number of companies e and the number of companies s0. This serves as the analysis result of the selected recruitment information in the position salary dimension.

[0036] In some preferred embodiments, the competitor portrait may include multiple measurement dimensions at multiple time points, and the candidate portrait may indicate the number of people who applied for each position through each recruitment channel at multiple time points. The recruitment analysis model may also analyze the current talent shortage based on the current number of recruitment positions and the historical number of recruitment positions in the competitor portrait, as well as the current number of job seekers and the historical number of job seekers in the candidate portrait, and output the analysis results of the current talent shortage. Specifically, the processor of the construction device may first obtain the current recruitment quantity p0 of the corresponding position from the competitor portrait based on the position indicated by the recruitment information, and obtain the current number of candidates c0 for the corresponding position from the candidate portrait. The recruitment analysis model may calculate the current job search ratio based on the current recruitment quantity p0 and the current number of candidates c0. When TSI>1, it means that the talent supply is insufficient to meet the demand. On the contrary, when TSI<1, it means that the talent supply exceeds the demand. Then, according to the rules of the recruitment analysis model, the processor can compare the current job search ratio TSI with each historical job search ratio data TSI (tsi1, tsi2, ..., tsi n ) to determine the number m of those that are lower than the current job seeker ratio TSI. The recruitment analysis model can determine the current talent shortage index based on the percentile of the current job seeker ratio TSI in the historical TSI ranking. This is used as the analysis result of the selected recruitment information in the talent saturation dimension. When x3=0, it means that the current TSI is at its lowest point in history, the recruitment market is in a state of talent supply exceeding demand, and talent is in short supply. On the contrary, when x3=1, it means that the current TSI is at its highest point in history, the recruitment market is in a state of talent supply exceeding demand, and talent is relatively saturated.

[0037] In some preferred embodiments, the above recruitment analysis model can also analyze the talent shortage degree of each current recruitment channel based on the recruitment channel portrait and the candidate portrait, and output the analysis result of the talent shortage degree of each current recruitment channel. Specifically, the processor of the construction device can first calculate the highest job search ratio TSI of the corresponding position in each current recruitment channel based on the recruitment channel portrait and the candidate portrait. h and the minimum job seeker ratio TSI l According to the rules of the recruitment analysis model, the processor can also filter the recruitment channel profile according to the recruitment channel information of the training sample to determine the current job search ratio TSI of the corresponding recruitment channel. t After that, the recruitment analysis model can be used to select the candidates with the highest job-seeking ratio TSI. h , the lowest job seeker ratio TSI l and current job seekers ratio TSI t , calculate the channel talent shortage index of the corresponding recruitment channel This serves as the analysis result of the selected recruitment information in the recruitment channel dimension.

[0038] In some preferred embodiments, the above recruitment analysis model can be constructed using a classification model or a regression model, and the attractiveness of the advertisement text in the above sample recruitment information is evaluated and analyzed according to a pre-trained three-point evaluation mapping function, and the analysis result of the attractiveness of the advertisement text is output. In some embodiments, the function of the recruitment analysis model to map the advertisement text linear layer to the attractiveness score can be realized by pre-using the advertisement texts of multiple existing recruitment information in the competitor portrait and the corresponding attractiveness annotation training.

[0039] Specifically, the processor of the construction device can first select multiple advertisement texts of recruitment information from the competitor portrait, and obtain the annotation of the attractiveness of each advertisement text from the outside. The annotation can be the average of the scores manually annotated by a large number of experts in the field of human resources based on the specific content of the advertisement text, or it can be the score manually annotated based on the number and quality of talents actually recruited by these existing advertisement texts. In some embodiments, the annotation can be implemented based on a three-point evaluation system. If the advertisement text is very attractive, it will be scored 2 points; if the advertisement text is mediocre, it will be scored 1 point; if the advertisement text has no highlights and is easy to discourage job seekers, it will be scored 0 points.

[0040] According to the rules of the recruitment analysis model, the processor can first abstractly represent the advertisement text used for training through the traditional bag-of-words model or encoding models such as CNN and RNN. Taking the encoding using the CNN model as an example, the processor can first represent the advertisement text used for training in units of words or phrases as Word = [w1, w2, ..., w n ], where w i In order to make the advertisement text suitable for convolution calculation, the processor can use word embedding matrix such as word2vec or glove to further convert the advertisement text into a two-dimensional matrix [e1, e2, ..., e k ], where k represents the dimension of the word vector matrix. It can be understood that the above word vector matrix can correspond each word or phrase in the advertisement text to a vector, and use multiple vectors to represent the corresponding words or phrases. The closer the cosine distance between the vectors, the more similar the semantics between the words or phrases. After obtaining the word vector matrix of the advertisement text, the processor can use a convolution kernel with a height of h With a window i:i+h-1 Perform convolution calculation, so that The dimension range is obtained Among them, e i:i-h-1 represents a window of size h×k consisting of the i-th row to the i+h-1th row of the input matrix, represented by e i ,ei+1 ,...,e i+h-1 concatenated, h represents the number of words in the window, w is the h×k-dimensional weight matrix (that is, the number of parameters that a convolution kernel needs to learn is h*k), b is the bias parameter, and f is the nonlinear function. Afterwards, the processor can process w and e i:i+h-1 Perform a dot product operation, sum the results of element-by-element multiplication, and then apply the convolution kernel to the ad text, moving one step at a time from top to bottom (i=1...n-h+1). 1:h After convolution operation, c1 is obtained. 2:h+1 After convolution operation, c2 is obtained. Similarly, by concatenating the convolution results, the feature f of the advertisement text after convolution operation can be obtained. ad =[c1,c2,...,c n Each convolution operation is equivalent to extracting a feature vector. By defining different windows, different feature vectors can be extracted to form the output of the convolution layer. After that, the processor can use a 1*d pooling vector e through the pooling operation. ad = pooling(f ad ) to represent the advertisement text. Taking maxpooling as an example, the processor can filter out a maximum feature from the feature vector generated by each sliding window, and then concatenate the filtered features to form a pooling vector e ad = pooling(f ad ).

[0041] In this way, the processor can sequentially perform the above-mentioned abstract representation, convolution calculation, and pooling operation on the selected multiple existing recruitment information advertisement texts to obtain the pooled vectors of multiple advertisement texts. Afterwards, the processor can use the pooled vectors of each advertisement text as input, and calculate the output score y by using regression or classification method through linear layer mapping. predict . The output score y is calculated by predict By calculating the loss with the correct annotated score output label y and adjusting the learning parameter w of the convolution calculation through the gradient descent method, the recruitment analysis model can be trained to realize the linear layer mapping function from the advertising text to the attractiveness score.

[0042] After completing the functional training of mapping the linear layer of the advertisement text to the attractiveness score, the processor can extract the advertisement text from the sample recruitment information and input the extracted advertisement text into the recruitment analysis model. The recruitment analysis model can perform abstract representation, convolution calculation, pooling operation and linear layer mapping on the extracted advertisement text to determine the advertisement attractiveness score x5 of the advertisement text of the sample recruitment information as the analysis result of the selected recruitment information in the advertisement attractiveness dimension.

[0043] In summary, in some preferred embodiments, the above recruitment analysis model can output analysis results X=[x1,x2,x3,x4,x5] in multiple dimensions, such as enterprise scale dimension, position salary dimension, talent saturation dimension, recruitment channel dimension and advertising attractiveness dimension, based on multiple types of position information to be input, competitor portraits, candidate portraits and recruitment channel portraits. By inputting multiple types of position information of multiple sample recruitment information stored in the database into the recruitment analysis model in sequence, the analysis results corresponding to each sample recruitment information can be obtained in sequence.

[0044] like Figure 1 As shown, in some embodiments of the present invention, the method for constructing the above-mentioned comprehensive recruitment difficulty model may further include the steps of: marking the comprehensive recruitment difficulty of the selected multiple recruitment information according to the analysis results of multiple dimensions; and taking the analysis results of the selected multiple recruitment information in multiple dimensions as input, and taking the corresponding comprehensive recruitment difficulty score as output, so as to train the comprehensive recruitment difficulty model.

[0045] In the above embodiment, the processor of the construction device can obtain the annotated score y of the comprehensive recruitment difficulty of each sample recruitment information from the outside. In some embodiments, the annotated score can be the average of the scores manually annotated by a large number of experts in the field of human resources based on the analysis results X = [x1, x2, x3, x4, x5] of the corresponding sample recruitment information in multiple dimensions such as enterprise scale dimension, position salary dimension, talent saturation dimension, recruitment channel dimension and advertising attractiveness dimension. In particular, the annotated score can be implemented based on a five-point evaluation system, corresponding to the competitiveness of the sample recruitment information in the above five dimensions. Optionally, in other embodiments, the annotated score can also be a score manually annotated based on the number and quality of talents actually recruited by the existing sample recruitment information.

[0046] In the comprehensive recruitment difficulty model, the processor can use a regression model or a classification model to implement the linear layer mapping function of the input multi-dimensional analysis results X=[x1,x2,x3,x4,x5] to the output comprehensive recruitment difficulty score y. For example, the comprehensive recruitment difficulty model can adopt a logistic regression model. The processor can first input X=[x1,x2,x3,x4,x5] into the logistic regression model to obtain z=w0+w1x1+w2x2+w3x3+w4x4+w5x5, where w0, w1, w2, w3, w4, and w5 are the learning parameters of the logistic regression model. Afterwards, the processor can use the Softmax function to model the probability of each score, which is in the form of The decision function of the comprehensive recruitment difficulty score is y * = argmax iP(y=i|x,w), the loss function is The processor can take the analysis results X = [x1, x2, x3, x4, x5] of the above-mentioned multiple sample recruitment information in multiple dimensions as input, and take the corresponding comprehensive recruitment difficulty score y as output, and modify each learning parameter w according to the loss function J(θ) i , in order to train the comprehensive recruitment difficulty model to realize the linear layer mapping function from the analysis results X = [x1, x2, x3, x4, x5] of multiple dimensions to the comprehensive recruitment difficulty score y.

[0047] In some embodiments, after completing the training of the comprehensive recruitment difficulty model, the processor of the construction device can also select another part of the recruitment information from the database, and obtain the comprehensive recruitment difficulty of these recruitment information from the outside for annotation. Afterwards, the processor of the construction device can input the analysis results X=[x1, x2, x3, x4, x5] of these recruitment information in multiple dimensions into the comprehensive recruitment difficulty model, and compare the comprehensive recruitment difficulty score y output by the comprehensive recruitment difficulty model with the obtained annotation score to verify whether the evaluation result of the comprehensive recruitment difficulty model is accurate.

[0048] Based on the above description, it can be known that by executing the construction method of the above-mentioned comprehensive recruitment difficulty model provided by the present invention, a comprehensive recruitment difficulty model can be constructed. The comprehensive recruitment difficulty model can give a comprehensive recruitment difficulty score based on multiple types of job information in the input recruitment information, so as to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty, thereby helping recruiting companies to attract the attention of high-quality talents and submit resumes, and improve the efficiency of corporate recruitment of talents.

[0049] In order to better demonstrate the technical application of the above-mentioned comprehensive recruitment difficulty model, the present invention also provides an embodiment of a recruitment difficulty evaluation method, which is used to illustrate how to use the above-mentioned comprehensive recruitment difficulty model to provide analysis results in multiple dimensions and comprehensive recruitment difficulty scores to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty; and how to guide HR to modify recruitment advertisements based on the analysis results in multiple dimensions and the comprehensive recruitment difficulty scores, thereby helping recruiting companies to attract the attention of high-quality talents and submit resumes, and improve the efficiency of companies in recruiting talents.

[0050] In some non-limiting embodiments, the above-mentioned recruitment difficulty assessment method can be implemented by a recruitment difficulty assessment device. Specifically, the recruitment difficulty assessment device can include a memory and a processor. The memory can include a computer-readable storage medium on which computer instructions are stored. The processor is connected to the memory and configured to execute the computer instructions stored in the memory to implement the above-mentioned recruitment difficulty assessment method.

[0051] The following will be described in conjunction with some embodiments of the recruitment difficulty assessment method and the recruitment difficulty assessment device. Those skilled in the art will understand that the embodiments of these recruitment difficulty assessment methods and recruitment difficulty assessment devices are only some non-limiting embodiments provided by the present invention, which are intended to clearly demonstrate the concept of the present invention and provide some specific solutions that are convenient for the public to implement, rather than to limit the scope of protection of the present invention.

[0052] Please refer to Figure 2 , Figure 2 A schematic flow chart of a recruitment difficulty assessment method provided according to some embodiments of the present invention is shown.

[0053] like Figure 2 As shown, in some embodiments of the present invention, the above-mentioned recruitment difficulty assessment method may include the steps of: classifying and arranging the recruitment information to be assessed to obtain multiple types of position information of the recruitment information to be assessed.

[0054] The above-mentioned recruitment information to be evaluated may be information in recruitment advertisements provided by client companies that have purchased recruitment difficulty assessment services, which may include job titles, work locations, salary and benefits, job responsibilities, work content, the company size of the client company, and one or more types of job information in the advertisement text.

[0055] The user of the recruitment difficulty assessment device can input the recruitment information to be assessed into the recruitment difficulty assessment device through the information interaction interface. Afterwards, the processor of the assessment device can classify and sort the input recruitment information according to the job location, enterprise size, job position, salary and other job information categories to obtain multiple types of job information such as recruitment channel information, job location information, job salary information, enterprise size information, and advertisement text.

[0056] like Figure 2 As shown, in some embodiments of the present invention, the above-mentioned recruitment difficulty assessment method may further include the step of: determining the analysis results of the recruitment information to be evaluated in multiple dimensions based on multiple categories of position information and candidate information of multiple recruitment information corresponding to the positions stored in the database.

[0057] In some non-limiting embodiments, the processor of the evaluation device may input multiple types of position information of the recruitment information to be evaluated into a recruitment analysis model, so as to use the recruitment analysis model to output analysis results of the recruitment information to be evaluated in multiple dimensions.

[0058] As described above, the recruitment analysis model can be constructed based on multiple types of job information to be input, competitor portraits, and candidate portraits. The competitor portrait can be constructed by counting multiple types of job information of multiple recruitment information stored in the database by job. The competitor portrait can include multiple measurement dimensions, each measurement dimension corresponding to a type of job information. The candidate portrait can indicate the number of people who applied for each job through each recruitment channel, and can be constructed by counting candidate information by recruitment channel. The candidate information can include recruitment channel information and application job information of multiple candidates.

[0059] Specifically, in some embodiments, the processor of the evaluation device may first obtain the work location information and enterprise size information of the recruitment information to be evaluated. Then, according to the rules of the recruitment analysis model, the processor may first screen the competitor portrait according to the obtained work location information to determine the total number of enterprises recruiting corresponding positions in the same city e; and then perform a second screening of the competitor portrait according to the obtained enterprise size information to determine the number of enterprises e0 whose enterprise size is smaller than the enterprise size information. Then, the recruitment analysis model may calculate and output the scale percentile of the recruitment information to be evaluated in the city according to the total number of enterprises e and the number of enterprises e0. This is used as the analysis result of the recruitment information in the dimension of enterprise size.

[0060] In some embodiments, the processor of the evaluation device can also obtain the work location information and position salary information from the recruitment information to be evaluated. According to the rules of the recruitment analysis model, the processor can first screen the competitor portrait based on the obtained work location information to determine the total number of companies recruiting corresponding positions in the same city e; then perform a second screening of the competitor portrait based on the obtained position salary information to determine the number of companies s0 whose position salaries are lower than the position salary information. Afterwards, the recruitment analysis model can calculate and output the salary percentile of the recruitment information to be evaluated in the city based on the total number of companies e and the number of companies s0. This is used as the analysis result of the recruitment information in the position salary dimension.

[0061] In some preferred embodiments, the competitor portrait may include multiple measurement dimensions at multiple time points, and the candidate portrait may indicate the number of people who applied for each position through each recruitment channel at multiple time points. The processor of the evaluation device may also obtain the current number of vacancies p0 for the corresponding position from the competitor portrait and the current number of candidates c0 for the corresponding position from the candidate portrait according to the position indicated by the recruitment information to be evaluated. The recruitment analysis model may calculate the current job search ratio based on the current number of vacancies p0 and the current number of candidates c0. Afterwards, according to the rules of the recruitment analysis model, the processor can compare the current job search ratio TSI with each historical job search ratio data TSI (tsi1, tsi2, ..., tsi n ) to determine the number m of those that are lower than the current job seeker ratio TSI. The recruitment analysis model can determine the current talent shortage index based on the percentile of the current job seeker ratio TSI in the historical TSI ranking. This is used as the analysis result of the selected recruitment information in the talent saturation dimension.

[0062] In some preferred embodiments, the processor of the evaluation device can also calculate the highest job seeker ratio TSI of the corresponding position in each current recruitment channel according to the recruitment channel profile and the candidate profile. h and the minimum job seeker ratio TSI l According to the rules of the recruitment analysis model, the processor can screen the recruitment channel profile according to the recruitment channel information in the recruitment information to be evaluated to determine the current job search ratio TSI of the corresponding recruitment channel. t After that, the recruitment analysis model can be used to select the candidates with the highest job-seeking ratio TSI. h , the lowest job seeker ratio TSI l and current job seekers ratio TSI t , calculate the channel talent shortage index of the corresponding recruitment channel This serves as the analysis result of the selected recruitment information in the recruitment channel dimension.

[0063] In some preferred embodiments, the processor of the evaluation device can also extract the advertisement text from the recruitment information to be evaluated, and input the extracted advertisement text into the recruitment analysis model. The recruitment analysis model can perform abstract representation, convolution calculation, pooling operation and linear layer mapping on the extracted advertisement text, and output the advertisement attractiveness score x5 of the extracted advertisement text. As described above, the function of the recruitment analysis model to map the advertisement text linear layer to the attractiveness score is trained by using the advertisement text of multiple recruitment information of competitor portraits and the labeled score of the attractiveness of each advertisement text, and the specific training process will not be repeated here.

[0064] In summary, in some preferred embodiments, the above recruitment analysis model can analyze the recruitment information to be evaluated in multiple dimensions such as enterprise scale dimension, position salary dimension, talent saturation dimension, recruitment channel dimension and advertising attractiveness dimension, and output the corresponding analysis result X client =[x1,x2,x3,x4,x5].

[0065] Afterwards, the processor of the evaluation device may convert the analysis result X into client=[x1,x2,x3,x4,x5] Input comprehensive recruitment difficulty model. Comprehensive recruitment difficulty model can use pre-trained mapping function to input analysis result X client =[x1,x2,x3,x4,x5] Output the corresponding comprehensive recruitment difficulty score y client As mentioned above, the function of mapping the output results of the recruitment analysis model into a comprehensive recruitment difficulty score is trained by using multiple types of job information and candidate information of multiple recruitment information stored in the database, as well as the annotated scores of the comprehensive recruitment difficulty of each recruitment information. The specific training process will not be repeated here.

[0066] like Figure 2 As shown, in some embodiments of the present invention, the above-mentioned recruitment difficulty assessment method may further include the step of generating a recruitment difficulty assessment report based on the analysis results of multiple dimensions and the comprehensive recruitment difficulty score.

[0067] In some embodiments, the database may store three types of assessment report templates: primary, intermediate, and advanced. Each assessment report template provides different levels of detail in the assessment information and modification suggestions. The assessment device may provide the client with the above three job description templates and their corresponding charging standards for the client to choose from.

[0068] If the client selects the primary assessment service, the processor of the assessment device may call the primary assessment report template. The primary assessment report template may include a column for the comprehensive recruitment difficulty evaluation of the client's job position, a column for the analysis of the size of other competitors recruiting the same job position, and a column for the competitiveness analysis of the job salary given by the client. client Fill in the comprehensive recruitment difficulty evaluation column to help HR evaluate the attractiveness of the recruitment advertisement and the difficulty of recruitment. The processor can fill in the size percentile x1 into the enterprise size analysis column, and call the corresponding suggestion text based on the score of the size percentile x1 to guide HR to modify the recruitment advertisement. The processor can also fill in the salary percentile x2 into the salary competitiveness analysis column, and call the corresponding suggestion text based on the score of the salary percentile x2 to guide HR to modify the recruitment advertisement.

[0069] For example, the value range of the salary percentile x2 is [0,1]. The processor can judge the salary competitiveness of the client's recruitment advertisement based on the preset custom parameters μ1 and μ2. When x2<μ1, the processor can judge that the salary competitiveness of the client's recruitment advertisement is weak, and can call the suggestion text "The upper limit of your position salary is too low. The current upper limit is 13,000, which is only higher than 25% of the recruiting companies. The salary competitiveness is weak. It is recommended to increase the upper limit of the salary" from the database. The 25% data can be determined based on the actual value of the salary percentile x2. When μ1<x2<μ2, the processor can judge that the salary competitiveness of the client's recruitment advertisement is moderate, and can call the suggestion text "The salary setting of your position is consistent with the salary of most companies. If you need to improve your competitiveness, you can appropriately increase the upper limit of the salary" from the database. When x2>μ2, the processor can judge that the salary competitiveness of the client's recruitment advertisement is strong, and can call the suggestion text "The salary setting of your position has exceeded that of most companies, and you have strong competitiveness in salary setting" from the database. In some embodiments, the above-mentioned custom parameters μ1 and μ2 can be configured as adjustable parameters, which can be flexibly set according to customer needs. Generally, the custom parameters μ1 and μ2 can be set equally.

[0070] If the customer selects the intermediate assessment service, the processor of the assessment device can call the intermediate assessment report template. The intermediate assessment report template can further include a talent saturation analysis column and a recruitment channel analysis column for the recruitment position. The processor can fill the current talent shortage index x3 into the talent saturation analysis column, and call the corresponding suggestion text according to the score of the current talent shortage index x3 to guide HR to adjust the salary range provided in the recruitment advertisement. The processor can fill the channel talent shortage index x4 of each recruitment platform into the corresponding position of the recruitment channel analysis column, and call the corresponding suggestion text according to the score of the channel talent shortage index x4 of each recruitment platform to guide HR to change the channel for publishing recruitment advertisements.

[0071] In some embodiments, the intermediate assessment report template may also include a statistical graph display method. Specifically, the processor may use time as the x-axis and TSI as the y-axis to draw a trend chart of the current talent shortage index x3 to indicate the future changes of the current talent shortage index x3. In addition, the processor may use time as the x-axis and TSI as the y-axis to draw a trend chart of the channel talent shortage index x4 for each recruitment channel to indicate the future changes of the channel talent shortage index x4.

[0072] If the customer selects the advanced evaluation service, the processor of the evaluation device may call the advanced evaluation report template. The advanced evaluation report template may further include an analysis column for the attractiveness of the advertisement text. The processor may fill the advertisement attractiveness score x5 into the advertisement text attractiveness analysis column, and call the corresponding suggestion text according to the score of the advertisement attractiveness score x5 to guide the HR to modify the content and wording of the advertisement text.

[0073] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0074] According to another aspect of the present invention, this article also provides a device for constructing a comprehensive recruitment difficulty model. Figure 3 , Figure 3 A schematic diagram of the architecture of a device for constructing a comprehensive recruitment difficulty model provided according to some embodiments of the present invention is shown.

[0075] like Figure 3 As shown, the construction device 30 of the above-mentioned comprehensive recruitment difficulty model provided by the present invention includes a memory 31 and a processor 32. The processor 32 is connected to the memory 31 and is configured to execute computer instructions stored in the memory 31 to implement a construction method of a comprehensive recruitment difficulty model. The construction device 30 can construct a comprehensive recruitment difficulty model. The comprehensive recruitment difficulty model can give a comprehensive recruitment difficulty score based on multiple types of job information in the input recruitment information, so as to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty, thereby helping recruiting companies attract high-quality talents to pay attention to and submit resumes, and improve the efficiency of companies in recruiting talents.

[0076] According to another aspect of the present invention, a computer-readable storage medium is also provided. The computer-readable storage medium stores computer instructions. When the computer instructions are executed by the processor 32, a method for constructing a comprehensive recruitment difficulty model can be implemented. The comprehensive recruitment difficulty model constructed by the construction method can give a comprehensive recruitment difficulty score based on multiple types of job information in the input recruitment information, so as to help HR evaluate the attractiveness of recruitment advertisements and recruitment difficulty, thereby helping recruiting companies attract high-quality talents to pay attention to and submit resumes, and improve the efficiency of corporate recruitment of talents.

[0077] According to another aspect of the present invention, a recruitment difficulty assessment device is also provided. Figure 4 , Figure 4 A schematic diagram of the architecture of a recruitment difficulty assessment device provided according to some embodiments of the present invention is shown.

[0078] like Figure 4 As shown, the above-mentioned recruitment difficulty evaluation device 40 provided by the present invention includes a memory 41 and a processor 42. The processor 42 is connected to the memory 41 and is configured to execute computer instructions stored in the memory 41 to implement a recruitment difficulty evaluation method. The evaluation device 40 can use the above-mentioned comprehensive recruitment difficulty model to give the analysis results of the input recruitment advertisement in multiple dimensions and the comprehensive recruitment difficulty score, so as to help HR evaluate the attractiveness of the recruitment advertisement and the recruitment difficulty, and guide HR to modify the recruitment advertisement according to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score, so as to help the recruiting company attract the attention of high-quality talents and submit resumes, and improve the efficiency of the company's recruitment of talents.

[0079] According to another aspect of the present invention, a computer-readable storage medium is also provided herein. Computer instructions are stored on the computer-readable storage medium. When the computer instructions are executed by the processor 42, a recruitment difficulty evaluation method can be implemented. The evaluation method can use the above-mentioned comprehensive recruitment difficulty model to give the analysis results of the input recruitment advertisement in multiple dimensions and the comprehensive recruitment difficulty score, so as to help HR evaluate the attractiveness and recruitment difficulty of the recruitment advertisement, and guide HR to modify the recruitment advertisement according to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score, thereby helping the recruiting company to attract the attention of high-quality talents and submit resumes, and improve the efficiency of the company's recruitment of talents.

[0080] Those skilled in the art will appreciate that information, signals, and data may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips cited throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or optical particles, or any combination thereof.

[0081] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. The technician may implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention.

[0082] Although the processors of each device described in the above embodiments can be implemented by a combination of software and hardware. However, it is understood that these processors can also be implemented in software or hardware. For hardware implementation, these processors can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, other electronic devices for performing the above functions, or a selection combination of the above devices. For software implementation, these processors can be implemented by independent software modules such as program modules (procedures) and function modules (functions) running on a general-purpose chip, each of which can perform one or more functions and operations described herein.

[0083] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a comprehensive recruitment difficulty model, characterized in that: include: Obtaining, from at least one recruitment channel, recruitment information of at least one position from multiple recruiting companies, and candidate information of multiple candidates seeking employment in the at least one recruitment channel, wherein the candidate information includes recruitment channel information and job application information of the candidate; Classifying and arranging the recruitment information by position to obtain multiple types of position information of each recruitment information, and storing the multiple types of position information of each recruitment information into a database by position, wherein the multiple types of position information include multiple ones of recruitment channel information, work location information, position salary information and enterprise scale information; Selecting multiple categories of job information from multiple recruitment information from the database, and performing statistics on the multiple categories of job information by job position to construct a competitor profile for each of the jobs, wherein the competitor profile includes multiple measurement dimensions, each measurement dimension corresponding to a category of the job information; The candidate information is counted by recruitment channel to construct a candidate profile for each recruitment channel, wherein the candidate profile indicates the number of people who applied for each position through each recruitment channel; Building a recruitment analysis model based on the multiple types of job information to be input, the competitor profiles and the candidate profiles; Inputting the multiple types of job information of the selected multiple recruitment information into the recruitment analysis model in sequence, so as to use the recruitment analysis model to respectively determine the analysis results of the selected multiple recruitment information in multiple dimensions, wherein the analysis results of the multiple dimensions include multiple dimensions of enterprise scale, job salary, talent saturation, recruitment channel, and advertising attractiveness; According to the analysis results of the multiple dimensions, respectively marking the comprehensive recruitment difficulty of the selected multiple recruitment information; and The analysis results of the selected multiple recruitment information in multiple dimensions are used as input, and the corresponding comprehensive recruitment difficulty scores are used as output to train the comprehensive recruitment difficulty model.

2. The construction method according to claim 1, characterized in that: The steps of determining the analysis results of the selected recruitment information in multiple dimensions include: Obtaining the work location information and enterprise size information from the selected recruitment information; The competitor profile is screened according to the work location information to determine the total number of companies recruiting for corresponding positions in the same city. ; Perform secondary screening on the competitor profile based on the enterprise scale information to determine the number of enterprises with a scale smaller than the enterprise scale information ;as well as According to the total number of enterprises and the number of enterprises mentioned , calculate the size percentile of the selected job posting in the city .

3. The construction method according to claim 1, characterized in that: The steps of obtaining the analysis results of the selected recruitment information in multiple dimensions include: Obtaining the work location information and position salary information from the selected recruitment information; The competitor profile is screened according to the work location information to determine the total number of companies recruiting for corresponding positions in the same city. ; Perform a secondary screening of the competitor profile based on the position salary information to determine the number of companies whose position salary is lower than the position salary information ;as well as According to the total number of enterprises and the number of enterprises mentioned , calculate the salary percentile of the selected job posting in the city .

4. The construction method according to claim 1, characterized in that: The competitor portrait includes multiple measurement dimensions at multiple time points, and the candidate portrait indicates the number of people who applied for each position through each recruitment channel at multiple time points. The step of obtaining the analysis results of the selected recruitment information in multiple dimensions includes: According to the recruitment information, the current recruitment quantity of the corresponding position is obtained from the competitor profile. ; Obtain the current number of candidates for the corresponding position from the candidate portrait according to the recruitment information ; According to the current recruitment number and the current number of candidates stated , calculate the current job search ratio ;as well as Calculate the current talent shortage index based on the ranking of the current job-seeking ratio in the job-seeking ratio at each time point .

5. The construction method according to claim 4, characterized in that: Also includes: The recruitment channel information of each of the recruitment information stored in the database is counted by position to construct a recruitment channel portrait of each of the positions, wherein the recruitment channel portrait indicates the number of recruitments for the corresponding position in each of the recruitment channels, The step of obtaining the analysis results of the selected recruitment information in multiple dimensions also includes: According to the recruitment channel profile and the candidate profile, the highest job-seeking ratio of the corresponding position in each of the current recruitment channels is calculated. and minimum job seeker ratio ; The recruitment channel profile is screened according to the recruitment channel information of the selected recruitment information to determine the current job search ratio of the corresponding recruitment channel. ;as well as According to the highest job-seeking ratio , the minimum job-seeking ratio and the current job search ratio , calculate the channel talent shortage index of the corresponding recruitment channel .

6. The construction method according to claim 1, characterized in that: The multiple types of job information also include advertisement texts, and the steps of constructing the recruitment analysis model include: Obtaining advertisement texts of multiple recruitment information from the competitor portrait, and marking the attractiveness of each advertisement text; Performing abstract representation, convolution calculation, and pooling operation on each of the advertisement texts respectively to obtain a pooled vector of each of the advertisement texts; and Taking the pooled vector of each of the advertisement texts as input and the corresponding annotation as output, the learning parameters of the convolution calculation are adjusted to train the recruitment analysis model to achieve a linear layer mapping of the advertisement text to the attractiveness score.

7. The construction method according to claim 6, characterized in that: The steps of obtaining the analysis results of the selected recruitment information in multiple dimensions include: Extracting the advertisement text from the selected job information; and The extracted advertisement text is input into the recruitment analysis model, and abstract representation, convolution calculation, pooling operation and linear layer mapping are performed to determine the advertisement attractiveness score of the extracted advertisement text. .

8. The construction method according to claim 1, characterized in that: The step of marking the comprehensive recruitment difficulty of the selected multiple recruitment information includes: The analysis results of multiple dimensions output by the recruitment analysis model , mark the comprehensive recruitment difficulty of the corresponding recruitment information; and Score each comprehensive recruitment difficulty The corresponding analysis results The information is stored in a database in an interrelated manner for use in training the comprehensive recruitment difficulty model.

9. A method for evaluating recruitment difficulty, characterized in that: include: Classifying and arranging the recruitment information to be evaluated to obtain multiple types of position information of the recruitment information to be evaluated, wherein the multiple types of position information include multiple ones of recruitment channel information, work location information, position salary information and enterprise scale information; Inputting multiple categories of job information of the recruitment information to be evaluated into a pre-trained recruitment analysis model, and the recruitment analysis model outputs analysis results of the recruitment information to be evaluated in multiple dimensions, wherein the recruitment analysis model is constructed based on the multiple categories of job information to be input, competitor portraits, and candidate portraits; the competitor portraits are constructed by statistically analyzing multiple categories of job information of multiple recruitment information stored in a database according to positions, and include multiple measurement dimensions, each of which corresponds to a category of the job information; the candidate portraits are constructed by statistically analyzing candidate information according to recruitment channels, and are used to indicate the number of people who have applied for each of the positions through each of the recruitment channels; and the candidate information includes the candidate's recruitment channel information and application position information; and The analysis results of the recruitment information to be evaluated in multiple dimensions are input into a pre-trained comprehensive recruitment difficulty model to obtain a corresponding comprehensive recruitment difficulty score, wherein the analysis results of the multiple dimensions include but are not limited to multiple dimensions of enterprise size, position salary, talent saturation, recruitment channel and advertising attractiveness, and the comprehensive recruitment difficulty model is constructed via the construction method of the comprehensive recruitment difficulty model according to any one of claims 1 to 8.

10. The evaluation method according to claim 9, characterized in that: The step of outputting the analysis results of the recruitment information to be evaluated in multiple dimensions further includes: Obtaining the work location information and enterprise size information from the recruitment information to be evaluated; The competitor profile is screened according to the work location information to determine the total number of companies recruiting for corresponding positions in the same city. ; Perform secondary screening on the competitor profile based on the enterprise scale information to determine the number of enterprises with a scale smaller than the enterprise scale information ;as well as According to the total number of enterprises and the number of enterprises mentioned , calculate the size percentile of the job posting to be evaluated in the city .

11. The evaluation method according to claim 9, characterized in that: The step of outputting the analysis results of the recruitment information to be evaluated in multiple dimensions further includes: Obtaining the work location information and position salary information of the recruitment information to be evaluated; The competitor profile is screened according to the work location information to determine the total number of companies recruiting for corresponding positions in the same city. ; Perform a secondary screening of the competitor profile based on the position salary information to determine the number of companies whose position salary is lower than the position salary information ;as well as According to the total number of enterprises and the number of enterprises mentioned , calculate the salary percentile of the job posting to be evaluated in the city .

12. The evaluation method according to claim 9, characterized in that: The competitor portrait includes multiple measurement dimensions at multiple time points, the candidate portrait indicates the number of people who applied for each position through each recruitment channel at multiple time points, and the step of outputting the analysis results of the recruitment information to be evaluated in multiple dimensions further includes: According to the recruitment information to be evaluated, the current recruitment quantity of the corresponding position is obtained from the competitor portrait. ; Obtain the current number of candidates for the corresponding position from the candidate portrait according to the recruitment information to be evaluated ; According to the current recruitment number and the number of current candidates stated , calculate the current job search ratio ;as well as Calculate the current talent shortage index based on the ranking of the current job-seeking ratio in the job-seeking ratio at each time point .

13. The evaluation method according to claim 12, characterized in that: The multiple types of job information stored in the database also include a recruitment channel portrait, which is constructed by statistically analyzing the recruitment channel information of the multiple job information stored in the database according to the job position, wherein the recruitment channel portrait indicates the number of recruitments for the corresponding job position in each of the recruitment channels, and the step of outputting the analysis results of the job information to be evaluated in multiple dimensions further includes: According to the recruitment channel profile and the candidate profile, the highest job-seeking ratio of the corresponding position in each of the current recruitment channels is calculated. and minimum job seeker ratio ; The recruitment channel profile is screened according to the recruitment channel information of the recruitment information to be evaluated to determine the current job search ratio of the corresponding recruitment channel. ;as well as According to the highest job-seeking ratio , the minimum job-seeking ratio and the current job search ratio , calculate the channel talent shortage index of the corresponding recruitment channel .

14. The evaluation method according to claim 9, characterized in that: The multiple types of job information also include advertisement texts, and the step of outputting the analysis results of the job information to be evaluated in multiple dimensions further includes: Extracting the advertisement text from the job information to be evaluated; and The extracted advertisement text is input into the recruitment analysis model, and abstract representation, convolution calculation, pooling operation and linear layer mapping are performed to determine the advertisement attractiveness score of the extracted advertisement text. , wherein the function of mapping the linear layer of the advertisement text to the attractiveness score is trained by using the advertisement texts of multiple recruitment information of the competitor portrait and the annotation of the attractiveness of each of the advertisement texts.

15. The evaluation method according to claim 9, characterized in that: The steps for obtaining the comprehensive recruitment difficulty score include: The analysis results of multiple dimensions output by the recruitment analysis model Input the comprehensive recruitment difficulty model, and the comprehensive recruitment difficulty model outputs the corresponding comprehensive recruitment difficulty score The function of mapping the output result of the recruitment analysis model into a comprehensive recruitment difficulty score is trained by utilizing multiple types of position information and candidate information of multiple recruitment information stored in the database, as well as the annotation of the comprehensive recruitment difficulty of each recruitment information.

16. The evaluation method according to claim 15, characterized in that: Also includes: According to the analysis results of the multiple dimensions and the comprehensive recruitment difficulty score Call the corresponding suggestion text; as well as The analysis results of the multiple dimensions 、The comprehensive recruitment difficulty score The suggested text is filled into a plurality of corresponding input slots in a preset assessment report template to generate an assessment report on recruitment difficulty.

17. A device for constructing a comprehensive recruitment difficulty model, characterized in that: It includes a memory and a processor, wherein the processor is connected to the memory and configured to execute computer instructions stored in the memory to implement the method for constructing a comprehensive recruitment difficulty model as described in any one of claims 1 to 8.

18. A recruitment difficulty assessment device, characterized in that: It comprises a memory and a processor, wherein the processor is connected to the memory and configured to execute computer instructions stored in the memory to implement the recruitment difficulty assessment method as described in any one of claims 9 to 16.

19. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by the processor, the method for constructing a comprehensive recruitment difficulty model as described in any one of claims 1 to 8 is implemented.

20. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by the processor, the recruitment difficulty assessment method as described in any one of claims 9 to 16 is implemented.