An optimization method and service system for precise push based on big data analysis

Abstract

The present invention discloses an optimization method and service system for precise push based on big data analysis, which relates to the field of human resources technology. In S1, various job requirements and user resume data are collected and integrated in advance to generate a static data set, which provides basic information support for subsequent matching. S2 uses the job matching factor Gpyz to preliminarily screen users and jobs that have not communicated, so as to accurately locate potential compatible target users and target jobs, thereby improving matching efficiency. In S3, the method generates a dynamic data set and monitors the activity level by counting the behaviors of target users and actively interacting users, analyzes the similarity Xsd, and constructs the satisfaction coefficient of active users for various jobs, further enhancing the understanding of job requirements and user preferences. S4 predicts the satisfaction coefficient and combines the evaluation threshold Q for further comparative analysis, so that the system can further screen out candidates that are highly matched with job requirements.

Description

Technical Field

[0001] The present invention relates to the field of human resources technology, and specifically to an optimization method and service system for precise push based on big data analysis. Background Art

[0002] Precision push technology based on big data analysis is widely used in many fields, especially in the field of human resources. With the diversification of recruitment platforms and job search methods, how to accurately push suitable positions to suitable users through massive data analysis has become one of the core challenges of modern recruitment system optimization. Especially in the process of matching users and positions, matching push between positions and users can not only improve the efficiency of recruitment, but also enable job seekers to find suitable job opportunities faster.

[0003] However, current job push systems often face some practical problems. For example, many systems rely on static data, ignoring the dynamic behavior and activity of users on recruitment platforms. In addition, traditional job recommendation methods rely more on direct matching of job tags, resulting in inaccurate matching of recommendations. The similarities between users, active interactive behaviors, and actual user satisfaction with a certain type of job are usually not fully considered, resulting in low accuracy of recommended jobs and poor user experience. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an optimization method and service system for precise push based on big data analysis, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an optimization method for accurate push based on big data analysis, comprising the following steps:

[0006] S1. Based on big data technology, various job requirements and user resumes are collected in advance to obtain relevant demand data information of various positions and relevant resume data information submitted by each user on the recruitment platform, and generate a static data set;

[0007] S2. Based on the static data set, analyze the job matching degree between users and uncommunicated jobs, construct the uncommunicated job matching factor Gpyz, and preliminarily screen out target users that are suitable for the corresponding types of jobs, and take the jobs corresponding to the target users as target jobs;

[0008] S3. Based on the target users selected in S2, the users who actively interact in this type of position are counted to generate the active user set A, and the active users and target users are monitored for various types of positions on the recruitment platform to obtain a dynamic data set. Based on the dynamic data set, the similarity Xsd between the target users and the active users is analyzed, and the active users' satisfaction coefficient for each type of position is constructed in combination with the active users' activeness on the recruitment platform. ;

[0009] S4. Satisfaction coefficient of active users for various types of positions , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

[0010] Preferably, S1 specifically includes:

[0011] S11. Collecting various job requirements and user resumes to obtain relevant demand data information for various jobs and relevant resume data information submitted by each user on the recruitment platform, wherein the relevant demand data information includes job requirement descriptions for various jobs and the total number of all types of job description documents ;

[0012] The relevant resume data information includes the resume description submitted by the user on the recruitment platform;

[0013] S12. Preprocess the relevant demand data information of various positions and the relevant resume data information submitted by each user on the recruitment platform to generate a static data set, wherein the preprocessing includes noise removal, missing value filling and data smoothing operations, and the missing value filling methods include mean filling, median filling, interpolation filling and regression filling, and use dimensionless processing technology to scale the information in the static data set so that the information range in the static data set falls between 0 and 1.

[0014] Preferably, S2 specifically includes:

[0015] S21. Use natural language processing technology to segment the static data set, remove stop words, and extract keywords in the relevant demand data information and the relevant resume data information;

[0016] S22, based on the keywords in the relevant demand data information and the keywords in the relevant resume data information obtained in S21, respectively obtain the word frequency Cp1 of the keywords in the relevant demand data information, the word frequency Cp2 of the keywords in the relevant resume data information, the inverse document frequency Ndp1 of the keywords in the relevant demand data information in all types of positions, and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of positions;

[0017] S221, the inverse document frequency Ndp1 of the keywords in the relevant demand data information in all types of positions and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of positions are obtained by the following method:

[0018] ;

[0019] In the formula, It is expressed as the inverse document frequency of the ith keyword in all types of positions. Represented as the total number of job description documents of all types, It is represented by the number of times the i-th keyword appears in the job description document.

[0020] Preferably, S2 also includes:

[0021] S23. Based on the keywords in the relevant demand data information and the keywords in the relevant resume data information obtained in S21, the job matching degree between the user and the uncommunicated job is analyzed to calculate the uncommunicated job matching factor Gpyz, which is specifically obtained by the following formula:

[0022] ;

[0023] In the formula, k is a common keyword; K is a set of keywords that appear in the relevant demand data information and the relevant resume data information; It is represented by the frequency of common keyword k in the relevant demand data information; It is expressed as the frequency of common keywords in relevant resume data information. It is expressed as the inverse document frequency of common keywords in all types of positions in the relevant demand data information. It is represented by the inverse document frequency of common keywords in all types of positions in the relevant resume data information. It is represented as the weight vector of common keywords in relevant demand data information, It is represented as the weight vector of common keywords in relevant resume data information, The common keyword k is a keyword that appears in both the relevant demand data information and the relevant resume data information.

[0024] Preferably, S2 also includes:

[0025] S24. Preset a matching threshold W, and compare and analyze the uncommunicated job matching factor Gpyz with the matching threshold W to preliminarily screen out target users that are suitable for the corresponding type of job. The specific preliminary judgment content is as follows:

[0026] If the uncommunicated job matching factor Gpyz exceeds the matching threshold W, it is preliminarily determined that the corresponding user resume is compatible with the requirements of the corresponding type of job. At this time, the user and the type of job will be labeled as the target user and target job respectively;

[0027] If the uncommunicated job matching factor Gpyz does not exceed the matching threshold W, it is preliminarily judged that the corresponding user resume is temporarily unmatched with the requirements of the corresponding type of job. At this time, no labeling operation will be performed on the user and the type of job.

[0028] Preferably, S3 specifically includes:

[0029] S31. Based on the target users screened out in S2, count the users who actively consulted the target position in the historical period, mark the users as active users, and generate an active user set A;

[0030] S32, real-time monitoring and recording of the activeness of active users and target users on the recruitment platform for various positions, so as to obtain a dynamic data set, wherein the dynamic data set includes the types of positions actively inquired by active users, the types of positions actively inquired by target users, and the number of views Lgcs of the positions actively inquired by active users;

[0031] S321. Construct a set of types of positions actively inquired by active users based on the types of positions actively inquired by active users , based on the types of positions that the target users actively ask about, build a set of types of positions that the target users actively ask about .

[0032] Preferably, S3 also includes:

[0033] S33: Based on the dynamic data set, the similarity Xsd between the target user and the active user is analyzed, which is specifically obtained by the following method:

[0034] ;

[0035] In the formula, It represents the set of job types that the target user actively inquires about. It represents the set of job types that active users actively inquire about. is represented as the intersection of two sets, Represented as the union of two sets;

[0036] S34. According to the activeness of active users on the recruitment platform for various positions, the number of views Lgcs and the proportion of views Lzb are linearly normalized, and the corresponding data values ​​are mapped to the interval To construct the satisfaction coefficient of active users for various types of positions , which can be obtained by:

[0037] ;

[0038] Where n is the number of active users browsing the same type of posts, j=1, 2, 3, ..., n, It is represented by the number of views of the j-th type of job, Expressed as the average number of views, It is expressed as the browsing proportion of the j-th type of job, Expressed as the average browsing share, and are weight values, Expressed as a correction constant.

[0039] Preferably, S4 specifically includes:

[0040] S41. Based on active users’ satisfaction with various types of positions , combined with the similarity Xsd between the target user and the active user, and after linear normalization, the target user's satisfaction coefficient for each type of position is predicted , the specific contents are as follows:

[0041] ;

[0042] In the formula, a represents the index of the active user, A represents the set of active users, Indicates that active user a belongs to the active user set, It is expressed as the similarity between the target user and the active user a, It is expressed as the satisfaction coefficient of active user a for the j-th type of job.

[0043] Preferably, S4 also includes:

[0044] S42. Obtain the target user's satisfaction coefficient for each type of position based on S41 , extract the target user's satisfaction coefficient for the target position;

[0045] S43, pre-set the evaluation threshold Q, and compare and analyze the target user's satisfaction coefficient with the target position with the evaluation threshold Q to screen out the candidates to be interviewed for the target position. The specific contents are as follows:

[0046] If the target user's satisfaction coefficient for the target position exceeds the evaluation threshold Q, the target position will be pushed to the target user through the recruitment platform, and the target user will be pushed to the target position, and the target user will be considered as a candidate for interview;

[0047] If the target user's satisfaction coefficient for the target position does not exceed the evaluation threshold Q, the target position will not be pushed to the target user through the recruitment platform at this time, and the target user will not be pushed to the target position at the same time.

[0048] A precision push service system based on big data analysis, including a static data monitoring module, a preliminary matching analysis module, a push analysis module and a locking module;

[0049] The static data monitoring module collects various job requirements and user resumes in advance based on big data technology to obtain relevant demand data information of various jobs and relevant resume data information submitted by each user on the recruitment platform, and generates a static data set;

[0050] The preliminary matching analysis module constructs an uncommunicated job matching factor Gpyz based on a static data set, and preliminarily screens out target users that are suitable for the corresponding type of job, and uses the job corresponding to the target user as the target job;

[0051] The push analysis module counts the users who actively interact in this type of position based on the screened target users to generate an active user set A, and monitors the active users and target users' activeness on the recruitment platform for various positions to obtain a dynamic data set. Based on the dynamic data set, the similarity Xsd between the target users and the active users is analyzed, and the active users' satisfaction coefficient for each type of position is constructed in combination with the active users' activeness on the recruitment platform for various positions. ;

[0052] The locking module is based on the satisfaction coefficient of active users for various types of positions. , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

[0053] The present invention provides an optimization method and service system for accurate push based on big data analysis, which has the following beneficial effects:

[0054] (1) First, in step S1, various job requirements and user resume data are collected and integrated in advance to generate a static data set, which provides basic information support for subsequent matching. Then, in step S2, the job matching factor Gpyz is used to preliminarily screen users and jobs that have not communicated, so as to accurately locate potential target users and target jobs, thereby improving matching efficiency. In step S3, this method generates a dynamic data set and monitors the activity level by counting the behaviors of target users and actively interacting users, analyzes the similarity Xsd, and constructs the satisfaction coefficient of active users for various jobs, further enhancing the understanding of job requirements and user preferences. Finally, in step S4, the satisfaction coefficient is predicted and combined with the evaluation threshold Q for further comparison and analysis, so that the system can further screen out candidates who are highly matched with job requirements, ensuring that only those candidates who are truly interested in the job and have a high degree of match can enter the interview stage, thereby improving the quality of interview candidates, reducing the matching blind spots between employers and job seekers, and optimizing the overall effect of job recruitment.

[0055] (2) Based on the similarity between users, the target user's preference for unknown items is predicted through the behavior and ratings of active users. In real scenarios, data sparsity is often a big problem. For example, in a large human resources system, the target user may have only rated or browsed a few positions, while the types and number of positions in the system are far greater than the number of these ratings. At this time, the system can help solve this sparsity problem: by using the ratings of other positions by users similar to the target user, the prediction bias caused by insufficient target user data can be further compensated. This is an idea based on collective wisdom. Through the behavior and ratings of other users, the prediction accuracy of the target user's preferences can be improved. At the same time, by dynamically using the latest behavior and ratings of active users to continuously update the prediction, the problem that the target user's interests and preferences will change over time and experience can be captured. Directly relying on the historical ratings of the target user may be difficult to reflect these dynamic changes, while predictions through active users can better adapt to these changes.

[0056] (3) This method uses natural language processing technology to extract keywords from job descriptions and user resumes, and further constructs an uncommunicated job matching factor Gpyz through the analysis of common keywords, thereby accurately measuring the degree of matching between users and jobs. This method is based on the comprehensive calculation of keyword frequency, inverse document frequency, and weight vectors, ensuring in-depth analysis of user resumes and job requirements, achieving more refined matching, and being able to quickly exclude resumes that do not meet job requirements in the initial screening stage, further reducing the subsequent screening burden. Through the comparison and analysis of matching factors, the system can automatically identify users with high adaptability and mark them as target users, and at the same time mark the corresponding target jobs. This dynamic marking and judgment process effectively improves the efficiency of job recommendations and user resume screening, and further avoids the rough or one-sided screening methods in traditional matching.

[0057] (4) Similarity analysis is based on the intersection and union of the target user and the active user in the job query set, ensuring that the recommendation system can more accurately identify active users with similar behaviors to the target user, thereby providing more relevant job recommendations for the target user. Combined with the active user's satisfaction coefficient for each type of job, the system can more accurately predict the target user's potential satisfaction with each type of job. This screening mechanism can effectively improve the relevance and effectiveness of job recommendations, thereby reducing mismatched recommendation results and optimizing the user's job search experience. Through this data-based precise push method, the recruitment platform can more effectively match users and jobs, reduce users' search costs, and increase the chances of successful recruitment.

[0058] (5) Through a refined satisfaction prediction and screening mechanism, the recruitment platform can more efficiently match users with positions, reduce the time for invalid recommendations and screening, and improve the overall efficiency of the recruitment process. This optimization not only improves the service quality of the platform, but also enhances the user's sense of participation and the platform's competitiveness. Accurate job recommendations and user matching can improve user experience and user satisfaction as much as possible. Users can more easily find positions that match their interests and abilities, thereby increasing the user stickiness and activity of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a flow chart of an optimization method for accurate push based on big data analysis of the present invention;

[0060] Figure 2 This is a block diagram of a service system for precise push based on big data analysis in the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] Example 1

[0063] See also Figure 1 The present invention provides an optimization method for accurate push based on big data analysis, comprising the following steps:

[0064] S1. Based on big data technology, various job requirements and user resumes are collected in advance to obtain relevant demand data information of various positions and relevant resume data information submitted by each user on the recruitment platform, and generate a static data set;

[0065] S2. Based on the static data set, analyze the job matching degree between users and uncommunicated jobs, construct the uncommunicated job matching factor Gpyz, and preliminarily screen out target users that are suitable for the corresponding types of jobs, and take the jobs corresponding to the target users as target jobs;

[0066] S3. Based on the target users selected in S2, the users who actively interact in this type of position are counted to generate the active user set A, and the active users and target users are monitored for various types of positions on the recruitment platform to obtain a dynamic data set. Based on the dynamic data set, the similarity Xsd between the target users and the active users is analyzed, and the active users' satisfaction coefficient for each type of position is constructed in combination with the active users' activeness on the recruitment platform. ;

[0067] S4. Satisfaction coefficient of active users for various types of positions , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

[0068] In this embodiment, the method collects and analyzes the demand data and user resume information of various positions, establishes a static data set, and then constructs a position matching factor Gpyz. Through this matching factor, users who are suitable for the target position can be preliminarily screened out, further improving the matching accuracy between users and positions, thereby reducing the error in the matching process. In addition, combined with the dynamic behavior of users, an active user set A is further generated, and the activeness of active users and target users on the recruitment platform is dynamically monitored to generate a dynamic data set, further improving the real-time and accuracy of the data; the method also calculates the satisfaction coefficient of active users for various positions by analyzing the similarity Xsd between active users and target users, combined with the actual behavior data of users on the platform, and predicts the satisfaction of target users with the target position. This combination of dual static and dynamic data enables the system to more accurately push positions that meet the expectations of users and enterprises, and by setting the evaluation threshold Q, it can automatically screen and optimize interview candidates, further improving the efficiency and accuracy of the recruitment process. This method effectively solves the problems of inaccurate matching and delayed response in traditional push systems, and improves the user experience of the recruitment platform.

[0069] Example 2

[0070] Please refer to Figure 1 , specifically: S1 specifically includes:

[0071] S11. Collecting various job requirements and user resumes to obtain relevant demand data information for various jobs and relevant resume data information submitted by each user on the recruitment platform, wherein the relevant demand data information includes job requirement descriptions for various jobs and the total number of all types of job description documents ;

[0072] The relevant resume data information includes the resume description submitted by the user on the recruitment platform;

[0073] S12. Preprocess the relevant demand data information of various positions and the relevant resume data information submitted by each user on the recruitment platform to generate a static data set, wherein the preprocessing includes removing noise, filling missing values ​​and data smoothing operations to reduce dimensions and improve calculation efficiency. The missing value filling methods include mean filling, median filling, interpolation filling and regression filling. The dimensionless processing technology is used to scale the information in the static data set so that the information range in the static data set falls between 0 and 1.

[0074] In this embodiment, the method makes the processing of job requirements and user resume information more efficient and accurate by introducing data preprocessing technology. First, in step S11, the job requirement data information and the resume information submitted by users are collected from multiple dimensions, ensuring the integrity and coverage of the data and laying a data foundation for subsequent matching analysis. The relevant requirement data information includes the detailed job description and the total number of description documents for all jobs, further enhancing the comprehensiveness of job matching. In step S12, the method performs a series of preprocessing operations, such as removing noise, filling in missing values, and data smoothing operations, to ensure the accuracy and consistency of the data. The methods for filling in missing values include mean filling, median filling, interpolation filling, and regression filling to ensure data integrity and further avoid the impact of data missing on subsequent analysis. At the same time, the dimensionless processing technology is used to scale the data, restricting the data range between 0 and 1, and improving the calculation efficiency and stability of the model as much as possible. In short, the static data set generated through this step provides accurate input data for the subsequent matching algorithm, thus effectively improving the accuracy and efficiency of job recommendations.

[0075] Embodiment 3

[0076] Please refer to Figure 1 , specifically: S2 specifically includes:

[0077] S21. Use natural language processing technology to segment the static data set, remove stop words (such as "de", "shi"), and extract the keywords in the relevant requirement data information and the keywords in the relevant resume data information;

[0078] S22. Based on the keywords in the relevant requirement data information and the keywords in the relevant resume data information obtained in S21, respectively obtain the word frequency Cp1 of the keywords in the relevant requirement data information, the word frequency Cp2 of the keywords in the relevant resume data information, the inverse document frequency Ndp1 of the keywords in the relevant requirement data information in all types of jobs, and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of jobs;

[0079] Among them, the word frequency refers to counting the number of occurrences of each keyword in a single description document to calculate the relative word frequency; the inverse document frequency refers to counting the number of documents containing the keyword in all types of jobs, indicating the rarity of a certain keyword in all types of jobs;

[0080] S221. The inverse document frequency Ndp1 of the keywords in the relevant requirement data information in all types of jobs and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of jobs are both obtained through the following method:

[0081] ;

[0082] In the formula, It is expressed as the inverse document frequency of the ith keyword in all types of positions. Represented as the total number of job description documents of all types, It is expressed as the number of job description documents in which the i-th keyword appears, that is, the number of job description documents in which the i-th keyword appears; the purpose of adding 1 is to avoid the denominator being 0 when the word t does not appear in all documents.

[0083] In this embodiment, natural language processing technology is used to segment the job requirements and user resumes in the static data set and remove stop words, extract keywords, and calculate the weight of keywords in combination with word frequency and inverse document frequency, so that the system can effectively capture the key information in the job description and user resume. In this way, the accuracy of the job matching algorithm can be improved, thereby further avoiding interference from irrelevant information. At the same time, the introduction of inverse document frequency prevents the weight of high-frequency commonly used words from being too high, ensuring that rare but important keywords can be highlighted.

[0084] Example 4

[0085] Please refer to Figure 1 Specifically, S2 also includes:

[0086] S23. Based on the keywords in the relevant demand data information and the keywords in the relevant resume data information obtained in S21, the job matching degree between the user and the uncommunicated job is analyzed to calculate the uncommunicated job matching factor Gpyz, which is specifically obtained by the following formula:

[0087] ;

[0088] In the formula, k is a common keyword, which is used to indicate the position of the keyword in the relevant demand data information and the relevant resume data information; K is a set of keywords that appear in the relevant demand data information and the relevant resume data information; It is represented by the frequency of common keyword k in the relevant demand data information, It is expressed as the frequency of common keywords in relevant resume data information. It is expressed as the inverse document frequency of common keywords in all types of positions in the relevant demand data information. It is represented by the inverse document frequency of common keywords in all types of positions in the relevant resume data information. It is represented as the weight vector of common keywords in relevant demand data information, It is represented as the weight vector of common keywords in relevant resume data information, The common keyword k is a keyword that appears in both the relevant demand data information and the relevant resume data information.

[0089] S2 also includes:

[0090] S24. Preset a matching threshold W, and compare and analyze the uncommunicated job matching factor Gpyz with the matching threshold W to preliminarily screen out target users that are suitable for the corresponding type of job. The specific preliminary judgment content is as follows:

[0091] If the uncommunicated job matching factor Gpyz exceeds the matching threshold W, it is preliminarily determined that the corresponding user resume is compatible with the requirements of the corresponding type of job. At this time, the user and the type of job will be labeled as the target user and target job respectively;

[0092] If the uncommunicated job matching factor Gpyz does not exceed the matching threshold W, it is preliminarily judged that the corresponding user resume is temporarily unmatched with the requirements of the corresponding type of job. At this time, no labeling operation will be performed on the user and the type of job.

[0093] In this embodiment, the method can accurately calculate the matching factor Gpyz between the user and the position that has not been communicated by using the word frequency and inverse document frequency analysis of keywords. This matching factor constructs a weight vector based on the common keywords of the job requirements and the user's resume, ensuring that the content of the job requirements and the resume description can be accurately compared, further improving the matching accuracy. By setting the matching threshold W, the target users who are relatively suitable for the job requirements can be effectively screened out, and the target positions and target users are automatically marked, thereby reducing manual screening time and improving recruitment efficiency. At the same time, for unsuitable users and positions, the system automatically removes them, further avoiding unnecessary subsequent operations and optimizing push accuracy.

[0094] It should be noted that the matching threshold W can be obtained in the following way:

[0095] By performing statistical calculations on multiple groups of uncommunicated job matching factors Gpyz, the average job matching factors of uncommunicated jobs are obtained. and the standard deviation of the job matching factor that was not communicated , based on the average job match factor without communication and the standard deviation of the job matching factor that was not communicated , set the matching threshold W: Among them, h is a constant, usually taking values ​​of 1-3, corresponding to different confidence levels, and the specific value is set by the user (according to actual conditions); according to the setting method of the matching threshold W here, the setting method of the evaluation threshold Q in S4 can also be set in this way.

[0096] Example 5

[0097] Please refer to Figure 1 , specifically: S3 specifically includes:

[0098] S31. Based on the target users screened out in S2, count the users who actively consulted the target position in the historical period, mark the users as active users, and generate an active user set A;

[0099] S32, real-time monitoring and recording of the activeness of active users and target users on the recruitment platform for various positions, so as to obtain a dynamic data set, wherein the dynamic data set includes the types of positions actively inquired by active users, the types of positions actively inquired by target users, and the number of views Lgcs of the positions actively inquired by active users;

[0100] S321. Construct a set of types of positions actively inquired by active users based on the types of positions actively inquired by active users , based on the types of positions that the target users actively ask about, build a set of types of positions that the target users actively ask about .

[0101] S3 also includes:

[0102] S33: Based on the dynamic data set, the similarity Xsd between the target user and the active user is analyzed, which is specifically obtained by the following method:

[0103] ;

[0104] In the formula, It represents the set of job types that the target user actively inquires about. It represents the set of job types that active users actively inquire about. is represented as the intersection of two sets, Represented as the union of two sets;

[0105] It should be noted that although the target user has browsed or rated some positions, he has not browsed or rated all positions, and even if the target user has browsed or rated some positions, these ratings may not be sufficient to reflect his potential interest in other positions in many aspects; therefore, introducing the ratings of other similar users at this time can help further improve the understanding of the target user's interests. Here, it is assumed that the target user has already rated several positions, and the system finds through calculation that there is a similarity in interest between the target user and the active user. Then the active user's ratings on certain positions can provide a new perspective to help the system predict that the target user may have similar evaluations of these positions.

[0106] In addition, the rating behavior of a single user may be affected by many factors, such as mood, short-term demand changes, etc., which may not be long-term interests. Therefore, making predictions based solely on the rating behavior of target users may have certain noise. By combining the ratings of active users, the system can smooth out the noise through "collective wisdom" and obtain more accurate recommendations.

[0107] A threshold can also be set here. If the similarity Xsd exceeds the threshold, it is judged that there is an obvious similarity between the active user and the target user. If the similarity Xsd does not exceed the threshold, it is judged that there is no obvious similarity between the active user and the target user. Active users with obvious similarities are used to predict the target user's satisfaction coefficient for various types of positions.

[0108] S34. According to the activeness of active users on the recruitment platform for various positions, the number of views Lgcs and the proportion of views Lzb are linearly normalized, and the corresponding data values ​​are mapped to the interval To construct the satisfaction coefficient of active users for various types of positions , which can be obtained by:

[0109] ;

[0110] Where n is the number of active users browsing the same type of posts, j=1, 2, 3, ..., n, It is represented by the number of views of the j-th type of job, Expressed as the average number of views, It is expressed as the browsing proportion of the j-th type of job, Expressed as the average browsing share, and are weight values, where 0 < ≤1,0< ≤1, and + =1, Expressed as a correction constant.

[0111] In this embodiment, through statistics and real-time monitoring based on historical data, users who actively consult the target position can be further accurately marked as active users, and an active user set A is generated. This marking mechanism helps to perform more detailed user classification on the recruitment platform, thereby providing reliable basic data for subsequent analysis and recommendation. By recording the activity of active users and target users in real time, a dynamic data set is generated. The construction of the dynamic data set enables the system to reflect the real interest and activity of active users in the position in real time, providing real-time and accurate data support for subsequent matching analysis. Based on the dynamic data set, the similarity Xsd between the target user and the active user is analyzed. By calculating the intersection and union of the two user sets, the similarity between users can be quantified. This similarity analysis helps to understand the similarity between the target user and the active user in the position interest, thereby optimizing the matching effect. By linearly normalizing the number of active user browsing and browsing proportion, the satisfaction coefficient of active users for various positions is constructed. This method maps the user's browsing behavior to a unified interval, and can accurately evaluate the user's satisfaction with the position. This evaluation not only considers the number of browsing, but also comprehensively considers the browsing proportion, thereby improving the accuracy and reliability of the satisfaction index. Combining the similarity and satisfaction coefficient between the target user and the active user, the target user's preferences and needs for various positions can be predicted more accurately. This process effectively improves the intelligence and accuracy of the job recommendation system, enabling the recruitment platform to provide job recommendations that are more in line with user interests as much as possible, thereby improving the user's job search experience and recruitment success rate. In summary, this method not only improves the accuracy of job recommendations through dynamic data analysis and similarity calculation, but also optimizes the overall matching efficiency of the recruitment platform, bringing significant benefits to recruiting companies and job seekers, and further improving the shortcomings of the traditional recruitment process.

[0112] Example 6

[0113] Please refer to Figure 1 , specifically: S4 specifically includes:

[0114] S41. Based on active users’ satisfaction with various types of positions , combined with the similarity Xsd between the target user and the active user, and after linear normalization, the target user's satisfaction coefficient for each type of position is predicted , the specific contents are as follows:

[0115] ;

[0116] In the formula, a represents the index of the active user, A represents the set of active users, Indicates that active user a belongs to the active user set, It is expressed as the similarity between the target user and the active user a, It is expressed as the satisfaction coefficient of active user a for the j-th type of job.

[0117] S4 also includes:

[0118] S42. Obtain the target user's satisfaction coefficient for each type of position based on S41 , extract the target user's satisfaction coefficient for the target position;

[0119] S43, pre-set the evaluation threshold Q, and compare and analyze the target user's satisfaction coefficient with the target position with the evaluation threshold Q to screen out the candidates to be interviewed for the target position. The specific contents are as follows:

[0120] If the target user's satisfaction coefficient for the target position exceeds the evaluation threshold Q, the target position will be pushed to the target user through the recruitment platform, and the target user will be pushed to the target position, and the target user will be considered as a candidate for interview;

[0121] If the target user's satisfaction coefficient for the target position does not exceed the evaluation threshold Q, the target position will not be pushed to the target user through the recruitment platform at this time, and the target user will not be pushed to the target position at the same time.

[0122] In this embodiment, the system makes job recommendations more personalized by accurately predicting the target user's satisfaction coefficient for each type of job, which can better match the user's needs and interests, and further improve the user's experience of using the recruitment platform. According to the target user's satisfaction coefficient for the target job and the preset evaluation threshold Q, the system can effectively screen out suitable interview candidates. When the target user's satisfaction coefficient for the target job exceeds the evaluation threshold Q, the system marks the target user as a candidate for interview and pushes the target job to the user. This method ensures that only those candidates who are truly interested in the job and have a high degree of match can enter the interview stage, thereby improving the quality of interview candidates. Through effective screening and accurate recommendation, the recruitment platform can reduce unnecessary interview arrangements and job recommendations, thereby saving recruitment resources and time. This method further avoids the situation of interviewing candidates who are not suitable for the job, reduces the waste of resources in the recruitment process, and improves recruitment efficiency.

[0123] Accurate job recommendations and reasonable candidate screening not only improve the recruitment experience of target users, but also enhance the trust of the platform. When users find that the recommended positions meet their needs, they will increase their satisfaction and loyalty to the platform, thereby prompting users to use the platform for a long time and improving the platform's user stickiness and market competitiveness; this method further realizes the intelligent and adaptive optimization of the recruitment system through comprehensive analysis of user behavior data and satisfaction. In addition, the system can automatically learn user preferences and make adjustments based on actual conditions, so that the recruitment platform can continuously improve the accuracy and adaptability of recommendations. Based on real-time feedback and evaluation results, the recruitment platform can dynamically adjust the recommendation strategy and evaluation criteria. This flexible adjustment mechanism enables the platform to optimize according to market demand and user feedback, ensuring that the recruitment strategy always remains effective and adaptable.

[0124] Example 7

[0125] Please refer to Figure 2 ,Specifically: A service system for accurate push based on big data analysis, including a static data monitoring module, a preliminary matching analysis module, a push analysis module and a locking module;

[0126] The static data monitoring module collects various job requirements and user resumes in advance based on big data technology to obtain relevant demand data information of various jobs and relevant resume data information submitted by each user on the recruitment platform, and generates a static data set;

[0127] The preliminary matching analysis module constructs an uncommunicated job matching factor Gpyz based on a static data set, and preliminarily screens out target users that are suitable for the corresponding type of job, and uses the job corresponding to the target user as the target job;

[0128] The push analysis module counts the users who actively interact in this type of position based on the screened target users to generate an active user set A, and monitors the active users and target users' activeness on the recruitment platform for various positions to obtain a dynamic data set. Based on the dynamic data set, the similarity Xsd between the target users and the active users is analyzed, and the active users' satisfaction coefficient for each type of position is constructed in combination with the active users' activeness on the recruitment platform for various positions. ;

[0129] The locking module is based on the satisfaction coefficient of active users for various types of positions. , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

[0130] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An optimization method for accurate push based on big data analysis, characterized by: The following steps are included: S1. Based on big data technology, various job requirements and user resumes are collected in advance to obtain relevant demand data information of various positions and relevant resume data information submitted by each user on the recruitment platform, and generate a static data set; S2. Based on the static data set, analyze the job matching degree between users and uncommunicated jobs, construct the uncommunicated job matching factor Gpyz, and preliminarily screen out target users that are suitable for the corresponding types of jobs, and use the jobs corresponding to the target users as target jobs; S31. Based on the target users screened out in S2, count the users who actively consulted the target position in the historical period, mark the users as active users, and generate an active user set A; S32, real-time monitoring and recording of the activeness of active users and target users on the recruitment platform for various positions, so as to obtain a dynamic data set, wherein the dynamic data set includes the types of positions actively inquired by active users, the types of positions actively inquired by target users, and the number of views Lgcs of the positions actively inquired by active users; S321. Construct a set of types of positions actively inquired by active users based on the types of positions actively inquired by active users , based on the types of positions that the target users actively ask about, build a set of types of positions that the target users actively ask about ; S33: Based on the dynamic data set, the similarity Xsd between the target user and the active user is analyzed, which is specifically obtained by the following method: ； In the formula, It represents the set of job types that the target user actively inquires about. It represents the set of job types that active users actively inquire about. is represented as the intersection of two sets, Represented as the union of two sets; S34. According to the activeness of active users on the recruitment platform for various positions, the number of views Lgcs and the proportion of views Lzb are linearly normalized, and the corresponding data values ​​are mapped to the interval To construct the satisfaction coefficient of active users for various types of positions , which can be obtained by: ； Where n is the number of active users browsing the same type of posts, j=1, 2, 3, ..., n, It is represented by the number of views of the j-th type of job, Expressed as the average number of views, It is expressed as the browsing proportion of the j-th type of job, Expressed as the average browsing share, and are weight values, Expressed as a correction constant; S4. Satisfaction coefficient of active users for various types of positions , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

2. According to the optimization method of the big data analysis based on the precision push according to claim 1, it is characterized by: S1 specifically includes: S11. Collecting various job requirements and user resumes to obtain relevant demand data information for various jobs and relevant resume data information submitted by each user on the recruitment platform, wherein the relevant demand data information includes job requirement descriptions for various jobs and the total number of all types of job description documents ; The relevant resume data information includes the resume description submitted by the user on the recruitment platform; S12. Preprocess the relevant demand data information of various positions and the relevant resume data information submitted by each user on the recruitment platform to generate a static data set, wherein the preprocessing includes noise removal, missing value filling and data smoothing operations, and the missing value filling methods include mean filling, median filling, interpolation filling and regression filling, and use dimensionless processing technology to scale the information in the static data set so that the information range in the static data set falls between 0 and 1.

3. The optimization method for accurate push based on big data analysis according to claim 2 is characterized in that: S2 specifically includes: S21. Use natural language processing technology to segment the static data set, remove stop words, and extract keywords in the relevant demand data information and the relevant resume data information; S22, based on the keywords in the relevant demand data information and the keywords in the relevant resume data information obtained in S21, respectively obtain the word frequency Cp1 of the keywords in the relevant demand data information, the word frequency Cp2 of the keywords in the relevant resume data information, the inverse document frequency Ndp1 of the keywords in the relevant demand data information in all types of positions, and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of positions; S221, the inverse document frequency Ndp1 of the keywords in the relevant demand data information in all types of positions and the inverse document frequency Ndp2 of the keywords in the relevant resume data information in all types of positions are obtained by the following method: ； In the formula, It is expressed as the inverse document frequency of the ith keyword in all types of positions. Represented as the total number of job description documents of all types, It is represented by the number of times the i-th keyword appears in the job description document.

4. The optimization method for accurate push based on big data analysis according to claim 3 is characterized in that: S2 also includes: S23. Based on the keywords in the relevant demand data information and the keywords in the relevant resume data information obtained in S21, the job matching degree between the user and the uncommunicated job is analyzed to calculate the uncommunicated job matching factor Gpyz, which is specifically obtained by the following formula: ； In the formula, k is a common keyword; K is a set of keywords that appear in the relevant demand data information and the relevant resume data information; It is represented by the frequency of common keyword k in the relevant demand data information; It is expressed as the frequency of common keywords in relevant resume data information. It is expressed as the inverse document frequency of common keywords in all types of positions in the relevant demand data information. It is represented by the inverse document frequency of common keywords in all types of positions in the relevant resume data information. It is represented as the weight vector of common keywords in relevant demand data information, It is represented as the weight vector of common keywords in relevant resume data information, The common keyword k is a keyword that appears in both the relevant demand data information and the relevant resume data information.

5. The optimization method for accurate push based on big data analysis according to claim 1 is characterized in that: S2 also includes: S24. Preset a matching threshold W, and compare and analyze the uncommunicated job matching factor Gpyz with the matching threshold W to preliminarily screen out target users that are suitable for the corresponding type of job. The specific preliminary judgment content is as follows: If the uncommunicated job matching factor Gpyz exceeds the matching threshold W, it is preliminarily determined that the corresponding user resume is compatible with the requirements of the corresponding type of job. At this time, the user and the type of job will be labeled as the target user and target job respectively; If the uncommunicated job matching factor Gpyz does not exceed the matching threshold W, it is preliminarily judged that the corresponding user resume is temporarily unmatched with the requirements of the corresponding type of job. At this time, no labeling operation will be performed on the user and the type of job.

6. The optimization method for accurate push based on big data analysis according to claim 1 is characterized in that: S4 specifically includes: S41. Based on active users’ satisfaction with various types of positions , combined with the similarity Xsd between the target user and the active user, and after linear normalization, the target user's satisfaction coefficient for each type of position is predicted , the specific contents are as follows: ； In the formula, a represents the index of the active user, A represents the set of active users, Indicates that active user a belongs to the active user set, It is expressed as the similarity between the target user and the active user a, It is expressed as the satisfaction coefficient of active user a for the j-th type of job.

7. The optimization method for accurate push based on big data analysis according to claim 6 is characterized in that: S4 also includes: S42. Obtain the target user's satisfaction coefficient for each type of position based on S41 , extract the target user's satisfaction coefficient for the target position; S43, pre-set the evaluation threshold Q, and compare and analyze the target user's satisfaction coefficient with the target position with the evaluation threshold Q to screen out the candidates to be interviewed for the target position. The specific contents are as follows: If the target user's satisfaction coefficient for the target position exceeds the evaluation threshold Q, the target position will be pushed to the target user through the recruitment platform, and the target user will be pushed to the target position, and the target user will be considered as a candidate for interview; If the target user's satisfaction coefficient for the target position does not exceed the evaluation threshold Q, the target position will not be pushed to the target user through the recruitment platform at this time, and the target user will not be pushed to the target position at the same time.

8. A service system for accurate push based on big data analysis, used to implement the optimization method for accurate push based on big data analysis as described in any one of claims 1 to 7, characterized in that: It includes static data monitoring module, preliminary matching analysis module, push analysis module and locking module; The static data monitoring module collects various job requirements and user resumes in advance based on big data technology to obtain relevant demand data information of various jobs and relevant resume data information submitted by each user on the recruitment platform, and generates a static data set; The preliminary matching analysis module constructs an uncommunicated job matching factor Gpyz based on a static data set, and preliminarily screens out target users that are suitable for the corresponding type of job, and uses the job corresponding to the target user as the target job; The push analysis module counts the users who actively interact in this type of position based on the screened target users to generate an active user set A, and monitors the active users and target users' activeness in various positions on the recruitment platform to obtain a dynamic data set. Based on the dynamic data set, the similarity Xsd between the target users and the active users is analyzed, and the active users' satisfaction coefficient for each type of position is constructed in combination with the active users' activeness in various positions on the recruitment platform. ; The locking module is based on the satisfaction coefficient of active users for various types of positions. , predict the target user's satisfaction coefficient for the target position, and set the evaluation threshold Q in advance based on the target position demand. Compare and analyze the target user's satisfaction coefficient for the target position with the evaluation threshold Q to optimize the candidates to be interviewed for the target position.

Images