Occupational planning system and method based on big data

Abstract

The invention discloses an occupational planning system and method based on big data, and relates to the technical field of occupational planning, and the system comprises an interest score module which collects user data to construct a user file, quantifies occupational data, calculates an occupational direction selection probability through a multi-arm machine algorithm, and calculates a comprehensive dominant interest score; and the occupational direction sorting module is used for acquiring occupational related behavior data of the user based on the social media data of the user and dynamically calculating the implicit interest score. According to the method, the comprehensive dominant interest score is the weighted sum of dominant interest scores in multiple occupational directions, the overall interest preference of the user in all occupational directions at present is reflected, the implicit interest score of the user in each occupational direction can be dynamically adjusted, the frequency and the concentration ratio of user behaviors are considered in cooperation with the TF-IDF algorithm, and the user experience is improved. Through the combination of dominant and implicit interest scores, the system can comprehensively evaluate the interests of the user in each occupational direction and sort the interests.

Description

Technical Field

[0001] The present invention relates to the technical field of career planning, and in particular to a career planning system and method based on big data. Background Art

[0002] Traditional career planning usually involves simple career planning based on individual users’ academic progress, certificates, and work experience, and through market recruitment information. As users pay more attention to career planning, they will also rely on manual consultation to obtain more external market career information, thereby adjusting their own career planning. This leads to a strong demand for large amounts of data in career planning. Recommendation results based on big data can reduce the limitations of subjective judgment. In recent years, career planning systems have gradually introduced big data analysis and algorithm models to recommend career directions by analyzing user resumes, work experience, educational background, and other information.

[0003] However, existing career planning methods based on big data usually adopt static data analysis, which easily ignores the dynamic changes of user interests and market demand. At the same time, when processing user career interest scores and career path predictions, the existing systems lack in-depth mining of users' implicit interests, resulting in inaccurate recommended career directions. In addition, career path planning lacks an effective dynamic adjustment mechanism and cannot fully meet users' needs for personalized career development. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing career planning systems and methods based on big data, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that the existing career planning methods based on big data usually adopt static data analysis, which easily ignores the dynamic changes of user interests and market demand. At the same time, when processing user career interest scores and career path predictions, the existing systems lack in-depth mining of users' implicit interests, resulting in the recommended career directions being not accurate enough, and the career path planning lacks an effective dynamic adjustment mechanism, which cannot fully meet the users' needs for personalized career development.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a career planning system based on big data, which includes:

[0007] The interest score module collects user data to build user profiles, quantifies career data and uses a multi-armed bandit algorithm to calculate the probability of career selection, and calculates a comprehensive explicit interest score;

[0008] The career direction ranking module obtains the user's career-related behavior data based on the user's social media data, dynamically calculates the implicit interest score, and ranks the career directions according to the comprehensive explicit interest score increment between the career directions, and determines the number of career directions in the career direction ranking;

[0009] The graph neural network module identifies the characteristics of specific occupations in the career direction based on the graph neural network and analyzes the transition probability between occupations through the softmax function;

[0010] The career path prediction module determines the career path length based on the information gain of career transformation and determines the optimal career path based on the complete path probability of career path combination.

[0011] As a preferred solution of the career planning system based on big data described in the present invention, wherein: the user data is collected to build a user profile, the career data is quantified using a multi-armed bandit algorithm to calculate the probability of career direction selection, and the comprehensive explicit interest score is calculated, including:

[0012] Obtain the user's resume content, historical work experience and educational background information, and based on the user's professional information, posting content and interest tags on the social media platform;

[0013] Standardize the collected user data and build user profiles;

[0014] Based on user profile information, the number of vocational trainings, the number of vocational certificates, and the learning and training time are quantified, the maximum values ​​of the number of vocational trainings, the number of vocational certificates, and the training time of all users are counted, and the explicit interest score of the user is calculated, which is expressed as:

[0015]

[0016] Where S ei represents the explicit interest score, N ti represents the number of vocational trainings, C ti represents the number of professional certificates, T ti Indicates the learning and training time, max(N ti )、max(C ti ) and max(T ti ) represent the maximum number of trainings participated, the number of certificates obtained, and the maximum working or learning time among all users, which are used for normalization processing;

[0017] Based on the user's career direction, the probability of each career direction is initialized. The multi-armed bandit algorithm is used to randomly select a career direction based on the current selection probability at each time step h and calculate the explicit interest score of the corresponding direction as the calculation reward;

[0018] Based on the calculated reward, the selection probability of this career direction is updated, expressed as:

[0019] P i (t+1)=P i (t)+η·(R i (t)-P i (t));

[0020] Where P i (t+1) represents the probability of selecting career direction i at time step t+1, P i (t) represents the probability of selecting career direction i at time step t, R i (t) represents the calculated reward of career direction i at time step t, and η represents the learning rate

[0021] Based on the update of the selection probability, the weighted sum of the explicit interest scores is calculated according to different career directions to calculate the comprehensive explicit interest score.

[0022] As a preferred solution of the career planning system based on big data described in the present invention, the method of obtaining the user's career-related behavior data based on the user's social media data, dynamically calculating the implicit interest score, and ranking the career directions in combination with the comprehensive explicit interest score includes:

[0023] Based on the API of the integrated social media platform, the interaction data between users and corresponding occupations is retrieved and obtained according to occupation-related keywords, and the collected interaction data is classified according to occupational direction.

[0024] By integrating browser plug-ins and search engine APIs, the user's career-related keywords entered in the search engine and their search times are obtained, the user's search keywords are classified according to career direction, and the search times for each direction are counted;

[0025] Obtain user behavior data on career-related websites through web browsing records or integrated behavior analysis tools, including browsing frequency and dwell time, classify career-related websites browsed by users according to career direction, and count browsing frequency and browsing time for each direction;

[0026] The user's mention frequency is based on the collected interaction data and search times related to the corresponding profession;

[0027] The number of keyword searches related to the corresponding occupations collected by the user is used as the user's search frequency;

[0028] Based on the collected interaction data between users and corresponding professions, the user's interaction frequency is used;

[0029] The TF-IDF algorithm is used in combination with user behavior data to calculate the implicit interest scores of users in different career directions, expressed as:

[0030]

[0031] Where S im,i represents the user's implicit interest score for the i-th career direction, f m,i represents the frequency of users mentioning the i-th career direction, f s,i represents the user's search frequency for the i-th career direction, f n,i represents the user's interaction frequency with the i-th career direction, t b,i represents the total time users spend browsing websites related to the i-th career direction, m represents the total number of different career directions, N represents the total number of users, and d represents the number of users who mention the current career direction;

[0032] Collect user data periodically and dynamically adjust the user's implicit interest score by updating the probability distribution;

[0033] Based on the Gaussian distribution in Bayesian inference, an initial implicit interest score probability distribution is established for each career direction;

[0034] Collect and update users' interaction data, search frequency and behavior data on different career directions in real time;

[0035] Based on Bayesian inference combined with weighted summation, the probability distribution of implicit interest scores in career direction i is updated, and the mean and variance are updated based on the update rule of Gaussian distribution in Bayesian inference;

[0036] The expected value is calculated by weighted average based on the updated probability distribution as the updated implicit interest score;

[0037] The sum of the comprehensive explicit interest score and the updated implicit interest score is calculated as the comprehensive interest score;

[0038] The comprehensive interest scores based on all career directions are sorted from high to low.

[0039] As a preferred solution of the career planning system based on big data of the present invention, wherein: the number of career directions to be sorted according to the increment of the comprehensive explicit interest scores between the career directions is determined, including:

[0040] Based on the ranking results of the comprehensive interest scores, the increment between each two career directions is calculated, and the cumulative increment is calculated;

[0041] Plot the cumulative interest increment C qThe curve that changes with the number of career directions q is used to check the inflection point in the cumulative interest increment curve through the elbow method, and the top K career directions ranked by the comprehensive interest scores are determined as the comprehensive interest career directions.

[0042] As a preferred solution of the career planning system based on big data of the present invention, wherein: the characteristics of specific occupations in career directions are identified based on graph neural networks, and the conversion probability between occupations is analyzed by softmax function, including:

[0043] A career path prediction model is built based on the graph neural network (GNN). Based on the specific occupations included in the top K career directions, the specific occupations in the career directions are used as nodes of the graph neural network (GNN). The node feature vectors are built based on the skill requirement level, market demand, and years of experience of the specific occupations.

[0044] Based on historical data, we collect the conversion data between different occupations in the occupation change records and build an occupation change dataset, including the records of each user transferring from one occupation to another;

[0045] Preprocess the data set and normalize the time series, and count the conversion frequency from occupation i to occupation j and the total conversion times of occupation i in the historical data;

[0046] The conversion probability from occupation i to occupation j is calculated by dividing the conversion frequency by the total number of conversions, and the conversion probability between occupations is used as the edge of the graph neural network GNN;

[0047] Based on the graph neural network, the characteristics of occupational transformation are learned through information exchange between nodes and neighboring nodes, which can be expressed as:

[0048]

[0049] in represents the feature representation of occupation node i in the k+1th layer, represents the feature representation of the occupation node j at the kth layer, σ represents the activation function, b represents the bias term, Ν(i) represents the set of neighbor nodes of node i, deg(i) and deg(j) are the number of connecting edges between node i and node j respectively, and W represents the linear transformation weight matrix of the neighbor node information;

[0050] After k+1 layers of information exchange, the features of occupation node i will contain relevant information about its neighbor's occupations, and the final feature aggregation will be performed through average pooling;

[0051] The final node features will be used as the input for career path prediction. The probability distribution of future career paths is output through the softmax function. The probability of the user's current career node i being transformed to its neighboring career node j is calculated through the inner product of the feature vector, expressed as:

[0052]

[0053] Where P ij represents the conversion probability of the user from the current occupation i to the target occupation j, A feature representation that indicates the user's current occupation, Represents the feature representation of target occupation j.

[0054] As a preferred solution of the career planning system based on big data of the present invention, the information gain based on career transformation is used to determine the career path length, including:

[0055] Based on the conversion probability P of users from current occupation i to target occupation j ij , repeatedly calculate the conversion probability P ij As a starting point, predict the user's conversion probability from career node j to the next career node, and recursively predict the user's possible future career conversion path;

[0056] Based on the conversion probability calculated each time, the information gain brought by occupational conversion is further analyzed through conversion probability and logarithmic calculation, which is expressed as:

[0057]

[0058] Among them, IG(P ij ) represents the information gain of the conversion probability from current occupation i to occupation j;

[0059] Following the repeated calculation of conversion probability, the information gain is calculated synchronously, and the change value of the information gain between every two conversion probabilities is calculated, and the information gain change curve is drawn. The elbow method is used to check the inflection point in the information gain change curve, terminate the repeated calculation of information gain and the repeated calculation of conversion probability, and the number of repeated calculations is used as the number of career conversion paths.

[0060] As a preferred solution of the career planning system based on big data of the present invention, wherein: the determination of the optimal career path according to the complete path probability of the career path combination includes:

[0061] Based on the conversion probability calculated repeatedly for the number of career conversion paths, a combination of career paths is formed;

[0062] Based on the combination of career paths, the product of each conversion probability of the career path combination is taken as the complete path probability;

[0063] Sort by the probability of a complete path for each career path;

[0064] The career path with the highest complete path probability is taken as the user's optimal career path.

[0065] Another object of the present invention is to provide a method for a career planning system based on big data, which comprises:

[0066] Collect users' career-related data, dynamically adjust the probability of career direction selection through the multi-armed bandit algorithm, and calculate the explicit interest score of each career direction;

[0067] Analyze users' career behavior data through methods such as TF-IDF algorithm and dynamically calculate the implicit interest score of each career direction;

[0068] The comprehensive explicit interest score and implicit interest score are combined to rank the career directions, and the number of career directions is determined according to the increment of the comprehensive explicit interest score;

[0069] Identify the specific career characteristics in each career direction and learn the conversion probability between career nodes through graph neural networks;

[0070] Based on the probability of career transition, the user's possible future career path is predicted, the path length is determined using information gain, and the optimal career path is calculated by accumulating probabilities.

[0071] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned method of a career planning system based on big data when executing the computer program.

[0072] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method of a career planning system based on big data.

[0073] The beneficial effects of the present invention are as follows: the comprehensive explicit interest score is the weighted sum of the explicit interest scores of multiple career directions, which reflects the user's current overall interest preference for all career directions, and can dynamically adjust the user's implicit interest score for each career direction. The frequency and concentration of user behavior are considered in conjunction with the TF-IDF algorithm. Through the combination of explicit and implicit interest scores, the system can comprehensively and comprehensively evaluate the user's interest in each career direction and rank them. The top K career directions are automatically determined by calculating the interest increment and using the elbow method. Through multi-layer feature updates, the career node contains information about its neighboring careers, so that the career transformation prediction can more deeply reflect the complex career transformation path. By taking the product of each transformation probability of each career transformation path as the probability of the entire path and sorting the paths according to the probability, it is ensured that the user chooses the most likely and feasible career path. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0075] Figure 1 This is a structural diagram of the career planning system based on big data.

[0076] Figure 2 This is a flowchart of the career planning method based on big data. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0078] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0079] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0080] Example 1, reference Figure 1 , which is the first embodiment of the present invention, and which provides a career planning system based on big data. The career planning system based on big data includes:

[0081] S1, collect user data to build user profiles, quantify career data, use multi-armed bandit algorithm to calculate the probability of career selection, and calculate the comprehensive explicit interest score;

[0082] Preferably, user data is collected to build a user profile, and the career data is quantified to calculate the probability of career direction selection using a multi-armed bandit algorithm, and a comprehensive explicit interest score is calculated, including:

[0083] Obtain the user's resume content, historical work experience and educational background information, and based on the user's professional information, posting content and interest tags on the social media platform;

[0084] Standardize the collected user data and build user profiles;

[0085] Based on user profile information, the number of vocational trainings, the number of vocational certificates, and the learning and training time are quantified, the maximum values ​​of the number of vocational trainings, the number of vocational certificates, and the training time of all users are counted, and the explicit interest score of the user is calculated, which is expressed as:

[0086]

[0087] Where S ei represents the explicit interest score, N ti represents the number of vocational trainings, C ti represents the number of professional certificates, T ti Indicates the learning and training time, max(N ti )、max(C ti ) and max(T ti ) represent the maximum number of trainings participated, the number of certificates obtained, and the maximum working or learning time among all users, which are used for normalization processing;

[0088] Through normalization, these data items are unified into the range of 0 to 1, eliminating the calculation bias caused by dimensional differences, so that different data items can be directly added and compared on the same scale. The calculated explicit interest scores can also be calculated separately based on different career directions according to user needs. For example, for technical occupations, the number of training sessions, the number of certificates, and the learning and training time of technical occupations are counted to achieve segmentation according to different career directions, which can more accurately assess the user's interest in each career direction.

[0089] Based on the user's career direction, the probability of each career direction is initialized. The multi-armed bandit algorithm is used to randomly select a career direction based on the current selection probability at each time step h and calculate the explicit interest score of the corresponding direction as the calculation reward;

[0090] Based on the calculated reward, the selection probability of this career direction is updated, expressed as:

[0091] P i (t+1)=P i (t)+η·(R i (t)-P i (t));

[0092] Where P i (t+1) represents the probability of selecting career direction i at time step t+1, P i (t) represents the probability of selecting career direction i at time step t, R i(t) represents the calculated reward of career direction i at time step t, η represents the learning rate, which can be preset according to the needs of the system and is a small positive number, for example, set to a fixed value such as 0.1 to ensure the stability of the update process;

[0093] Based on the update of the selection probability, the weighted sum of the explicit interest scores is calculated according to different career directions to calculate the comprehensive explicit interest score.

[0094] Through the multi-armed bandit algorithm, the system can dynamically adjust the explicit interest scores of different career directions in the process of continuous exploration and utilization, and finally synthesize a comprehensive explicit interest score, so that the selection probability can be dynamically adjusted as the user behavior data changes, reflecting the system's preference for different career directions;

[0095] By constructing the user's career profile and quantifying the number of career trainings, the number of career certificates and the learning time based on the user's resume, historical work experience, educational background, and content posted on social media platforms, the system can comprehensively and dynamically quantify the user's career interests. Since the data is normalized, the system can ensure that different data items are calculated on the same scale, reducing calculation bias and making the calculation of explicit interest scores more scientific. Through the quantitative evaluation of career training and learning time, it can provide users with quantifiable career participation and provide data support for career decision-making. The multi-armed bandit algorithm enables the system to adjust the probability of career selection in the process of continuously exploring the user's interests. The system can balance exploration and utilization to try career directions with greater uncertainty. At the same time, it increases attention to high-yield career directions and improves the intelligence of path planning. As user behavior data is continuously updated, the system can dynamically adjust career direction choices to ensure that the recommendation results are more in line with user interests and career development needs. The comprehensive explicit interest score is the weighted sum of the explicit interest scores of multiple career directions, reflecting the user's current overall interest preference for all career directions, which can help users make a more comprehensive career direction assessment. Through the dynamic update of the comprehensive interest score, the system can provide users with accurate career choice suggestions to help users better find career development directions that suit their interests. Through the continuously updated career direction selection probability, the system can timely optimize the career path according to changes in user interests to ensure that users' career development directions can reflect their actual interests at any time.

[0096] S2, based on the user's social media data, obtain the user's career-related behavior data, dynamically calculate the implicit interest score, and rank the career directions according to the comprehensive explicit interest score increment between the career directions, and determine the number of career directions in the career direction ranking;

[0097] Preferably, the career-related behavior data of the user is obtained based on the user's social media data, the implicit interest score is dynamically calculated, and the career direction is ranked according to the comprehensive explicit interest score, including:

[0098] Based on the API of the integrated social media platform, the interaction data between users and corresponding occupations is retrieved and obtained according to occupation-related keywords, including the number of interaction behaviors such as likes, comments, and shares, and the collected interaction data is classified according to occupational direction.

[0099] By integrating browser plug-ins and search engine APIs, the user's career-related keywords entered in the search engine and their search times are obtained, the user's search keywords are classified according to career direction, and the search times for each direction are counted;

[0100] Obtain user behavior data on career-related websites through web browsing records or the integrated behavior analysis tool Mixpanel, including browsing frequency and dwell time. Categorize career-related websites browsed by users according to career direction, and count browsing frequency and browsing time for each direction.

[0101] The user's mention frequency is based on the collected interaction data and search times related to the corresponding profession;

[0102] The number of keyword searches related to the corresponding occupations collected by the user is used as the user's search frequency;

[0103] Based on the collected interaction data between users and corresponding professions, the user's interaction frequency is used;

[0104] The TF-IDF algorithm is used in combination with user behavior data to calculate the implicit interest scores of users in different career directions, expressed as:

[0105]

[0106] Where S im,i represents the user's implicit interest score for the i-th career direction, f m,i represents the frequency of users mentioning the i-th career direction, f s,i represents the user's search frequency for the i-th career direction, f n,i represents the user's interaction frequency with the i-th career direction, t b,i represents the total time users spend browsing websites related to the i-th career direction, m represents the total number of different career directions, N represents the total number of users, and d represents the number of users who mention the current career direction;

[0107] The implicit interest score calculated by the above formula based on the TF-IDF algorithm can effectively measure the relative importance of career directions and the concentration of user interests. This algorithm can avoid the problem of some high-frequency career directions being over-counted among all users, making the implicit interest score of each career direction more representative. By considering the mention frequency, search frequency, interaction frequency and browsing time, it can fully reflect the user's implicit interest in each career direction.

[0108] Collect user data periodically and dynamically adjust the user's implicit interest score by updating the probability distribution;

[0109] Based on the Gaussian distribution in Bayesian inference, an initial implicit interest score probability distribution is established for each career direction, expressed as:

[0110]

[0111] in represents the probability distribution of the user's implicit interest score in career direction i based on the initial mean and variance, σ0 represents the initial implicit interest score mean in career direction i, and μ0 represents the variance of the implicit interest score in career direction i;

[0112] Collect and update users' interaction data, search frequency and behavior data on different career directions in real time;

[0113] Based on Bayesian inference combined with weighted summation, the probability distribution of implicit interest scores in career direction i is updated, and the mean and variance are updated based on the update rule of Gaussian distribution in Bayesian inference, which is expressed as:

[0114]

[0115] Where P(S im,i |data) represents the updated probability distribution of the user’s implicit interest score in career direction i after updating data, P(data|S im,i ) represents the probability of observing the current data given the implicit interest score, μ n and σ n represents the mean and variance at time n respectively, and date represents the new user behavior data currently obtained, including the user's recent interaction data, search times and behavior data in a certain career direction;

[0116] The expected value is calculated based on the weighted average of the updated probability distribution as the updated implicit interest score, which is expressed as:

[0117]

[0118] Among them, Ε(S im,i) represents the weighted average of all possible implicit interest scores multiplied by their corresponding probabilities, S ip,i represents the updated implicit interest score of the i-th career direction;

[0119] The sum of the comprehensive explicit interest score and the updated implicit interest score is calculated as the comprehensive interest score;

[0120] The comprehensive interest scores based on all career directions are sorted from high to low.

[0121] By integrating multiple sources such as user interaction data on social media platforms, keyword search behavior, web browsing data, etc., the system can collect users' career-related behavior data in multiple dimensions, and then use the TF-IDF algorithm to combine these data to quantify implicit career interest scores, which can comprehensively evaluate users' interests. By collecting data from different dimensions, the system can more accurately explore users' interest levels in various career directions, including explicit interests and implicit interests. The TF-IDF algorithm takes into account the frequency and concentration of user behaviors, effectively avoiding over-calculation of high-frequency career directions, thereby more objectively reflecting the user's interest intensity in a certain direction. By periodically collecting users' interaction data, search times, and browsing data, combined with the probability distribution update mechanism, it can dynamically adjust users' preferences for various career directions. Implicit interest scores, and the system also calculates the expected value of interest in career directions based on these updated probability distributions. It calculates the user's interest by using the implicit interest score probability distribution based on the initial mean and variance, and adjusts the mean and variance in combination with the updated user behavior data, thereby dynamically adjusting the probability of career selection. Through the probability distribution update mechanism, the system can provide a reasonable selection probability for each career direction to ensure the rationality and stability of the recommendation results. Through the combination of explicit and implicit interest scores, the system can comprehensively and comprehensively evaluate the user's interest in each career direction and rank them to ensure the accuracy of the recommendation results. Through the sorted comprehensive interest scores, users can clearly see the career direction they are most interested in, and accordingly formulate personal career development plans to improve the matching degree of career paths.

[0122] Furthermore, according to the increment of comprehensive explicit interest scores among the career directions, the number of career directions for career direction ranking is determined, including:

[0123] Based on the ranking results of the comprehensive interest scores, the increment between each two career directions is calculated, and the cumulative increment is calculated, expressed as:

[0124]

[0125] Among them C q represents the cumulative interest increment of the first q career directions, ΔS to,jThe variable representing the comprehensive interest score of the jth career direction;

[0126] Plot the cumulative interest increment C q The curve that changes with the number of career directions q is used to check the inflection point in the cumulative interest increment curve through the elbow method, and the top K career directions ranked by the comprehensive interest scores are determined as the comprehensive interest career directions.

[0127] Through incremental calculation and cumulative curves, the system can identify the inflection point where the user's interest in certain career directions suddenly decreases, thereby optimizing the recommendation results and ensuring that the recommended career directions are the directions where the user's interest is highly concentrated. Through the elbow method, the system can effectively reduce redundant career directions in the career direction sorting, avoid recommending too many directions that have little impact on users, and improve the recommendation efficiency. By dynamically adjusting the number of career directions as the user's behavior data changes, rather than being limited to a fixed number, the system can automatically adjust the number of recommended career directions according to the interest curves of different users. Through the cumulative interest increment curve, the system can clearly show the trend of user interest changes in various career directions, so that the recommendation results are not only based on a single score sorting, but through comprehensive interest increment analysis, ensuring that the recommendation results are more reasonable. By calculating the interest increment and using the elbow method to automatically determine the top K career directions, the system can effectively improve the decision-making efficiency of users when making career plans. Users do not need to manually evaluate the interest level of each career direction, and the system can intelligently complete the optimization sorting.

[0128] S3, identifies the characteristics of specific occupations in career directions based on graph neural networks, and analyzes the transition probability between occupations through the softmax function;

[0129] Preferably, the characteristics of specific occupations in the career direction are identified based on the graph neural network, and the conversion probability between occupations is analyzed by the softmax function, including:

[0130] A career path prediction model is constructed based on the graph neural network GNN. Based on the specific occupations included in the first K career directions, the specific occupations in the career directions are used as nodes of the graph neural network GNN. The node feature vector is constructed according to the skill requirement level, market demand (which can be calculated based on the market recruitment share of the occupation in the corresponding career direction) and years of experience of the specific occupation.

[0131] The node feature vector may also be constructed by including the personality and value evaluation of historical work experience and the evaluation of personal image and temperament, so that the nodes of the career path fully consider the evaluation of the user's professional personality, ability, interest, temperament and values ​​by the recruiting company in the career change record;

[0132] Based on historical data, we collect the conversion data between different occupations in the occupation change records and build an occupation change dataset, including the records of each user transferring from one occupation to another;

[0133] Preprocess the data set and normalize the time series, and count the conversion frequency from occupation i to occupation j and the total conversion times of occupation i in the historical data;

[0134] The conversion probability from occupation i to occupation j is calculated by dividing the conversion frequency by the total number of conversions, and the conversion probability between occupations is used as the edge of the graph neural network GNN;

[0135] Based on the graph neural network, the characteristics of occupational transformation are learned through information exchange between nodes and neighboring nodes, which can be expressed as:

[0136]

[0137] in represents the feature representation of occupation node i in the k+1th layer, represents the feature representation of the occupation node j at the kth layer, σ represents the activation function, b represents the bias term, Ν(i) represents the set of neighbor nodes of node i, deg(i) and deg(j) are the number of connecting edges between node i and node j respectively, and W represents the linear transformation weight matrix of the neighbor node information;

[0138] After k+1 layers of information exchange, the features of occupation node i will contain relevant information about its neighbor's occupations, and the final feature aggregation is performed through average pooling, which is expressed as:

[0139]

[0140] in represents the final feature representation after average pooling, Represents the feature representation of the neighbor node j in the last F layer;

[0141] The final node features will be used as the input for career path prediction. The probability distribution of future career paths is output through the softmax function. The probability of the user's current career node i being transformed to its neighboring career node j is calculated through the inner product of the feature vector, expressed as:

[0142]

[0143] Where P ij represents the conversion probability of the user from the current occupation i to the target occupation j, A feature representation that indicates the user's current occupation, Represents the feature representation of target occupation j.

[0144] By constructing a graph neural network based on career nodes, the model can make accurate career path predictions based on the node's skill requirements, market demand, and years of experience, combined with the conversion probability between occupations. The information exchange between nodes and neighboring nodes ensures that the model can capture the complex relationship between occupations, thereby improving the accuracy of career conversion predictions. Through the statistics of historical career conversion data, the model can dynamically adjust the conversion probability between occupations. The conversion probability of career nodes is based on historical records and is obtained through the normalized career conversion frequency. After information exchange, career nodes update their features, so that the model can adapt to changes in market demand and dynamically update career conversion probabilities. Through the exchange of information between multiple layers of nodes, career nodes can not only learn their own information, but also obtain relevant career conversion information from neighboring career nodes. Through multi-layer feature updates, career nodes contain The information of its neighbors' occupations is used to make the career transition prediction more deeply reflect the complex career transition path. The career transition probability is calculated by the inner product of the career node features, and the probability distribution of career transition is output using the softmax function, which improves the prediction efficiency of the model. It can also provide users with a clear reference for career transition probabilities to help users make wise career choices. Through the inner product calculation and the softmax function, the model can efficiently and quickly calculate the probability of a user switching from their current occupation to their target occupation, greatly improving the calculation efficiency of career prediction. The calculation results of the career transition probability can provide users with clear and definite career choice suggestions, reducing the uncertainty in the user's decision-making process. The career node features of each user can be dynamically generated according to personal background and market demand, so that the model can provide personalized career path predictions for different users.

[0145] S4, determine the career path length based on the information gain of career transformation, and determine the optimal career path based on the complete path probability of career path combination;

[0146] Preferably, determining the career path length based on the information gain of career transformation includes:

[0147] Based on the conversion probability P of users from current occupation i to target occupation j ij , repeatedly calculate the conversion probability P ij As a starting point, predict the user's conversion probability from career node j to the next career node, and recursively predict the user's possible future career conversion path;

[0148] Based on the conversion probability calculated each time, the information gain brought by occupational conversion is further analyzed through conversion probability and logarithmic calculation, which is expressed as:

[0149]

[0150] Among them, IG(P ij) represents the information gain of the conversion probability from current occupation i to occupation j;

[0151] When a user switches from occupation A to occupation B, information gain can help us determine whether this conversion is helpful for the overall career path prediction. By taking the logarithm of the conversion probability, we can quantify the amount of information brought by the conversion. If a low-probability event (i.e., an occupation with a low conversion probability) occurs, it will bring a large information gain, while a high-probability event will bring a relatively small information gain.

[0152] Following the repeated calculation of conversion probability, the information gain is calculated synchronously, and the change value of the information gain between every two conversion probabilities is calculated, and the information gain change curve is drawn. The elbow method is used to check the inflection point in the information gain change curve, terminate the repeated calculation of information gain and the repeated calculation of conversion probability, and the number of repeated calculations is used as the number of career conversion paths.

[0153] By predicting the career transformation path and using information gain to calculate the uncertainty of each step of career transformation, the system can intelligently determine the length of the career path. The information gain change curve combined with the elbow method can help the system find the optimal end point of the path length and avoid the problem of predicting the path to be too long or too short. Through the calculation of information gain, the career transformation prediction of each step not only considers the transformation probability, but also measures the size of the information gain brought by the career transformation. The calculation of information gain helps to improve the decision-making quality of the model in career path prediction and ensure that the path prediction is more accurate. By synchronously calculating the changes in career transformation probability and information gain, the length of the transformation path is dynamically adjusted. The calculation of information gain and the synchronous update of transformation probability avoid redundant calculations, making career path prediction more efficient. By drawing the information gain change curve and combining it with the elbow method to find the optimal career path length, users can intuitively see the information gain change process of the system in path selection, helping users understand the decision-making basis of the career transformation path. By dynamically adjusting the length of the career transformation path, the system can adaptively recommend the optimal career path based on the user's actual career status and market changes.

[0154] Furthermore, the optimal career path is determined based on the complete path probability of the career path combination, including:

[0155] Based on the conversion probability calculated repeatedly for the number of career conversion paths, a combination of career paths is formed;

[0156] Based on the combination of career paths, the product of each conversion probability of the career path combination is taken as the complete path probability;

[0157] Sort by the probability of a complete path for each career path;

[0158] The career path with the highest complete path probability is taken as the user's optimal career path.

[0159] By calculating the complete path probability of each career path, the system can effectively select the optimal career development path from multiple paths. By taking the product of the conversion probabilities of each career conversion path as the probability of the entire path and sorting the paths according to the probability, it ensures that the user chooses the most likely and feasible career path. By conducting a combined analysis of multiple career conversion paths, the system can cover the various possibilities of the user's career choices and calculate the complete probability of each possible path. The combination allows the system to comprehensively evaluate each career path and provide more comprehensive career path recommendations. By sorting the probabilities of complete paths, the system can ensure that the user chooses the path with the highest probability of career conversion, thereby accurately reflecting the feasibility and success possibility of the career path, and providing users with data-supported career decision-making basis.

[0160] Example 2, reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides a method for a career planning system based on big data, including:

[0161] Collect users' career-related data, dynamically adjust the probability of career direction selection through the multi-armed bandit algorithm, and calculate the explicit interest score of each career direction;

[0162] Analyze users' career behavior data through methods such as TF-IDF algorithm and dynamically calculate the implicit interest score of each career direction;

[0163] The comprehensive explicit interest score and implicit interest score are combined to rank the career directions, and the number of career directions is determined according to the increment of the comprehensive explicit interest score;

[0164] Identify the specific career characteristics in each career direction and learn the conversion probability between career nodes through graph neural networks;

[0165] Based on the probability of career transition, the user's possible future career path is predicted, the path length is determined using information gain, and the optimal career path is calculated by accumulating probabilities.

[0166] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0167] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0168] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0169] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0170] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A career planning system based on big data, characterized by: include, The interest score module collects user data to build user profiles, quantifies career data and uses a multi-armed bandit algorithm to calculate the probability of career selection, and calculates a comprehensive explicit interest score; The career direction ranking module obtains the user's career-related behavior data based on the user's social media data, dynamically calculates the implicit interest score, and ranks the career directions according to the comprehensive explicit interest score increment between the career directions, and determines the number of career directions in the career direction ranking; The graph neural network module identifies the characteristics of specific occupations in the career direction based on the graph neural network and analyzes the transition probability between occupations through the softmax function; The career path prediction module determines the career path length based on the information gain of career transformation and determines the optimal career path based on the complete path probability of career path combination.

2. The career planning system based on big data as claimed in claim 1, characterized in that: The user data is collected to build a user profile, the career data is quantified using a multi-armed bandit algorithm to calculate the probability of career selection, and the comprehensive explicit interest score is calculated, including: Obtain the user's resume content, historical work experience and educational background information, and based on the user's professional information, posting content and interest tags on the social media platform; Standardize the collected user data and build user profiles; Based on user profile information, the number of vocational trainings, the number of vocational certificates, and the learning and training time are quantified, the maximum values ​​of the number of vocational trainings, the number of vocational certificates, and the training time of all users are counted, and the explicit interest score of the user is calculated, which is expressed as: Where S ei represents the explicit interest score, N ti represents the number of vocational trainings, C ti represents the number of professional certificates, T ti Indicates the learning and training time, max(N ti )、max(C ti ) and max(T ti ) represent the maximum number of trainings participated, the number of certificates obtained, and the maximum working or learning time among all users, which are used for normalization processing; Based on the user's career direction, the probability of each career direction is initialized. The multi-armed bandit algorithm is used to randomly select a career direction based on the current selection probability at each time step h and calculate the explicit interest score of the corresponding direction as the calculation reward; Based on the calculated reward, the selection probability of this career direction is updated, expressed as: P i (t+1)=P i (t)+η·(R i (t)-P i (t)); Where P i (t+1) represents the probability of selecting career direction i at time step t+1, P i (t) represents the probability of selecting career direction i at time step t, R i (t) represents the calculated reward of career direction i at time step t, and η represents the learning rate Based on the update of the selection probability, the weighted sum of the explicit interest scores is calculated according to different career directions to calculate the comprehensive explicit interest score.

3. The career planning system based on big data as claimed in claim 2, characterized in that: The method of obtaining the user's career-related behavior data based on the user's social media data, dynamically calculating the implicit interest score, and ranking the career directions in combination with the comprehensive explicit interest score includes: Based on the API of the integrated social media platform, the interaction data between users and corresponding occupations is retrieved and obtained according to occupation-related keywords, and the collected interaction data is classified according to occupational direction. By integrating browser plug-ins and search engine APIs, the user's career-related keywords entered in the search engine and their search times are obtained, the user's search keywords are classified according to career direction, and the search times for each direction are counted; Obtain user behavior data on career-related websites through web browsing records or integrated behavior analysis tools, including browsing frequency and dwell time, classify career-related websites browsed by users according to career direction, and count browsing frequency and browsing time for each direction; The user's mention frequency is based on the collected interaction data and search times related to the corresponding profession; The number of keyword searches related to the corresponding occupations collected by the user is used as the user's search frequency; Based on the collected interaction data between users and corresponding professions, the user's interaction frequency is used; The TF-IDF algorithm is used in combination with user behavior data to calculate the implicit interest scores of users in different career directions, expressed as: Where S im,i represents the user's implicit interest score for the i-th career direction, f m,i represents the frequency of users mentioning the i-th career direction, f s,i represents the user's search frequency for the i-th career direction, f n,i represents the user's interaction frequency with the i-th career direction, t b,i represents the total time users spend browsing websites related to the i-th career direction, m represents the total number of different career directions, N represents the total number of users, and d represents the number of users who mention the current career direction; Collect user data periodically and dynamically adjust the user's implicit interest score by updating the probability distribution; Based on the Gaussian distribution in Bayesian inference, an initial implicit interest score probability distribution is established for each career direction; Collect and update users' interaction data, search frequency and behavior data on different career directions in real time; Based on Bayesian inference combined with weighted summation, the probability distribution of implicit interest scores in career direction i is updated, and the mean and variance are updated based on the update rule based on Gaussian distribution in Bayesian inference; The expected value is calculated by weighted average based on the updated probability distribution as the updated implicit interest score; The sum of the comprehensive explicit interest score and the updated implicit interest score is calculated as the comprehensive interest score; The comprehensive interest scores based on all career directions are sorted from high to low.

4. The career planning system based on big data as claimed in claim 3, characterized in that: The number of career directions to be ranked is determined according to the increment of the comprehensive explicit interest scores between the career directions. include, Based on the ranking results of the comprehensive interest scores, the increment between each two career directions is calculated, and the cumulative increment is calculated; Plot the cumulative interest increment C q The curve that changes with the number of career directions q is used to check the inflection point in the cumulative interest increment curve through the elbow method, and the top K career directions ranked by the comprehensive interest scores are determined as the comprehensive interest career directions.

5. The career planning system based on big data as claimed in claim 4, characterized in that: The graph neural network is used to identify the characteristics of specific occupations in the occupational direction, and the softmax function is used to analyze the conversion probability between occupations, including: A career path prediction model is built based on the graph neural network (GNN). Based on the specific occupations included in the top K career directions, the specific occupations in the career directions are used as nodes of the graph neural network (GNN). The node feature vectors are built based on the skill requirement level, market demand, and years of experience of the specific occupations. Based on historical data, we collect the conversion data between different occupations in the occupation change records and build an occupation change dataset, including the records of each user transferring from one occupation to another; Preprocess the data set and normalize the time series, and count the conversion frequency from occupation i to occupation j and the total conversion times of occupation i in the historical data; The conversion probability from occupation i to occupation j is calculated by dividing the conversion frequency by the total number of conversions, and the conversion probability between occupations is used as the edge of the graph neural network GNN; Based on the graph neural network, the characteristics of occupational transformation are learned through information exchange between nodes and neighboring nodes, which can be expressed as: in represents the feature representation of occupation node i in the k+1th layer, represents the feature representation of the occupation node j at the kth layer, σ represents the activation function, b represents the bias term, Ν(i) represents the set of neighbor nodes of node i, deg(i) and deg(j) are the number of connecting edges between node i and node j respectively, and W represents the linear transformation weight matrix of the neighbor node information; After k+1 layers of information exchange, the features of occupation node i will contain relevant information about its neighbor's occupations, and the final feature aggregation will be performed through average pooling; The final node features will be used as the input for career path prediction. The probability distribution of future career paths is output through the softmax function. The probability of the user's current career node i being transformed to its neighboring career node j is calculated through the inner product of the feature vector, expressed as: Where P ij represents the conversion probability of the user from the current occupation i to the target occupation j, A feature representation representing the user's current occupation, Represents the feature representation of target occupation j.

6. The career planning system based on big data as claimed in claim 5, characterized in that: The information gain based on career transformation determines the career path length, include, Based on the conversion probability P of users from current occupation i to target occupation j ij , repeatedly calculate the conversion probability P ij As a starting point, predict the user's conversion probability from career node j to the next career node, and recursively predict the user's possible future career conversion path; Based on the conversion probability calculated each time, the information gain brought by occupational conversion is further analyzed through conversion probability and logarithmic calculation, which is expressed as: Among them, IG(P ij ) represents the information gain of the conversion probability from current occupation i to occupation j; Following the repeated calculation of conversion probability, the information gain is calculated synchronously, and the change value of the information gain between every two conversion probabilities is calculated, and the information gain change curve is drawn. The elbow method is used to check the inflection point in the information gain change curve, terminate the repeated calculation of information gain and the repeated calculation of conversion probability, and the number of repeated calculations is used as the number of career conversion paths.

7. The career planning system based on big data as claimed in claim 6, characterized in that: The determining of the optimal career path according to the complete path probability of the career path combination includes: Based on the conversion probability calculated repeatedly for the number of career conversion paths, a combination of career paths is formed; Based on the combination of career paths, the product of each conversion probability of the career path combination is taken as the complete path probability; Sort by the probability of a complete path for each career path; The career path with the highest complete path probability is taken as the user's optimal career path.

8. A method for a career planning system based on big data according to any one of claims 1 to 7, characterized in that: include, Collect users' career-related data, dynamically adjust the probability of career direction selection through the multi-armed bandit algorithm, and calculate the explicit interest score of each career direction; Analyze users' career behavior data through methods such as TF-IDF algorithm and dynamically calculate the implicit interest score of each career direction; The comprehensive explicit interest score and implicit interest score are combined to rank the career directions, and the number of career directions is determined according to the increment of the comprehensive explicit interest score; Identify the specific career characteristics in each career direction and learn the conversion probability between career nodes through graph neural networks; Based on the probability of career transition, the user's possible future career path is predicted, the path length is determined using information gain, and the optimal career path is calculated by accumulating probabilities.

9. A computer device comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the method of the career planning system based on big data described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method of the career planning system based on big data described in any one of claims 1 to 7 are implemented.

Images