Human resource sharing service platform based on big data

Abstract

The invention discloses a human resource sharing service platform based on big data, and relates to the technical field of human resource sharing service. Comprising a demand analysis and initial weight setting module, an information push and resource allocation module, a channel performance monitoring and data analysis module, a dynamic weight adjustment and optimization module and a resource reallocation and weight updating module. And the demand analysis and initial weight setting module comprehensively knows recruitment demands of the enterprise before the recruitment activity starts, and sets the initial push weight of each recruitment channel based on the recruitment demands. According to the method, the efficient channel is preferentially put by setting the initial push weight, the channel performance is monitored and analyzed in real time, the conversion effect is accurately predicted in combination with the machine learning model, the push weight is dynamically adjusted, the resources are inclined to the efficient channel, and the conversion rate and the recruitment quality are improved. Meanwhile, resource allocation is optimized, waste is avoided, cost is reduced, the period is shortened, competitive advantages of talents of enterprises are enhanced, and business growth and market competitiveness improvement are facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of human resource sharing services, and in particular to a human resource sharing service platform based on big data. Background Art

[0002] The human resources sharing service platform is a comprehensive service system built on digital and information technology. It aims to integrate the internal and external human resources management needs of enterprises, and provide enterprises and employees with efficient and convenient one-stop services through standardized, automated and intelligent processes. The platform usually covers modules such as recruitment management, salary and benefits, performance appraisal, employee training, labor relations and data analysis, and realizes the centralization and intelligence of human resources services through technologies such as cloud computing, big data and artificial intelligence. Its core goal is to optimize resource allocation, improve management efficiency, and reduce operating costs, while providing data support for corporate decision-making and providing employees with personalized service experience.

[0003] The prior art has the following deficiencies:

[0004] In multi-channel recruitment push, the existing human resource sharing service platform is usually unable to adjust the push resources in real time according to the actual conversion situation of the channel. This technical limitation leads to too fixed or average resource allocation, ignoring the difference in effect between channels, which leads to a series of serious consequences, especially in the highly competitive recruitment environment. When the actual conversion performance of efficient channels is excellent but the best time to reach the target candidates is missed due to insufficient resource investment, high-quality talents are often quickly snatched away by competitors. At the same time, although inefficient channels have higher resource investment, they waste resources due to poor results, significantly increase recruitment costs, and make it difficult to attract highly matched candidates, which directly weakens the return on investment and overall efficiency of recruitment.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a human resource sharing service platform based on big data, which ensures that recruitment resources are preferentially placed in the efficient channels that best meet the needs of the enterprise during the startup phase by reasonably setting the initial push weight, thereby maximizing information coverage and candidate reach. Through real-time monitoring and analysis of the performance indicators of each channel, timely identification of channels with excellent performance, and accurate prediction of conversion effects with the help of pre-trained machine learning models, dynamic adjustment of push weights, and more resources are tilted towards efficient channels, greatly improving conversion rates and recruitment quality. In addition, by dynamically optimizing resource allocation, the waste of resources in inefficient channels is effectively avoided, recruitment costs are significantly reduced and the return on investment is improved, which not only optimizes resource utilization efficiency and shortens the recruitment cycle, but also enhances the core advantages of enterprises in the fierce competition for talents, ensuring that high-quality talents can be quickly attracted and retained, and ultimately helping enterprises achieve business growth and continuous market competitiveness improvement, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above-mentioned purpose, the present invention provides the following technical solutions: a human resource sharing service platform based on big data, including a demand analysis and initial weight setting module, an information push and resource allocation module, a channel performance monitoring and data analysis module, a dynamic weight adjustment and optimization module, and a resource reallocation and weight update module;

[0008] Demand analysis and initial weight setting module: before the recruitment activity begins, first fully understand the company's recruitment needs, and based on the recruitment needs, set the initial push weights of various recruitment channels to ensure that resources can be allocated in an orderly manner to the channels that best meet the company's needs at the beginning;

[0009] The information push and resource allocation module, after completing the setting of the initial push weight, starts to push recruitment information on various channels according to the set weight ratio, so that each channel receives a specified number of recruitment advertisements or job information according to its assigned weight;

[0010] The channel performance monitoring and data analysis module continuously monitors the performance indicators of each channel during the recruitment information push period, extracts key data that can reflect the channel conversion effect, conducts in-depth analysis on the extracted data, and inputs the analysis results into the pre-trained machine learning model to predict the actual conversion of recruitment information;

[0011] Dynamic weight adjustment and optimization module, based on the prediction results of the machine learning model, dynamically adjusts the push weight of each recruitment channel, increases the push weight of channels with excellent prediction performance, optimizes resource allocation in real time, and prioritizes resources to efficient channels, thereby improving the overall recruitment effect;

[0012] The resource reallocation and weight update module, after completing the dynamic adjustment of push weights, reallocates the actual push weights of recruitment information, and readjusts the resource input of each channel based on the latest weight allocation strategy.

[0013] Preferably, the initial push weight refers to the initial resource ratio allocated to each channel at the beginning of the recruitment activity, based on the company's recruitment needs and the characteristics of each recruitment channel, reflecting the degree to which the company inclines resources to different channels when pushing recruitment information.

[0014] Preferably, during the push of recruitment information, the performance indicators of each channel are continuously monitored to extract key data that can reflect the channel conversion effect, wherein the extracted key data include the ratio of channel clicks converted to actual applications and the proportion of candidate interaction behaviors on the channel. Within the analysis window, after analyzing the ratio of the extracted channel clicks converted to actual applications and the proportion of candidate interaction behaviors on the channel, a click conversion factor and an interaction response factor are generated respectively. The actual effect of the click behavior is measured by the click conversion factor, and the activity of the channel is measured by the interaction response factor.

[0015] Preferably, after an in-depth analysis of the ratio of extracted channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel, the click conversion factor and interactive response factor generated by the analysis are input into a pre-trained machine learning model, and a push return index is output based on the model to predict the actual conversion of the recruitment information.

[0016] Preferably, within the analysis window, the specific steps of generating a click conversion factor after analyzing the ratio of the extracted channel click volume converted into actual applications are as follows:

[0017] For each channel, the total click volume of the channel and the actual number of applications are nonlinearly mapped according to the weight ratio to calculate the basic conversion contribution value of its click behavior to the actual application, reflecting the basic conversion effect of the channel click behavior on the actual application. The calculation expression is: ,in, is the basic conversion contribution value, A is the actual application volume, which indicates the number of job applications submitted through this channel, K is the channel click volume, which indicates the total number of times the recruitment information is clicked by candidates in this channel, L is the candidate's behavioral participation, and B is the channel characteristic coefficient, which reflects the advantages or disadvantages of the channel. The click conversion factor weight is used to adjust the mapping weight between the click volume and the actual application. is the candidate behavior factor weight, regulating the impact of candidate behavior engagement;

[0018] After calculating the basic conversion contribution value, the final click conversion factor is generated by combining the channel weight and resource allocation coefficient. The generation expression of the click conversion factor is: ,in, is the click conversion factor, which measures the actual conversion effect of click behavior; R is the channel resource utilization coefficient, which reflects the efficiency of the channel resource allocation; S is the resource saturation; is the resource saturation adjustment factor, which is used to smooth the impact of resource saturation on the click conversion factor. P is the number of high-quality candidates in the channel, which refers to the number of high-quality candidates converted from the channel. F is the total number of candidates, which refers to the number of all candidates converted from the channel. Adjust the weight for basic conversion and control the impact of basic conversion contribution value. The weight of high-quality candidates controls the contribution of the proportion of high-quality candidates to the click-through conversion factor.

[0019] Preferably, within the analysis window, the specific steps of generating the interaction response factor after analyzing the extracted candidate's interaction behavior ratio on the channel are as follows:

[0020] In the analysis window, the candidate's various interactive behaviors and their occurrence frequencies are recorded. Since the importance of each behavior is different, a weight is assigned to each interactive behavior, and the assigned weight is multiplied by the behavior frequency to obtain the weighted response contribution of each behavior. The calculation expression of the weighted response contribution is: ,in, is the weighted response contribution value of the i-th interactive behavior, is the total frequency of the candidate's ith interactive behavior within the analysis window. The square of the frequency is used to amplify the weight of high-frequency behaviors. is the weight of the i-th interactive behavior, reflecting the relative contribution of this behavior to the conversion. Introducing a logarithmic function to smooth the growth of high-frequency behavior;

[0021] The weighted response contributions of all interactive behaviors are nonlinearly superimposed to generate the total interactive response factor of the channel. In order to strengthen the identification of high-response channels, the exponential correction coefficient of the candidate's active coverage is added. The generation expression of the interactive response factor is: ,in, is the interactive response factor, n is the total number of candidate behavior categories, The nonlinear reinforcement factor that contributes to the behavior, is the active coverage rate of candidates in the channel, that is, the ratio of candidates participating in interactive behaviors to the total number of candidates in the channel. is the exponential correction coefficient of active coverage, which is used to adjust the impact of active coverage on the results. Represents the weighted response contribution value of the i-th interactive behavior by Raises the exponent to a power, increasing the influence of large weighted response contributions.

[0022] Preferably, based on the prediction results of the machine learning model, the push weights of each recruitment channel are dynamically adjusted. The specific steps are as follows:

[0023] During the push process, the key performance indicators of each channel are monitored in real time and input into the pre-trained machine learning model. The push return index generated by the model is used to evaluate the conversion potential of the channel.

[0024] Compare the push return index with the push return index reference threshold, calculate the channel return index deviation, and accurately evaluate the actual performance of each recruitment channel. The calculation expression of the return index deviation is: ,in, It is the reference threshold of the push return index, which is used to judge whether the channel performs well. is the push return index generated by the machine learning model prediction for channel k, is the return index deviation of channel k.

[0025] Preferably, according to the comparison result between the push return index and the push return index reference threshold, the push weight of the channel is dynamically adjusted, and the adjusted weight is calculated by the following formula: ,in: is the adjusted push weight of channel k, is the initial push weight of channel k, is the adjustment coefficient used to control the change range of push weight. Make sure to only weight positive deviations.

[0026] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0027] The present invention ensures that recruitment resources are preferentially placed on efficient channels that best meet the needs of the enterprise during the startup phase by reasonably setting the initial push weights, thereby maximizing information coverage and candidate reach. By real-time monitoring and analyzing the performance indicators of each channel, the channels with excellent performance are promptly identified, and the conversion effect is accurately predicted with the help of a pre-trained machine learning model, and the push weights are dynamically adjusted to tilt more resources toward efficient channels, greatly improving the conversion rate and recruitment quality. In addition, by dynamically optimizing resource allocation, the waste of resources in inefficient channels is effectively avoided, the recruitment cost is significantly reduced, and the return on investment is improved. It not only optimizes resource utilization efficiency and shortens the recruitment cycle, but also enhances the core advantages of enterprises in the fierce competition for talent, ensuring that high-quality talents can be quickly attracted and retained, and ultimately helping enterprises achieve business growth and continuous improvement in market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0029] Figure 1 This is a module schematic diagram of a big data-based human resources sharing service platform of the present invention. DETAILED DESCRIPTION

[0030] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0031] The present invention provides Figure 1 A human resource sharing service platform based on big data is shown, including a demand analysis and initial weight setting module, an information push and resource allocation module, a channel performance monitoring and data analysis module, a dynamic weight adjustment and optimization module, and a resource reallocation and weight update module;

[0032] Demand analysis and initial weight setting module: before the recruitment activity begins, first fully understand the company's recruitment needs, and based on the recruitment needs, set the initial push weights of various recruitment channels to ensure that resources can be allocated in an orderly manner to the channels that best meet the company's needs at the beginning;

[0033] Initial push weight refers to the initial resource ratio allocated to each channel at the beginning of the recruitment activity, based on the company's recruitment needs and the characteristics of each recruitment channel. It reflects the degree to which the company's resources are tilted towards different channels when pushing recruitment information. The setting of initial push weight is usually based on historical data (such as click-through rate, conversion rate, candidate quality score of each channel), target candidate profile of the recruitment position (such as skills, region, industry background, etc.), channel coverage, and budget allocation strategy. For example, for technical positions, a higher weight can be set for professional communities or technical forums (such as GitHub, StackOverflow), while for sales positions, resources may be allocated to recruitment websites or social media first. By setting the initial push weight reasonably, it is ensured that resources can be concentrated on the channels that are most likely to bring high conversion rates in the early stage of recruitment, laying the foundation for subsequent dynamic adjustments.

[0034] The information push and resource allocation module, after completing the setting of the initial push weight, starts to push recruitment information on various channels according to the set weight ratio, so that each channel receives a specified number of recruitment advertisements or job information according to its assigned weight;

[0035] For example, if a recruitment website has a higher initial weight, the website will get more job exposure opportunities. At the same time, ensure that the format and content of the pushed content conform to the characteristics of each channel and user habits to improve the acceptance and response rate of the information. The real-time push process needs to be seamlessly connected with the interfaces of each channel to ensure that the information can be delivered to the target candidate group in a timely and accurate manner. This step ensures that recruitment information can be widely and targetedly distributed on multiple channels according to corporate needs, improving the coverage and reach of information. At the same time, by allocating initial weights, ensure that resources are reasonably utilized in the early stage of push, and provide preliminary data support for subsequent performance evaluation and dynamic adjustment.

[0036] The channel performance monitoring and data analysis module continuously monitors the performance indicators of each channel during the recruitment information push period, extracts key data that can reflect the channel conversion effect, conducts in-depth analysis on the extracted data, and inputs the analysis results into the pre-trained machine learning model to predict the actual conversion of recruitment information;

[0037] During the push of recruitment information, the performance indicators of various channels are continuously monitored to extract key data that can reflect the channel conversion effect. The extracted key data include the ratio of channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel. Within the analysis window, after analyzing the ratio of the extracted channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel, the click conversion factor and interactive response factor are generated respectively. The actual effect of the click behavior is measured by the click conversion factor, and the activity of the channel is measured by the interactive response factor.

[0038] During the recruitment information push period, if the conversion rate of clicks on a certain channel to actual applications is high, it usually indicates that the actual conversion of recruitment information on this channel is good. A high click-to-conversion rate means that after browsing the recruitment information, the candidate has sufficient interest in the position and has taken the next step (such as submitting an application or resume). This reflects that the recruitment information has a high matching degree in this channel, and it may also indicate that the target candidates of this channel are highly consistent with the job requirements. In addition, a high click-to-conversion rate also indicates that the content design of the recruitment information (such as job description, salary and benefits display, corporate brand promotion, etc.) is attractive and can effectively stimulate candidates' job-seeking intentions. Therefore, the click-to-conversion rate is one of the important indicators to measure the actual conversion effect of recruitment information, which can be used to guide resource allocation and optimize recruitment strategies.

[0039] In the analysis window, the specific steps for generating the click conversion factor after analyzing the ratio of extracted channel clicks to actual applications are as follows:

[0040] For each channel, the total click volume of the channel and the actual number of applications are nonlinearly mapped according to the weight ratio to calculate the basic conversion contribution value of its click behavior to the actual application, reflecting the basic conversion effect of the channel click behavior on the actual application. The calculation expression is: ,in, is the basic conversion contribution value, A is the actual application volume, which indicates the number of job applications submitted through this channel, K is the channel click volume, which indicates the total number of times the recruitment information is clicked by candidates in this channel, L is the candidate's behavioral participation, such as the comprehensive score of behavioral indicators such as page dwell time and secondary clicks, B is the channel characteristic coefficient, which reflects the advantages or disadvantages of the channel (such as recommendation accuracy and user activity), The click conversion factor weight is used to adjust the mapping weight between the click volume and the actual application. is the candidate behavior factor weight, regulating the impact of candidate behavior engagement;

[0041] Candidate behavioral engagement is a key indicator that is gradually being applied to recruitment analysis in existing technologies. It evaluates candidates' interest and engagement in recruitment information by monitoring a series of interactive behaviors on the recruitment information page. Specific behaviors include page dwell time, number of clicks, browsing depth (such as viewing company profile, salary and benefits, etc.), secondary visit frequency, application intention (such as resume upload or job collection), etc. Through data analysis, these behaviors are quantified into a comprehensive score to reflect the candidate's potential interest in the position and the depth of engagement. At present, this technology has been widely used by some intelligent recruitment platforms, combined with big data and machine learning, to optimize the content and push strategy of recruitment information and improve candidate conversion rate. It not only provides a more accurate basis for recruitment effect evaluation, but also helps companies reach target candidates more targeted.

[0042] The channel characteristic coefficient is a comprehensive parameter used to quantify the unique advantages or disadvantages of a recruitment channel, reflecting the inherent characteristics of the channel in attracting target candidates. This coefficient takes into account factors such as the channel's recommendation accuracy, user activity, historical conversion effect, matching degree of covered candidate profiles, and cost-effectiveness. For example, professional technical communities (such as GitHub) may show a higher matching degree in attracting developers, so their channel characteristic coefficient will be higher; while traditional recruitment websites may have a lower coefficient due to their wide target group but low accuracy. Setting the channel characteristic coefficient requires a comprehensive evaluation of the channel's performance based on historical data analysis and industry experience through machine learning models or multi-indicator weighted methods. It is a coefficient in the existing technology and will not be elaborated on here. This coefficient plays an important role in resource allocation optimization, helping the platform to reasonably allocate resources according to channel characteristics and achieve more efficient recruitment results.

[0043] This step uses nonlinear mapping (logarithmic function and sigmoid function) to make a more detailed trade-off between the relationship between click volume and actual applications, avoiding analysis bias caused by direct linear correlation.

[0044] After calculating the basic conversion contribution value, the final click conversion factor is generated by combining the channel weight and resource allocation coefficient. The generation expression of the click conversion factor is: ,in, is the click conversion factor, which measures the actual conversion effect of click behavior; R is the channel resource utilization coefficient, which reflects the efficiency of the channel resource allocation; S is the resource saturation, which measures whether the channel resource usage is close to the upper limit (such as the amount of advertising delivery); is the resource saturation adjustment factor, which is used to smooth the impact of resource saturation on the click conversion factor. P is the number of high-quality candidates in the channel, which refers to the number of high-quality candidates converted from the channel. F is the total number of candidates, which refers to the number of all candidates converted from the channel. Adjust the weight for basic conversion and control the impact of basic conversion contribution value. The weight of high-quality candidates controls the contribution of the proportion of high-quality candidates to the click-through conversion factor.

[0045] The channel resource utilization coefficient is a parameter that measures the effectiveness or efficiency achieved by a recruitment channel when using allocated resources. It reflects the proportional relationship between the resources allocated to the channel (such as budget, number of exposures, push frequency, etc.) and the actual recruitment conversion effect it brings (such as the number of resumes submitted and the quality of candidates). The higher the resource utilization coefficient, the more efficient the channel is in using resources in recruitment activities, that is, each unit of allocated resources can bring more actual results.

[0046] When calculating the resource utilization coefficient, the following formula can be used:

[0047] ,in: The actual conversion effect generated by the channel (such as the actual number of resumes submitted, number of clicks, etc.). The amount of resources allocated to the channel (such as budget, number of pushes, etc.).

[0048] The resource utilization coefficient can help recruitment platforms dynamically evaluate the resource utilization efficiency of each channel and provide data support for resource allocation decisions. Channels with high utilization coefficients indicate that resources are effectively utilized and should be further invested; while channels with low utilization coefficients may need to optimize or reduce allocated resources to avoid waste.

[0049] Resource saturation is a parameter that measures whether a recruitment channel is close to its actual carrying capacity during the resource allocation process. It reflects the relationship between the channel resource usage and its maximum carrying capacity. When resource allocation is close to the saturation point, even if you continue to increase investment, you may not be able to significantly improve the conversion effect, and may even lead to diminishing marginal benefits.

[0050] Resource saturation can be calculated using the following formula:

[0051] ,in: The amount of resources that have been allocated. The maximum amount of resources that the channel can effectively carry (can be estimated through historical data or channel characteristics analysis).

[0052] Resource saturation can help platforms avoid over-investment in channel resources. When the resource saturation of a channel is close to 1, it should consider stopping adding resources or transferring resources to other efficient channels. This parameter can significantly improve the scientificity and economy of resource allocation.

[0053] The number of high-quality candidates in a channel refers to the number of highly matched candidates obtained through a certain recruitment channel. High-quality candidates usually refer to candidates whose resume, skills, experience and other conditions are highly consistent with the requirements of the recruitment position and have a high probability of being hired in the subsequent recruitment process. This indicator reflects the quality of channel conversions, not just the number of conversions.

[0054] The number of high-quality candidates can be determined by the following steps:

[0055] According to the requirements of the recruitment position, set the evaluation criteria for high-quality candidates (such as skill matching, years of work experience, etc.).

[0056] Screen the candidates converted from the channel and calculate the number of candidates that meet the standards, which is the number of high-quality candidates.

[0057] The number of high-quality candidates is an important indicator for evaluating the quality of channel recruitment. It helps recruitment platforms focus on the accuracy of channel conversion rather than just the accumulation of quantity. Channels with a higher number of high-quality candidates should receive more support in resource allocation to further improve overall recruitment results and efficiency.

[0058] The total number of candidates refers to the total number of all candidates obtained through a certain recruitment channel, including all job seekers who clicked on the recruitment information and submitted an application. This data is usually obtained directly through the back-end statistics of the channel or the monitoring system of the recruitment platform. For example, recruitment websites will record the total number of resumes submitted, social media recruitment ads will count the total number of people from ad clicks to completed applications, and professional communities may count the number of candidates who submitted and interacted. Through API interfaces, data export tools or real-time monitoring technology, these data can be integrated into the human resources shared service platform to form accurate statistical results, providing basic data support for subsequent analysis and resource allocation.

[0059] It can be seen from the click-through conversion factor that within the analysis window, the larger the performance value of the click-through conversion factor generated by analyzing the ratio of the extracted channel click volume to actual application, the better the actual conversion of the recruitment information of the channel. This is because the click-through conversion factor comprehensively considers the conversion effect from click behavior to actual application, and can reflect the efficiency and quality of the channel in attracting target candidates. A high click-through conversion factor usually means that the channel not only has a high number of clicks, but also that these clicks are more likely to be converted into actual job-seeking behavior, reflecting the high matching between the recruitment information and the target candidates. Conversely, a low click-through conversion factor indicates that the click behavior of the channel has not been effectively converted into actual application, which may be due to problems such as channel mismatch, insufficient information attractiveness, or inaccurate target population. Therefore, the click-through conversion factor is an important indicator to measure the conversion effect of the channel.

[0060] A high percentage of candidate interaction on a channel usually indicates that the actual conversion of recruitment information on that channel is good. This is because interactive behaviors (such as clicking to view, submitting resumes, adding positions to favorites, consulting or commenting, etc.) directly reflect candidates' interest and engagement in recruitment information. When the percentage of these interactive behaviors is high, it means that the information pushed by the channel has attracted the attention of the target group and prompted them to take further actions, thereby improving the reach and conversion rate of recruitment information. In addition, a high percentage of interactive behaviors usually also means that the user quality of the channel is high and the content of the recruitment information is more in line with the needs of candidates. Therefore, continuously monitoring candidates' interactive behavior indicators (such as interaction rate or interactive response factor) is one of the important methods to evaluate the conversion effect of channels, and provides data support for the subsequent optimization of push strategies and resource allocation.

[0061] Within the analysis window, the specific steps for generating the interaction response factor after analyzing the extracted candidate's interaction behavior ratio on the channel are as follows:

[0062] In the analysis window, the candidate's various interactive behaviors (such as clicks, favorites, inquiries, resume submissions, etc.) and their occurrence frequencies are recorded. Since the importance of each behavior is different, a weight is assigned to each interactive behavior, and the assigned weight is multiplied by the behavior frequency to obtain the weighted response contribution of each behavior. The calculation expression of the weighted response contribution is: ,in, is the weighted response contribution value of the i-th interactive behavior, is the total frequency of the candidate's ith interactive behavior within the analysis window (such as the number of clicks, the number of submissions). The square of the frequency is used to amplify the weight of high-frequency behaviors. is the weight of the ith interactive behavior, reflecting the relative contribution of the behavior to the conversion (e.g., the resume submission weight may be higher than the click weight). The weight is determined by the historical conversion data and the importance of the behavior. Introduce a logarithmic function to smooth the growth of high-frequency behavior and avoid excessive influence of extreme frequencies on the results;

[0063] Through this calculation, a weighted contribution value of each interactive behavior is generated. , comprehensively considering the frequency and importance of the behavior, providing a basis for the subsequent calculation of the overall activity of the channel.

[0064] The weighted response contributions of all interactive behaviors are nonlinearly superimposed to generate the total interactive response factor of the channel. In order to strengthen the identification of high-response channels, the exponential correction coefficient of the candidate's active coverage is added. The generation expression of the interactive response factor is: ,in, is the interactive response factor, n is the total number of candidate behavior categories, The nonlinear reinforcement factor that contributes to the behavior (controls the sensitivity of exponential growth), usually set to a positive value (such as 0.1-0.5), is the active coverage rate of candidates in the channel, that is, the ratio of candidates participating in interactive behaviors to the total number of candidates in the channel. is the exponential correction coefficient of active coverage, which is used to adjust the impact of active coverage on the results. Represents the weighted response contribution value of the i-th interactive behavior by Exponentiation of the exponents increases the influence of large weighted response contributions;

[0065] By nonlinearly superimposing the weighted contribution values ​​of each interactive behavior, the overall response of the channel is comprehensively measured, and combined with the active coverage of candidates, the score of highly active channels is further amplified. Index correction ensures that channel activity has a more sensitive reflection of the response quality of recruitment information.

[0066] It can be seen from the interactive response factor that within the analysis window, the larger the performance value of the interactive response factor generated by analyzing the proportion of the extracted candidates' interactive behaviors on the channel, the better the actual conversion of the recruitment information on the channel within the analysis window, and vice versa. This is because the interactive response factor combines the weights and frequencies of the candidates' various interactive behaviors on the channel (such as clicks, submissions, consultations, etc.), reflecting the candidates' interest in the recruitment information and the depth of their participation. In addition, the interactive response factor also introduces correction factors such as the candidate's active coverage rate, further strengthening the ability to measure the overall activity of the channel. Therefore, when the interactive response factor performance value is large, it means that the channel not only attracts more candidates' attention, but also prompts candidates to take more conversion behaviors, such as resume submission or job application. On the contrary, when the interactive response factor is low, it may mean that the exposure effect of the recruitment information on the channel is not good, or it does not match the needs of the target candidates, resulting in insufficient interaction and conversion behaviors. By monitoring the interactive response factor, the actual conversion performance of the channel can be intuitively evaluated, providing a scientific basis for optimizing the allocation of recruitment resources.

[0067] After an in-depth analysis of the ratio of extracted channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel, the click conversion factor and interactive response factor generated by the analysis are input into the pre-trained machine learning model, and the push return index is output based on the model. The actual conversion of recruitment information is predicted by the push return index.

[0068] Pre-trained machine learning models refer to algorithm models that are trained and optimized in advance using a large amount of historical data in the recruitment information push system. These models learn and identify the key factors that affect the recruitment conversion effect by analyzing past recruitment activity data, including the number of clicks, conversion rates, candidate interaction behaviors, etc. on various channels. Specifically, the model training process involves multiple steps such as data collection, feature engineering, model selection, training and verification. First, collect detailed data on the company's past on different recruitment channels, including the number of clicks, resume submissions, interaction behavior ratios, and final conversion rates for each channel. Next, through feature engineering, these raw data are converted into features suitable for machine learning model processing, such as standardizing clicks and conversion rates, extracting candidate behavior patterns, etc. Then, select an appropriate machine learning algorithm, such as decision tree, random forest, support vector machine, or deep learning model, for training. During the training process, the model continuously adjusts parameters to minimize prediction errors and improve the accuracy of predictions on recruitment conversions. Through cross-validation and test set evaluation, ensure the generalization ability of the model on unseen data and avoid overfitting. After training is completed, these models can be called in actual applications to analyze the performance indicators of various channels in current recruitment activities in real time and predict the conversion effects of different channels.

[0069] In practical applications, pre-trained machine learning models play a key role. By inputting the click-through conversion factor and interactive response factor monitored in real time, the push return index is output to predict the actual conversion of recruitment information. This process first relies on the model's understanding and association ability of the input features, that is, the model can evaluate the performance and potential of each channel in the current recruitment environment based on the current channel performance indicators. The push return index output by the model reflects the conversion effect that each channel may bring in the future, helping recruitment managers make more accurate and intelligent resource allocation decisions. This dynamic adjustment based on prediction not only improves resource utilization efficiency, but also significantly shortens the recruitment cycle, improves candidate matching and overall return on investment. In addition, the pre-trained model can continuously learn and adapt to new data. As more recruitment activities are carried out, the model's predictive ability and accuracy will continue to improve, further optimizing the recruitment strategy. This intelligent, data-driven recruitment management method enables companies to have a stronger competitive advantage in the fiercely competitive recruitment environment, and can quickly respond to market changes and attract and retain high-quality talents.

[0070] The machine learning model is not specifically limited here, and can achieve the conversion factor of clicks and the interactive response factor Conduct comprehensive analysis to generate push return index In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; push reward index The resulting calculation formula is: , where , Click-through conversion factor and the interactive response factor The preset scaling factor of , Both are greater than 0.

[0071] The preset scaling factor here ( and ) refers to the weight distribution coefficients for click conversion factor and interactive response factor in the push return index calculation formula. These coefficients determine the degree of influence of the two indicators on the final result. and The value of can flexibly adjust the relative importance of the two indicators according to actual needs.

[0072] For example:

[0073] If the recruitment conversion process relies more on the conversion effect of click behavior, it can improve to increase the click-through conversion factor The weight in the calculation.

[0074] If you pay more attention to the depth and activity of candidates' interactions, you can increase to increase the interactive response factor The weight of .

[0075] These preset ratio coefficients are usually derived from experience or historical data analysis, and can also be optimized through model training to ensure that the push return index can more accurately reflect the real conversion potential of the channel. This flexibility allows the calculation formula to adapt to the needs of different scenarios and recruitment strategies, improving the practicality and predictive ability of the system.

[0076] It can be seen from the push return index that, within the analysis window, the greater the performance value of the click conversion factor generated after analyzing the ratio of the extracted channel clicks to actual applications, the greater the performance value of the interactive response factor generated after analyzing the proportion of the extracted candidates' interactive behaviors on the channel. In other words, the greater the performance value of the push return index generated when predicting the actual conversion of the channel recruitment information within the analysis window through the pre-trained machine learning model, the better the actual conversion of the recruitment information of the channel within the analysis window, and vice versa.

[0077] Dynamic weight adjustment and optimization module, based on the prediction results of the machine learning model, dynamically adjusts the push weight of each recruitment channel, increases the push weight of channels with excellent prediction performance, optimizes resource allocation in real time, and prioritizes resources to efficient channels, thereby improving the overall recruitment effect;

[0078] Based on the prediction results of the machine learning model, dynamically adjust the push weight of each recruitment channel. The specific steps are as follows:

[0079] During the push process, the key performance indicators (click conversion factor and interactive response factor) of each channel are monitored in real time, and the key performance indicators are input into the pre-trained machine learning model. The push return index generated by the model is used to evaluate the conversion potential of the channel;

[0080] Compare the push return index with the push return index reference threshold, calculate the channel return index deviation, and accurately evaluate the actual performance of each recruitment channel. The calculation expression of the return index deviation is: ,in, It is the reference threshold of the push return index, which is used to judge whether the channel performs well. is the push return index generated by the machine learning model prediction for channel k, is the return index deviation of channel k;

[0081] By generating a push return index through a machine learning model and comparing it with the push return index reference threshold, it is possible to identify which channels are performing well or poorly, providing a scientific basis for the subsequent dynamic adjustment of push weights. The push return index reference threshold can be set based on historical data or the company's recruitment goals, making decisions more targeted.

[0082] According to the comparison result between the push return index and the reference threshold of the push return index, the push weight of the channel is dynamically adjusted. The adjusted weight is calculated by the following formula: ,in: is the adjusted push weight of channel k, is the initial push weight of channel k, is the adjustment coefficient used to control the change range of push weight. Make sure to only weight positive deviations (outperforming channels)

[0083] Dynamically adjusting push weights can ensure flexibility and accuracy in resource allocation. For channels with outstanding performance, increasing push weights can expand their advantages; while for channels with insufficient performance, reducing weights can avoid wasting resources. This dynamic adjustment mechanism enables real-time optimization of recruitment information push strategies to adapt to changing recruitment needs and channel performance.

[0084] When the push return index is less than or equal to the push return index reference threshold, continue to maintain the initial push weight of the recruitment channel. Its role is to avoid reducing resource investment too early and ensure that the normal fluctuation of the channel within a certain range does not interfere with the push strategy. This strategy can maintain the stability of channel resources and provide them with sufficient time and opportunities to realize their potential in subsequent pushes. The performance of some channels may be temporarily lower than the push return index reference threshold due to short-term factors (such as time period, target candidate activity, etc.), but it does not mean that the overall effect of the channel is poor. By maintaining the initial push weight, fluctuations in resource allocation caused by excessive adjustments can be avoided, thereby ensuring the robustness and continuity of the push strategy, and accumulating more data basis for subsequent dynamic optimization.

[0085] The resource reallocation and weight update module reallocates the actual push weight of recruitment information after completing the dynamic adjustment of push weights, and readjusts the resource investment of each channel based on the latest weight allocation strategy;

[0086] Re-adjust the resource investment in each channel, such as adjusting the advertising budget, reallocating the push frequency and exposure, etc. This step not only involves the reallocation of resources, but also requires fine-tuning the push strategy to adapt to the dynamic changes in the needs of different recruitment information. For example, for technical positions, the push ratio in professional communities and technical forums can be increased; for sales positions, the push intensity on social media and recruitment websites can be increased. In addition, it is necessary to continuously monitor the effect of the adjustment to ensure that the new push strategy can effectively improve the recruitment conversion rate, and further optimize the push weight based on feedback.

[0087] Through the above-mentioned human resources sharing service platform solution based on big data, enterprises can significantly improve recruitment efficiency and effectiveness, and realize precise and dynamic optimization of resource allocation. First, the reasonable setting of the initial push weight ensures that recruitment resources are concentrated on the efficient channels that best meet the needs of the enterprise at the start-up stage, maximizing information coverage and candidate reach. Secondly, real-time monitoring and analysis of the performance indicators of each channel enable the platform to promptly identify the channels with excellent performance, and make accurate predictions through pre-trained machine learning models, so as to dynamically adjust the push weight, allocate more resources to efficient channels first, and significantly improve the conversion rate and recruitment quality. In addition, the process of reallocating push weights effectively avoids the waste of resources in inefficient channels, reduces recruitment costs, and improves the return on investment. Overall, the solution not only optimizes resource utilization and shortens the recruitment cycle, but also enhances the advantages of enterprises in the fierce competition for talents, ensuring that high-quality talents can be quickly attracted and retained, and ultimately promoting the sustainable development of the enterprise's business and the improvement of competitiveness.

[0088] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0089] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0090] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A human resources sharing service platform based on big data, characterized in that: It includes demand analysis and initial weight setting module, information push and resource allocation module, channel performance monitoring and data analysis module, dynamic weight adjustment and optimization module, and resource reallocation and weight update module; Demand analysis and initial weight setting module: before the recruitment activity begins, first fully understand the company's recruitment needs, and based on the recruitment needs, set the initial push weights of various recruitment channels to ensure that resources can be allocated in an orderly manner to the channels that best meet the company's needs at the beginning; The information push and resource allocation module, after completing the setting of the initial push weight, starts to push recruitment information on various channels according to the set weight ratio, so that each channel receives a specified number of recruitment advertisements or job information according to its assigned weight; The channel performance monitoring and data analysis module continuously monitors the performance indicators of each channel during the recruitment information push period, extracts key data that can reflect the channel conversion effect, conducts in-depth analysis on the extracted data, and inputs the analysis results into the pre-trained machine learning model to predict the actual conversion of recruitment information; Dynamic weight adjustment and optimization module, based on the prediction results of the machine learning model, dynamically adjusts the push weight of each recruitment channel, increases the push weight of channels with excellent prediction performance, optimizes resource allocation in real time, and prioritizes resources to efficient channels, thereby improving the overall recruitment effect; The resource reallocation and weight update module, after completing the dynamic adjustment of push weights, reallocates the actual push weights of recruitment information, and readjusts the resource input of each channel based on the latest weight allocation strategy.

2. According to the big data-based human resources sharing service platform of claim 1, it is characterized in that: The initial push weight refers to the initial resource ratio allocated to each channel at the beginning of the recruitment activity, based on the company's recruitment needs and the characteristics of each recruitment channel. It reflects the degree to which the company inclines its resources to different channels when pushing recruitment information.

3. According to the big data-based human resources sharing service platform of claim 1, it is characterized in that: During the push of recruitment information, the performance indicators of various channels are continuously monitored to extract key data that can reflect the channel conversion effect. The extracted key data include the ratio of channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel. Within the analysis window, after analyzing the ratio of the extracted channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel, a click conversion factor and an interactive response factor are generated respectively. The click conversion factor is used to measure the actual effect of the click behavior, and the interactive response factor is used to measure the activity of the channel.

4. According to the big data-based human resources sharing service platform of claim 3, it is characterized in that: After an in-depth analysis of the ratio of extracted channel clicks converted to actual applications and the proportion of candidates' interactive behaviors on the channel, the click conversion factor and interactive response factor generated by the analysis are input into the pre-trained machine learning model, and the push return index is output based on the model. The actual conversion of recruitment information is predicted by the push return index.

5. According to the big data-based human resources sharing service platform of claim 3, it is characterized in that: In the analysis window, the specific steps for generating the click conversion factor after analyzing the ratio of extracted channel clicks to actual applications are as follows: For each channel, the total click volume of the channel and the actual number of applications are nonlinearly mapped according to the weight ratio to calculate the basic conversion contribution value of its click behavior to the actual application, reflecting the basic conversion effect of the channel click behavior on the actual application. The calculation expression is: ,in, is the basic conversion contribution value, A is the actual application volume, which indicates the number of job applications submitted through this channel, K is the channel click volume, which indicates the total number of times the recruitment information is clicked by candidates in this channel, L is the candidate's behavioral participation, and B is the channel characteristic coefficient, which reflects the advantages or disadvantages of the channel. The click conversion factor weight is used to adjust the mapping weight between the click volume and the actual application. is the candidate behavior factor weight, regulating the impact of candidate behavior engagement; After calculating the basic conversion contribution value, the final click conversion factor is generated by combining the channel weight and resource allocation coefficient. The generation expression of the click conversion factor is: ,in, is the click conversion factor, which measures the actual conversion effect of click behavior; R is the channel resource utilization coefficient, which reflects the efficiency of the channel resource allocation; S is the resource saturation; is the resource saturation adjustment factor, which is used to smooth the impact of resource saturation on the click conversion factor. P is the number of high-quality candidates in the channel, which refers to the number of high-quality candidates converted from the channel. F is the total number of candidates, which refers to the number of all candidates converted from the channel. Adjust the weight for basic conversion and control the impact of basic conversion contribution value. The weight of high-quality candidates controls the contribution of the proportion of high-quality candidates to the click-through conversion factor.

6. The human resources sharing service platform based on big data according to claim 3 is characterized in that: Within the analysis window, the specific steps for generating the interaction response factor after analyzing the extracted candidate's interaction behavior ratio on the channel are as follows: In the analysis window, the candidate's various interactive behaviors and their occurrence frequencies are recorded. Since the importance of each behavior is different, a weight is assigned to each interactive behavior, and the assigned weight is multiplied by the behavior frequency to obtain the weighted response contribution of each behavior. The calculation expression of the weighted response contribution is: ,in, is the weighted response contribution value of the i-th interactive behavior, is the total frequency of the candidate's ith interactive behavior within the analysis window. The square of the frequency is used to amplify the weight of high-frequency behaviors. is the weight of the i-th interactive behavior, reflecting the relative contribution of this behavior to the conversion. Introducing a logarithmic function to smooth the growth of high-frequency behavior; The weighted response contributions of all interactive behaviors are nonlinearly superimposed to generate the total interactive response factor of the channel. In order to strengthen the identification of high-response channels, the exponential correction coefficient of the candidate's active coverage is added. The generation expression of the interactive response factor is: ,in, is the interactive response factor, n is the total number of candidate behavior categories, The nonlinear reinforcement factor that contributes to the behavior, is the active coverage rate of candidates in the channel, that is, the ratio of candidates participating in interactive behaviors to the total number of candidates in the channel. is the exponential correction coefficient of active coverage, which is used to adjust the impact of active coverage on the results. Represents the weighted response contribution value of the i-th interactive behavior by Raises the exponent to a power, increasing the influence of large weighted response contributions.

7. The human resources sharing service platform based on big data according to claim 4 is characterized in that: Based on the prediction results of the machine learning model, dynamically adjust the push weight of each recruitment channel. The specific steps are as follows: During the push process, the key performance indicators of each channel are monitored in real time and input into the pre-trained machine learning model. The push return index generated by the model is used to evaluate the conversion potential of the channel. Compare the push return index with the push return index reference threshold, calculate the channel return index deviation, and accurately evaluate the actual performance of each recruitment channel. The calculation expression of the return index deviation is: ,in, It is the reference threshold of the push return index, which is used to judge whether the channel performs well. is the push return index generated by the machine learning model prediction for channel k, is the return index deviation of channel k.

8. The human resources sharing service platform based on big data according to claim 7 is characterized in that: According to the comparison result between the push return index and the reference threshold of the push return index, the push weight of the channel is dynamically adjusted. The adjusted weight is calculated by the following formula: ,in: is the adjusted push weight of channel k, is the initial push weight of channel k, is the adjustment coefficient used to control the change range of push weight. Make sure to only weight positive deviations.

Images