Intelligent Salary Inquiry Method and System

Abstract

This disclosure provides an intelligent salary query method and system. The method includes obtaining first job-seeking information from a job seeker, wherein the first job-seeking information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs; based on the first job-seeking information, searching for job information corresponding to the first job-seeking information through a first preset database and according to a preset job matching model, and determining a first matching degree between the job information and the first job-seeking information; if the first matching degree exceeds a first preset threshold, then displaying the first job information corresponding to the matching degree exceeding the first preset threshold. This job-seeking method analyzes a large number of salary samples, extracts and calculates the salary level of corresponding groups from massive amounts of data, provides relevant personnel with a clear cognitive reference for salary positioning, increases the success rate of job placement, and reduces unnecessary communication costs.

Description

Technical Field

[0001] This disclosure relates to the field of job search technology, and in particular to an intelligent salary query method and system. Background Technology

[0002] Traditional job salary reports rely on manually collecting salary data from various companies in the market and analyzing and calculating it using traditional offline Excel spreadsheets or tools. This method suffers from low accuracy and efficiency, and the human intervention makes the final analysis data less objective, reducing its credibility. When such data flows into the market or into the hands of clients, it can cause some damage to the salary environment of employees' jobs.

[0003] The information disclosed in the background section of this application is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] This disclosure provides an intelligent salary query method and system that can extract and calculate the salary level of a corresponding group from massive amounts of data by analyzing a large number of salary samples, providing relevant personnel with a clear cognitive reference for salary positioning, increasing the success rate of job placement and reducing unnecessary communication costs.

[0005] A first aspect of this disclosure provides an intelligent salary query method, including:

[0006] Obtain the job seeker's first job search information, wherein the first job search information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs;

[0007] Based on the first job application information, the job information corresponding to the first job application information is searched through the first preset database and according to the preset job matching model, and the first matching degree between the job information and the first job application information is determined.

[0008] If the first matching degree exceeds a first preset threshold, then the first job information corresponding to the matching degree exceeding the first preset threshold is displayed.

[0009] The first job information includes the reference salary corresponding to the target job.

[0010] In one alternative implementation,

[0011] After determining the first matching degree between the job information and the first job application information, the method further includes:

[0012] If the first matching degree is lower than the preset threshold, then based on the job seeker's first job-seeking information, a second preset database is used to search for second job-seeking information corresponding to the job seeker.

[0013] The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information.

[0014] Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information;

[0015] If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed.

[0016] The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

[0017] In one alternative implementation,

[0018] The method further includes training the job matching model, the training method comprising:

[0019] A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features.

[0020] The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data.

[0021] Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function.

[0022] Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature.

[0023] Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

[0024] In one alternative implementation,

[0025] The method further includes determining a reference salary, wherein the method for determining the reference salary includes:

[0026] Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database;

[0027] Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data.

[0028] Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores.

[0029] Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

[0030] In one alternative implementation,

[0031] The method also includes a training method for the weight allocation model, the training method for the weight allocation model comprising:

[0032] Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set;

[0033] Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified.

[0034] Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained;

[0035] By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified.

[0036] If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed;

[0037] If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

[0038] A second aspect of this disclosure provides an intelligent salary query system, comprising:

[0039] The first unit is used to obtain the job seeker's first job information, wherein the first job information includes the job seeker's target job name, the city where the target job is located, and the industry to which the target job belongs;

[0040] The second unit is used to search for job information corresponding to the first job information through a first preset database and a preset job matching model based on the first job information, and to determine the first matching degree between the job information and the first job information.

[0041] The third unit is used to display the first job information corresponding to the matching degree exceeding the first preset threshold if the first matching degree exceeds the first preset threshold.

[0042] The first job information includes the reference salary corresponding to the target job.

[0043] In one alternative implementation,

[0044] The device further includes a fourth unit, the fourth unit being used for:

[0045] If the first matching degree is lower than the preset threshold, then based on the job seeker's first job-seeking information, a second preset database is used to search for second job-seeking information corresponding to the job seeker.

[0046] The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information.

[0047] Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information;

[0048] If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed.

[0049] The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

[0050] In one alternative implementation,

[0051] The device further includes a fifth unit, the fifth unit being used for:

[0052] A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features.

[0053] The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data.

[0054] Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function.

[0055] Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature.

[0056] Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

[0057] In one alternative implementation,

[0058] The device further includes a sixth unit, the sixth unit being used for:

[0059] Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database;

[0060] Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data.

[0061] Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores.

[0062] Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

[0063] In one alternative implementation,

[0064] The sixth unit is also used for:

[0065] Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set;

[0066] Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified.

[0067] Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained;

[0068] By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified.

[0069] If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed;

[0070] If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

[0071] This disclosure provides an intelligent salary query method, including:

[0072] Obtain the job seeker's first job search information, wherein the first job search information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs;

[0073] By obtaining job seekers' initial job search information, it is possible to retrieve information from the database that matches their target profession. At the same time, the initial job search information only needs to include the job title, the city where the job is located, and the industry described in the job. Data matching can be completed with minimal information, which helps to improve the user experience for job seekers.

[0074] Based on the first job application information, the job information corresponding to the first job application information is searched through the first preset database according to the preset job matching model, and the first matching degree between the job information and the first job application information is determined.

[0075] This publicly disclosed job matching model can query job information corresponding to job seekers. By analyzing a large number of salary samples, it extracts and calculates the salary level of the corresponding group from massive amounts of data, providing relevant personnel with a clear cognitive reference for salary positioning, increasing the success rate of job placement and reducing unnecessary communication costs.

[0076] If the first matching degree exceeds a first preset threshold, then the first job information corresponding to the matching degree exceeding the first preset threshold is displayed.

[0077] The first job information includes the reference salary corresponding to the target job;

[0078] The first job information includes the reference salary for the target position. The reference salary is determined based on a comprehensive evaluation of data from multiple sources, taking into account the salary level in the same dimensions in the market and other person-related dimensions. This can provide an intuitive understanding of the job seeker's position and offer effective reference for job seekers. Attached Figure Description

[0079] Figure 1 This is a flowchart illustrating the intelligent salary query method according to an embodiment of the present disclosure;

[0080] Figure 2 This is a schematic diagram illustrating the correspondence between job information, weight values, and scores in an embodiment of this disclosure.

[0081] Figure 3 This is a schematic diagram of the structure of the intelligent salary query system according to an embodiment of the present disclosure. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0083] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in this disclosure, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such use of data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein.

[0084] It should be understood that in the various embodiments of this disclosure, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this disclosure.

[0085] It should be understood that in this disclosure, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0086] It should be understood that in this disclosure, "multiple" refers to two or more. "And / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, "and / or B" can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "Contains A, B, and C", "Contains A, B, and C" means that all three A, B, and C are contained; "Contains A, B, or C" means that one of A, B, and C is contained; "Contains A, B, and / or C" means that any one, two, or three of A, B, and C are contained.

[0087] It should be understood that in this disclosure, "B corresponding to A", "B corresponding to A", "A corresponds to B", or "B corresponds to A" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information. Matching A and B is defined as a similarity between A and B that is greater than or equal to a preset threshold.

[0088] Depending on the context, "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection."

[0089] The technical solutions of this disclosure will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0090] Figure 1 An exemplary flowchart of the intelligent salary query method according to an embodiment of this disclosure is shown, such as... Figure 1 As shown, the method includes:

[0091] Step S101: Obtain the job seeker's initial job search information;

[0092] The first job information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs;

[0093] For example, the categories and quantities of information included in the first job application information are merely illustrative. In actual application, they can be adjusted according to the actual situation. The first job application information may include other information besides the information listed above, such as the job seeker's age, education level, gender, etc. The above description of the first job application information does not constitute a limitation on the first job application information.

[0094] By obtaining job seekers' initial job search information, it is possible to retrieve information from the database that matches their target profession. At the same time, the initial job search information only needs to include the job title, the city where the job is located, and the industry described in the job. Data matching can be completed with minimal information, which helps to improve the user experience for job seekers.

[0095] Step S102: Based on the first job application information, search for job information corresponding to the first job application information through the first preset database and according to the preset job matching model, and determine the first matching degree between the job information and the first job application information.

[0096] For example, the first preset database of this disclosure stores a large amount of job-related data, including but not limited to job seeker gender, job seeker age, job seeker education, job seeker work experience, job seeker past occupation, job seeker city, job seeker industry, salary level of job seeker industry, job seeker user profile, job fixed-to-float ratio, and other information.

[0097] Based on the first job application information, a pre-defined job matching model can be used to search for job information corresponding to the first job application information in a first pre-defined database, and the first degree of matching between the job information and the first job application information can be determined.

[0098] For example, if the target job title of the first job search information is sales, the city of the target job is Beijing, and the industry of the target job is real estate, then job information that matches all three conditions can be searched in the first preset database, and the first matching degree can be calculated comprehensively.

[0099] In one alternative implementation,

[0100] The method further includes training the job matching model, the training method comprising:

[0101] A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features.

[0102] The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data.

[0103] For example, the training dataset for the salary dimension can be obtained by collecting daily operational data from third-party job search platforms or third-party job search operators, or by collecting customer data and market research data proactively collected by headhunters, as well as salary data and full data on the characteristics of the corresponding population, including information such as gender, age, and education level.

[0104] After constructing the training dataset for the salary dimension, each training data point can be labeled for subsequent feature recognition and filtering. Each training data point in the training dataset can also be vectorized to obtain standard training data. Vectorizing the training data not only reduces the amount of data but also reduces the difficulty of data processing.

[0105] It should be noted that the dataset constructed in the above way may have relatively coarse data dimensions. In order to reduce the pressure of subsequent data processing, statistical analysis methods can be used to identify possible error values ​​or outliers, such as bias analysis, identifying values ​​that do not follow the distribution or regression equation, and cleaning the dirty data in combination with business-specific rules.

[0106] At the same time, comparative analysis is used to find similar data to complete the incomplete data. For example, if the age field is missing in sample A, the same type of data is matched by the salary, city, industry and other known data of that data, and then sampled and compared to find the age with the highest proportion in this type of data and assigned to sample A.

[0107] Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function.

[0108] Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature.

[0109] For example, the job matching model of this disclosure can be built on a neural network to reduce the dimensionality of input features and obtain an efficient feature representation of the input features. It is understood that the function of the job matching model in the embodiments of this disclosure can include optimizing the input features, which can be built on a neural network. For example, the job matching model in the embodiments of this disclosure can include an autoencoder. It should be noted that the job matching model can include an autoencoder only for illustrative purposes, and the embodiments of this disclosure do not limit the specific type of job matching model.

[0110] An autoencoder is a type of neural network and an unsupervised learning algorithm. It can be trained using backpropagation to ensure that the output dimension of the autoencoder is the same as the input dimension. In practical applications, specific settings can be added to the autoencoder to learn valuable representations and information about the input features.

[0111] Optionally, the method for performing dimensionality-upgrading mapping on the standard training features may include:

[0112] Based on the first loss function, the first weight matrix corresponding to the first loss function, the first bias parameter corresponding to the first loss function, and the standard training features, the standard training features are subjected to dimensionality-up mapping processing. For example, the method for performing dimensionality-up mapping processing includes the following formula:

[0113]

[0114] Where H represents the feature after dimensionality upscaling of the standard training features, θ represents the first loss function, μ represents the first weight matrix, x represents the standard training feature, β represents the first bias parameter, and σ k Let k represent the standard deviation of each dimension of the standard training feature, k∈[1,D], where D represents the dimension of the standard training feature, and ∈ represents a non-zero constant.

[0115] Similarly, methods for performing dimensionality reduction mapping on the second intermediate feature may include:

[0116] Based on the second loss function, the second weight matrix corresponding to the second loss function, the second bias parameter corresponding to the second loss function, and the second intermediate feature, the second intermediate feature is subjected to dimensionality reduction mapping. For example, the method for performing dimensionality reduction mapping includes the following formula:

[0117]

[0118] Where H′ represents the feature after dimensionality reduction mapping of the second intermediate feature, θ′ represents the second loss function, μ′ represents the second weight matrix, x′ represents the second intermediate feature, β′ represents the second bias parameter, and σ ′k Let k represent the standard deviation of each dimension of the second intermediate feature, k∈[1,D], where D represents the dimension of the second intermediate feature and ∈ represents a non-zero constant.

[0119] Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

[0120] For example, the first and second loss functions of the job matching model to be trained can be iteratively optimized through backpropagation as shown in the following formula:

[0121]

[0122] in, μ represents the standard training features before dimensionality standardization. kLet represent the mean of each dimension in the standard training features, and x represent the standard training features. Indicates the third intermediate feature. This represents the loss function after iterative optimization.

[0123] This disclosure embodiment automatically learns the deep relationships between features to be optimized through the multi-layer network structure of the job matching model, which can uncover some hidden non-linear information features. For example, two seemingly unrelated accounts may actually belong to the same company, which is beneficial for further analysis of these two accounts, something that cannot be done manually.

[0124] In one alternative implementation,

[0125] After determining the first matching degree between the job information and the first job application information, the method further includes:

[0126] If the first matching degree is lower than the preset threshold, then based on the job seeker's first job-seeking information, a second preset database is used to search for second job-seeking information corresponding to the job seeker.

[0127] The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information.

[0128] Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information;

[0129] If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed.

[0130] The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

[0131] In practical applications, if there is limited job information, such as for less popular professions, or if the first job search information provided by the job seeker cannot effectively find matching jobs, a second preset database can be used to search for second job search information corresponding to the job seeker.

[0132] The second preset database may be access data of job seekers collected by the job search platform, such as job seekers' personal information, including but not limited to name, gender, address, etc.; job seekers' resume information, including but not limited to resumes actively submitted by job seekers or resumes generated through the job search platform; job seekers' historical search information, including but not limited to the occupations, industries, and cities searched by job seekers; and job seekers' historical employment information, including but not limited to the companies that job seekers have previously worked for.

[0133] The second preset database can effectively expand job search information related to job seekers, so that when the first matching degree is lower than the preset threshold, job search information can be further matched based on the second job search information.

[0134] Optionally, to obtain as much secondary job information as possible, you can drill down continuously by changing the level and granularity of the dimensions. From different dimensions such as job title, city, industry, and salary range, you can mine the group characteristics of the corresponding sample population. For example, when a user queries salary data for job A, the system calculates data across all dimensions by default. Users can freely select and combine data to query deeper and more granular data according to their query needs, such as Beijing → real estate industry → age 30 → 6 years of work experience → male → bachelor's degree for job A, etc.

[0135] This publicly disclosed job matching model can query job information corresponding to job seekers. By analyzing a large number of salary samples, it extracts and calculates the salary level of the corresponding group from massive amounts of data, providing relevant personnel with a clear cognitive reference for salary positioning, increasing the success rate of job placement and reducing unnecessary communication costs.

[0136] Step S103: If the first matching degree exceeds the first preset threshold, then display the first job information corresponding to the matching degree exceeding the first preset threshold.

[0137] The first job information includes the reference salary corresponding to the target job.

[0138] For example, the reference salary is determined based on a comprehensive evaluation of data from multiple sources. By providing the salary level in the same dimensions in the market and other person-related dimensions, it is possible to intuitively understand the job seeker's position and provide them with an effective reference.

[0139] In one alternative implementation,

[0140] The method further includes determining a reference salary, wherein the method for determining the reference salary includes:

[0141] Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database;

[0142] Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data.

[0143] Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores.

[0144] Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

[0145] For example, salary reference information may include job demand comparison, salary, gender, age, education, work experience, experience with well-known companies, and whether the candidate graduated from a prestigious university.

[0146] Among them, the job demand month-on-month comparison can be calculated by statistically analyzing the number of job postings in the market this month and last month, and then calculating (this month - last month) / last month * 100% to show the change in the demand for jobs in the market. This represents the change in the supply and demand of jobs in the market. Such changes will also affect the change in salary, thus enabling the acquisition of real-time dynamic salary information and providing more accurate salary information.

[0147] Fixed-to-variable ratio: The percentage of fixed salary to variable salary. The percentage of fixed salary has a significant impact on whether a job seeker is hired.

[0148] Salary distribution: Based on the actual sample situation, the sample salaries are divided into different categories, and the proportion of each category in the total sample is calculated.

[0149] For example, the weighting model can be a model built on a neural network, used to assign corresponding weight values ​​to each salary reference information.

[0150] The method for assigning a corresponding weight value to each sample data can be shown by the following formula:

[0151] P=(1-j)*array[i+1]+j*array[i+2]

[0152] Here, the array corresponding to each sample data can be sorted from smallest to largest, where n is the number of data in the data, the integer part is i, and the decimal part is j.

[0153] like Figure 2 As shown, Figure 2 An exemplary embodiment of this disclosure illustrates the correspondence between job information, weight values, and scores.

[0154] For example, company job postings can be compared with structured data in the system. Using job salary as the core and combining it with month-on-month changes in job market demand, scores can be assigned to seven characteristics of the job: salary, gender, age, education, work experience, experience at a reputable company, and whether the candidate graduated from a prestigious university. Figure 2 The simulated job's salary is 300,000. By comparing and calculating with sample data, it is found that 300,000 is in the 33rd percentile of the market, so the score is 33. Then, a weighting algorithm is used to assign different weight values ​​to each characteristic, and finally the comprehensive score of the company's job is obtained [33*30%+51.77*10%...], which provides a reference for companies to assess the competitiveness of their own positions.

[0155] In one alternative implementation,

[0156] The method also includes a training method for the weight allocation model, the training method for the weight allocation model comprising:

[0157] Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set;

[0158] Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified.

[0159] Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained;

[0160] By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified.

[0161] If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed;

[0162] If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

[0163] The first job information includes the reference salary for the target position. The reference salary is determined based on a comprehensive evaluation of data from multiple sources, taking into account the salary level in the same dimensions in the market and other person-related dimensions. This can intuitively help job seekers understand their position and provide them with effective references.

[0164] This disclosure provides an intelligent salary query method, including:

[0165] Obtain the job seeker's first job search information, wherein the first job search information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs;

[0166] By obtaining job seekers' initial job search information, it is possible to retrieve information from the database that matches their target profession. At the same time, the initial job search information only needs to include the job title, the city where the job is located, and the industry described in the job. Data matching can be completed with minimal information, which helps to improve the user experience for job seekers.

[0167] Based on the first job application information, the job information corresponding to the first job application information is searched through the first preset database according to the preset job matching model, and the first matching degree between the job information and the first job application information is determined.

[0168] This publicly disclosed job matching model can query job information corresponding to job seekers. By analyzing a large number of salary samples, it extracts and calculates the salary level of the corresponding group from massive amounts of data, providing relevant personnel with a clear cognitive reference for salary positioning, increasing the success rate of job placement and reducing unnecessary communication costs.

[0169] If the first matching degree exceeds a first preset threshold, then the first job information corresponding to the matching degree exceeding the first preset threshold is displayed.

[0170] The first job information includes the reference salary corresponding to the target job;

[0171] The first job information includes the reference salary for the target position. The reference salary is determined based on a comprehensive evaluation of data from multiple sources, taking into account the salary level in the same dimensions in the market and other person-related dimensions. This can intuitively help job seekers understand their position and provide them with effective references.

[0172] Figure 3 An exemplary schematic diagram of the intelligent salary query system according to an embodiment of this disclosure is shown, such as... Figure 3 As shown, the system includes:

[0173] The first unit 31 is used to obtain the job seeker's first job information, wherein the first job information includes the job seeker's target job name, the city where the target job is located, and the industry to which the target job belongs;

[0174] The second unit 32 is used to search for job information corresponding to the first job information through a first preset database and a preset job matching model based on the first job information, and to determine the first matching degree between the job information and the first job information.

[0175] The third unit 33 is used to display the first job information corresponding to the matching degree exceeding the first preset threshold if the first matching degree exceeds the first preset threshold.

[0176] The first job information includes the reference salary corresponding to the target job.

[0177] In one alternative implementation,

[0178] The device further includes a fourth unit, the fourth unit being used for:

[0179] If the first matching degree is lower than the preset threshold, then based on the job seeker's first job-seeking information, a second preset database is used to search for second job-seeking information corresponding to the job seeker.

[0180] The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information.

[0181] Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information;

[0182] If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed.

[0183] The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

[0184] In one alternative implementation,

[0185] The device further includes a fifth unit, the fifth unit being used for:

[0186] A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features.

[0187] The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data.

[0188] Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function.

[0189] Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature.

[0190] Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

[0191] In one alternative implementation,

[0192] The device further includes a sixth unit, the sixth unit being used for:

[0193] Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database;

[0194] Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data.

[0195] Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores.

[0196] Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

[0197] In one alternative implementation,

[0198] The sixth unit is also used for:

[0199] Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set;

[0200] Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified.

[0201] Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained;

[0202] By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified.

[0203] If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed;

[0204] If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

[0205] It should be noted that the beneficial effects of the intelligent salary query system in this embodiment can be referred to the beneficial effects of the aforementioned intelligent salary query method, and will not be repeated here.

[0206] This disclosure also provides a program product including executable instructions stored in a readable storage medium. At least one processor of the device can read the executable instructions from the readable storage medium, and the at least one processor executes the executable instructions to cause the device to implement the methods provided in the various embodiments described above.

[0207] The readable storage medium can be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of computer programs from one location to another. A computer storage medium can be any available medium accessible to a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application-Specific Integrated Circuit (ASIC). Alternatively, the ASIC can be located in a user equipment. Of course, the processor and the readable storage medium can also exist as discrete components in a communication device. The readable storage medium can be a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0208] In the above-described terminal or server embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in this disclosure can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0209] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit them. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A method for intelligent salary inquiry, characterized in that, include: Obtain the job seeker's first job search information, wherein the first job search information includes the job seeker's target job title, the city where the target job is located, and the industry to which the target job belongs; Based on the first job application information, the job information corresponding to the first job application information is searched through the first preset database and according to the preset job matching model, and the first matching degree between the job information and the first job application information is determined. If the first matching degree exceeds a first preset threshold, then the first job information corresponding to the matching degree exceeding the first preset threshold is displayed. The first job information includes the reference salary corresponding to the target job; If the first matching degree is lower than the first preset threshold, then based on the job seeker's first job information, a second job information corresponding to the job seeker is searched through a second preset database. The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information. Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information; If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed. The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

2. The method according to claim 1, characterized in that, The method further includes training the job matching model, the training method comprising: A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features. The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data. Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function. Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature. Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

3. The method according to claim 1, characterized in that, The method further includes determining a reference salary, wherein the method for determining the reference salary includes: Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database; Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data. Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores. Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

4. The method according to claim 3, characterized in that, The method also includes a training method for the weight allocation model, the training method for the weight allocation model comprising: Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set; Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified. Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained; By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified. If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed; If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

5. An intelligent salary inquiry system, characterized in that, include: The first unit is used to obtain the job seeker's first job information, wherein the first job information includes the job seeker's target job name, the city where the target job is located, and the industry to which the target job belongs; The second unit is used to search for job information corresponding to the first job information through a first preset database and a preset job matching model based on the first job information, and to determine the first matching degree between the job information and the first job information. The third unit is used to display the first job information corresponding to the matching degree exceeding the first preset threshold if the first matching degree exceeds the first preset threshold. The first job information includes the reference salary corresponding to the target job; The fourth unit is used to search for second job information corresponding to the job seeker through a second preset database based on the job seeker's first job information if the first matching degree is lower than the preset threshold. The second job information includes at least one of the job seeker's personal information, the job seeker's resume information, the job seeker's historical search information, and the job seeker's historical employment information. Based on the second job application information, and based on the job matching model, search for job information corresponding to the second job application information, and determine the second matching degree between the job information and the first job application information; If the second matching degree exceeds the second preset threshold, then the second job information corresponding to the second matching degree exceeding the second preset threshold is displayed. The second job information includes at least one of the following: the reference salary corresponding to the second job, the similarity between the second job and the target job, and the recommendation value of the second job.

6. The system according to claim 5, characterized in that, The system further includes a fifth unit, which is used for: A training dataset for salary dimensions is constructed, and each training data point in the dataset is labeled. Then, the features of each training data point are vectorized to obtain standard training features. The training dataset for the salary dimension includes at least one of the following: job application salary data, salary structure data, job application profile data corresponding to the salary data, and job requirement change data. Based on the first loss function of the job matching model to be trained, the standard training features are subjected to dimensionality-up mapping processing, and the standard training features are mapped to a second intermediate feature with the same spatial dimension as the mapping layer corresponding to the first loss function. Based on the second loss function of the job matching model to be trained, the second intermediate feature is subjected to dimensionality reduction mapping, and the second intermediate feature is mapped to a third intermediate feature with the same dimension as the standard training feature. Based on the feature error between the standard training features and the third intermediate features, the first loss function and the second loss function of the job matching model to be trained are iteratively optimized through the backpropagation algorithm so that the feature error meets the preset convergence condition, thereby completing the training of the job matching model.

7. The system according to claim 5, characterized in that, The system further includes a sixth unit, the sixth unit being used for: Based on predetermined salary reference information, corresponding sample data is assigned to each type of salary reference information according to the first preset database; Based on the sample data corresponding to each type of salary reference information and the preset weight allocation model, assign a corresponding weight value to each sample data. Based on the weight values ​​and the preset first correspondence, the salary score corresponding to each weight value is determined, wherein the salary score is used to indicate the salary proportion of the industry in which the sample data is located, and the first correspondence is used to indicate the correspondence between the weight values ​​and the salary scores. Based on the salary score and the first job information, a reference salary corresponding to the target job is determined according to a preset second correspondence, wherein the second correspondence is used to indicate the correspondence between the salary score and the reference salary.

8. The system according to claim 7, characterized in that, The sixth unit is also used for: Construct a salary data training set, and vectorize each salary data in the salary data training set into a feature vector to obtain a first feature vector set; Based on the salary level corresponding to the salary data, the first feature of the first feature vector set is classified. Based on the first feature after classification, the weight value corresponding to the first feature is determined by the loss function of the weight allocation model to be trained; By using a pre-constructed salary data test set, the third degree of matching between the weight value corresponding to the first feature and the reference weight value of the salary data test set is verified. If the third matching degree exceeds the third preset threshold, then the training of the weight allocation model is completed; If the third matching degree is lower than the third preset threshold, the loss function of the weight allocation model to be trained is iteratively optimized until the third matching degree exceeds the third preset threshold.

Images