Academic talent success rule prediction system based on resume analysis method
Through a system based on resume analysis method, using decision tree model and data adjustment technology, the problem of failure to predict the success laws of academic talents in the existing technology is solved, and the scientific prediction of academic talents and dynamic description of career laws is realized, which improves the Prediction accuracy and coverage.
Patent Information
- Application Number
- CN202411152055.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2025-01-10
AI Technical Summary
The existing technology has failed to effectively predict the success laws of academic talents, mainly because of the lack of methods to analyze and predict talent data in large samples. Most of the existing methods are classification, analysis and feature description, and lack of prediction functions.
The system based on resume analysis method is adopted, including database module, prediction module and execution module. By extracting and classifying resume data of talents, multiple decision trees are built, the success time is calculated, and the parameters of the decision tree are adjusted to improve prediction accuracy. Form a prediction model for the success laws of academic talents.
It realizes the prediction of the success laws of academic talents based on large-scale resume data, dynamically describes the growth characteristics of different academic career stages, and scientifically predicts possible successful academic talents in the future, which improves the accuracy and coverage of the prediction.
Smart Images
Figure CN119293305A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data analysis, and in particular to a system for predicting the success rules of academic talents based on a resume analysis method. Background Art
[0002] In the field of higher education research, researchers usually use resume analysis to explore the career trajectories and activity patterns of high-knowledge and high-skill groups such as university teachers, scientists, and engineers. Although resume analysis in the field of higher education research and resume analysis in the field of human resource management have something in common, both analyze the personal background, work and life experience of the research object, but the purpose of using resume analysis in the field of human resource management is to judge the adaptation of individuals to future jobs. In the field of higher education research, resume analysis is mainly used to examine the various explicit "outputs" (such as academic publications, patents, etc.) of the research object and their influencing factors, in order to summarize the development characteristics and group laws of the research object in their career. It can be said that in the field of higher education, the application scenarios of resume analysis are richer. It can not only objectively describe the group characteristics and career laws of the research object, but also achieve longitudinal research goals such as long-term resume tracking, quasi-natural experiment and quasi-causal effect analysis.
[0003] Most of the existing talent data analysis methods are aimed at classifying and analyzing talent data, or implementing talent management activities such as career function assessment and talent resume construction, or describing and analyzing the characteristics of specific talent groups based on small samples. They have not yet reached the stage of predicting talent success.
[0004] Therefore, there is an urgent need to provide a prediction system for the success patterns of academic talents based on resume analysis. Summary of the invention
[0005] The present invention solves the technical problems existing in the prior art and provides a system for predicting the success rules of academic talents based on resume analysis.
[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0007] The academic talent success law prediction system based on resume analysis method comprises a database module, a prediction module and an execution module, wherein the output end of the database module is connected to the input end of the prediction module, and the output end of the execution module is connected to the input end of the database module;
[0008] The database module is used to extract resume data of talents and establish resume templates;
[0009] The prediction module forms multiple decision trees based on the resume data of the talent, and calculates the success time based on the multiple decision trees;
[0010] The execution module obtains the resume data of the talent to be predicted, and the execution module inputs the resume data into the database module for processing, and the processed resume data is input into the prediction module to obtain the success time of the talent to be predicted.
[0011] Furthermore, the prediction module includes a resume feature extraction module, a feature classification module and a calculation module. The resume feature extraction module is used to extract the resume data processed by the database module; the feature classification module obtains the resume data extracted by the resume feature extraction module, and N education stages are set in the feature classification module. The feature classification module classifies the extracted resume data of each talent according to the N education stages to form N decision trees; the calculation module calculates the success time of each talent according to the N decision trees formed for each talent by the feature classification module.
[0012] Furthermore, there are three educational stages set in the feature classification module, namely undergraduate, master and doctoral.
[0013] Furthermore, in the training process of the decision tree in the feature classification module: the best splitting point is selected, and each split subset is recursively split until the stopping condition is reached; the personal information of each talent is used as the input feature of the decision tree in the feature classification module, and the identity identifier of the institution attended by each talent in each educational stage is used as a label, thereby forming 3 decision trees corresponding to each talent.
[0014] Furthermore, the calculation module calculates the success time by the following formula:
[0015]
[0016] In the above formula, Indicates the success time. represents the actual weight of the i-th decision tree, represents the actual deviation of the i-th decision tree, represents the ideal value of weight, Indicates the deviation from the ideal value.
[0017] Furthermore, the ideal value of the weight is calculated by the following formula:
[0018]
[0019] in, represents the ideal value of weight, w i represents the feature weight of the i-th decision tree, and N represents the total number of decision trees.
[0020] Furthermore, the ideal value of the deviation is calculated by the following formula:
[0021]
[0022] in, represents the deviation from the ideal value, b i represents the feature deviation of the i-th decision tree, and N represents the total number of decision trees.
[0023] Furthermore, the calculation module calculates the error value, which is the absolute value of the difference between the actual success time of the talent and the calculated success time. The actual success time of the talent is the time when the talent title is first obtained. A threshold is set in the calculation module. When the absolute value is greater than the threshold, the decision tree established by the feature classification module is corrected; the decision tree is corrected by adjusting the maximum depth, minimum sample split number, and minimum sample node number of the decision tree, specifically:
[0024] max_depth_new=max_depth_old-α×error;
[0025] min_samples_split_new=min_samples_split_old+β×error;
[0026] min_samples_leaf_new=min_samples_leaf_old+γ×error;
[0027] In the above formula, max_depth_new represents the maximum depth after correction, max_depth_old represents the maximum depth before correction, error represents the error value, min_samples_split_new represents the minimum number of sample splits after correction, min_samples_split_old represents the minimum number of sample splits before correction, min_samples_leaf_new represents the minimum number of sample nodes after correction, min_samples_leaf_old represents the minimum number of sample nodes before correction, and α, β, and γ represent three different adjustment coefficients.
[0028] Furthermore, the database module includes a resume extraction module and a template creation module. The resume extraction module is used to extract resume data of talents, and the template creation module is used to create a resume template.
[0029] Furthermore, the resume data includes talent information and achievement information. The talent information includes name, age, gender, years of work experience, major, and schools attended at different stages. The achievement information includes information on scientific research, teaching, professional titles and honors, and talent titles. The resume template is divided into three categories: humanities, social sciences, and natural sciences.
[0030] Compared with the prior art, the present invention has the following beneficial effects:
[0031] (1) The present invention uses resume big data to construct talent portraits based on the personal information and success information of academic talents, and dynamically describes the characteristics of growth and development at different stages of academic careers. It uses machine learning models to extract the growth laws of academic talents and form a prediction model for the success laws of academic talents. It scientifically predicts the academic talents most likely to succeed in the future, uses large-scale resume data rather than small samples to discover the laws of talent resumes, and realizes reasoning with talent feature data. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0033] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention, and all other embodiments obtained by ordinary technicians in the field without making creative work are within the protection scope of the present invention.
[0034] like Figure 1 As shown, the present invention provides an academic talent success law prediction system based on resume analysis method, including a database module, a prediction module and an execution module, the output end of the database module is connected to the input end of the training data set module, the output end of the training data set module is connected to the input end of the prediction module, and the output end of the prediction module is connected to the input end of the execution module.
[0035] The database module includes a resume extraction module, a template creation module and a resume encoding module. The output end of the resume extraction module is connected to the input end of the resume template module, and the output end of the resume template module is connected to the input end of the resume encoding module.
[0036] The resume extraction module is used to extract the resume data of talents, and obtain the resume data of the corresponding talents at each stage according to the talents' working years. The resume data of the talents at each stage includes personal information and performance information. The personal information includes name, age, gender, years of work, major, and schools attended at each stage. The performance information includes information on the talents in scientific research, teaching, professional titles and honors, and talent titles.
[0037] The template establishment module is used to establish a resume template based on the resume data extracted by the resume extraction module. Specifically, based on the resume data of the acquired talents, the acquired talents are classified according to humanities, social sciences and natural sciences to form three sample sets, namely, the humanities sample set, the social science sample set and the natural science sample set. Each sample set includes multiple samples, and each sample includes the personal information and performance information of a talent.
[0038] The resume encoding module is used to encode the data in the established template, standardize the data in the established template, and process it using existing standardization methods. The specific methods will not be repeated here.
[0039] The prediction module includes a resume feature extraction module, a feature classification module and a calculation module. The input end of the resume feature extraction module is connected to the output end of the resume encoding module, the output end of the resume feature extraction module is connected to the input end of the feature classification module, and the output end of the feature classification module is connected to the input end of the calculation module.
[0040] The resume feature extraction module is used to extract personal information and performance information from the samples encoded by the resume encoding module.
[0041] In the feature classification module, N education stages are set, and N is preferably 3, that is, 3 education stages are set, namely undergraduate, master and doctoral degrees. For each talent, the personal information and academic performance information extracted by the resume feature extraction module are classified according to the 3 education stages. For the information of each education stage, the personal information (such as gender, age, major, the level of the school attended in the previous stage, etc.) is used as the input feature, and the identity identifier of the school attended in the education stage is used as the label. The identity identifiers include 985 schools, 211 schools, and ordinary schools. Then, the decision tree algorithm is used for training.
[0042] The training process of the decision tree includes: 1) selecting the best split point, 2) recursively splitting each subset after the split until the stopping condition (such as the maximum depth or the minimum number of nodes) is reached. The decision trees corresponding to each stage are formed, and each person forms 3 decision trees.
[0043] In the calculation module, based on the decision tree established by the feature classification module, the ideal weight value and ideal deviation value of different characteristics of each talent in the three education stages are calculated, and the success time of the talent is calculated based on the ideal weight value and ideal deviation value.
[0044] Specifically, the ideal weight value and the ideal deviation value are calculated by the following formula:
[0045]
[0046] in, represents the ideal value of weight, w i represents the feature weight of the i-th decision tree, represents the deviation from the ideal value, b i represents the characteristic deviation of the i-th decision tree, N represents the total number of decision trees, and N is 3.
[0047] The success time is calculated by the following formula:
[0048]
[0049] In the above formula, Indicates the success time, w i represents the actual weight of the i-th decision tree, b i Represents the actual deviation of the i-th decision tree.
[0050] The calculation module also calculates the error value, which is the absolute value of the difference between the actual success time of the talent and the calculated success time. The actual success time of the talent is the time when the talent first obtains the title of Changjiang or Outstanding Youth. When the absolute value is greater than the threshold, the decision tree established by the feature classification module is corrected.
[0051] The decision tree is modified by adjusting the maximum depth and minimum number of samples of the decision tree to prevent overfitting or underfitting.
[0052] Maximum depth: The larger the error value, the smaller the maximum depth is, to prevent overfitting. The smaller the error value is, the larger the maximum depth is, to prevent underfitting. Minimum number of samples: The larger the error value is, the larger the minimum number of samples is, to prevent overfitting. The smaller the error value is, the smaller the minimum number of samples is, to prevent underfitting. The minimum number of samples includes the minimum number of sample splits and the minimum number of sample nodes. The specific adjustment formula is as follows:
[0053] max_depth_new=max_depth_old-α×error
[0054] min_samples_split_new=min_samples_split_old+β×error
[0055] min_samples_leaf_new=min_samples_leaf_old+γ×error
[0056] In the above formula, max_depth_new represents the maximum depth after correction, max_depth_old represents the maximum depth before correction, error represents the error value, min_samples_split_new represents the minimum number of sample splits after correction, min_samples_split_old represents the minimum number of sample splits before correction, min_samples_leaf_new represents the minimum number of sample nodes after correction, min_samples_leaf_old represents the minimum number of sample nodes before correction, α, β, and γ represent three different adjustment coefficients, which are constants.
[0057] In the execution module, the personal information and performance information of the talent to be predicted is obtained, and the obtained information is input into the database module. After being processed by the database module, the information is input into the prediction module, and the success time of the talent is obtained through the prediction module.
[0058] The present invention uses resume big data to construct talent portraits based on the personal information and success information of academic talents, and dynamically describes the characteristics of growth and development at different stages of academic career. It uses machine learning models to extract the growth laws of academic talents, form a prediction model for the success laws of academic talents, and scientifically predict the academic talents most likely to succeed in the future. It uses large-scale resume data rather than small samples to discover the laws of talent resumes and realize reasoning with talent characteristic data.
[0059] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.
Claims
1. The academic talent success law prediction system based on resume analysis method is characterized by: It includes a database module, a prediction module and an execution module, wherein the output end of the database module is connected to the input end of the prediction module, and the output end of the execution module is connected to the input end of the database module; The database module is used to extract resume data of talents and establish resume templates; The prediction module forms multiple decision trees based on the resume data of the talent, and calculates the success time based on the multiple decision trees; The execution module obtains the resume data of the talent to be predicted, and the execution module inputs the resume data into the database module for processing, and the processed resume data is input into the prediction module to obtain the success time of the talent to be predicted.
2. The academic talent success rule prediction system based on resume analysis method according to claim 1 is characterized in that: The prediction module includes a resume feature extraction module, a feature classification module and a calculation module. The resume feature extraction module is used to extract the resume data processed by the database module; the feature classification module obtains the resume data extracted by the resume feature extraction module, and N education stages are set in the feature classification module. The feature classification module classifies the extracted resume data of each talent according to the N education stages to form N decision trees; the calculation module calculates the success time of each talent based on the N decision trees formed for each talent by the feature classification module.
3. The academic talent success rule prediction system based on resume analysis method according to claim 2 is characterized in that: There are three educational stages set in the feature classification module, namely undergraduate, master and doctoral.
4. The academic talent success rule prediction system based on resume analysis method according to claim 3 is characterized in that: In the training process of the decision tree in the feature classification module: selecting the best split point, and recursively continuing to split each split subset until a stop condition is reached; In the feature classification module, the personal information of each talent is used as the input feature of the decision tree, and the identity identifier of the school attended by each talent in each educational stage is used as a label, thereby forming three decision trees corresponding to each talent.
5. The academic talent success rule prediction system based on resume analysis method according to claim 2 is characterized in that: The calculation module uses the following formula to calculate the success time: In the above formula, Indicates the success time. represents the actual weight of the i-th decision tree, represents the actual deviation of the i-th decision tree, represents the ideal value of weight, Indicates the deviation from the ideal value.
6. The academic talent success rule prediction system based on resume analysis method according to claim 5 is characterized in that: The ideal value of the weight is calculated by the following formula: in, represents the ideal value of weight, w i represents the feature weight of the i-th decision tree, and N represents the total number of decision trees.
7. The academic talent success rule prediction system based on resume analysis method according to claim 5 is characterized in that: The ideal value of the deviation is calculated by the following formula: in, represents the deviation from the ideal value, b i represents the feature deviation of the i-th decision tree, and N represents the total number of decision trees.
8. The academic talent success rule prediction system based on resume analysis method according to claim 5 is characterized in that: The calculation module calculates the error value, which is the absolute value of the difference between the actual success time of the talent and the calculated success time. The actual success time of the talent is the time when the talent title is first obtained. A threshold is set in the calculation module. When the absolute value is greater than the threshold, the decision tree established by the feature classification module is corrected; the decision tree is corrected by adjusting the maximum depth of the decision tree, the minimum number of sample splits, and the minimum number of sample nodes. Specifically: max_depth_new=max_depth_old-α×error; min_samples_split_new=min_samples_split_old+β×error; min_samples_leaf_new=min_samples_leaf_old+γ×error; In the above formula, max_depth_new represents the maximum depth after correction, max_depth_old represents the maximum depth before correction, error represents the error value, min_samples_split_new represents the minimum number of sample splits after correction, min_samples_split_old represents the minimum number of sample splits before correction, min_samples_leaf_new represents the minimum number of sample nodes after correction, min_samples_leaf_old represents the minimum number of sample nodes before correction, and α, β, and γ represent three different adjustment coefficients.
9. The academic talent success rule prediction system based on resume analysis method according to claim 1 is characterized in that: The database module includes a resume extraction module and a template creation module. The resume extraction module is used to extract resume data of talents, and the template creation module is used to create resume templates.
10. The academic talent success rule prediction system based on resume analysis method according to claim 9 is characterized in that: The resume data includes talent information and achievement information. The talent information includes name, age, gender, years of work experience, major, and schools attended at different stages. The achievement information includes information on scientific research, teaching, professional titles and honors, and talent titles. The resume template is divided into three categories: humanities, social sciences, and natural sciences.