A precision guidance system and method for employment intention of college students

Abstract

The application discloses a kind of for college student employment intention accurate guidance system and method, it is related to big data analysis technical field.The present application is based on student job-seeking intention information to establish corresponding employment intention graph, extracts the employment intention graph of each student to generate employment intention vector, and sets up career perception space, the employment intention vector of all students is input into career perception space, according to employment intention vector distribution in career perception space, intention cluster is divided out, the real-time evolution process of each intention cluster is continuously tracked, and intention cluster distribution in career perception space is adjusted synchronously, set up individualized guidance scheme library, individualized guidance scheme library generates individualized guidance scheme according to the real-time distribution of intention cluster distribution, and establishes employment process closed loop tracking mechanism, whenever judge student does not complete employment contract and updates student job-seeking intention information, according to the new individualized guidance scheme of updated student job-seeking intention information generation.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to a precise guidance system and method for college students' employment intentions. Background Technology

[0002] Currently, employment guidance for college students generally relies on traditional methods such as questionnaires, counselors' experience-based judgment, and students' self-consultation. These methods have significant shortcomings: First, the data collection dimensions are singular and lagging, making it difficult to capture the dynamic changes in students' job intentions in real time. For example, key changes in students' intentions, such as shifting from "taking postgraduate entrance exams" to "taking civil service exams" or from "big internet companies" to "state-owned enterprises," are often discovered with delay.

[0003] Secondly, the guidance programs are highly homogenized and lack in-depth adaptation to students' individual career planning timelines, skill requirements, and non-cognitive traits, resulting in resume revision and interview coaching becoming mere formalities.

[0004] Third, there is a lack of systematic spatial modeling and evolutionary analysis of group employment intentions, making it impossible to identify "popular convergence directions" or "differentiation trends" in advance, and difficult to provide forward-looking employment assistance decision support for classes and departments. Although there are employment recommendation systems based on simple tags in existing technologies, they ignore key factors such as the intensity of intention, conversion costs, and time-series attention, and cannot construct a structured and evolvable employment intention map, nor can they track intention clusters in real time and provide anomaly warnings in a dynamic career perception space.

[0005] Therefore, a precise guidance system and method for college students' employment intentions are provided. Summary of the Invention

[0006] The purpose of this invention is to provide a precise guidance system and method for college students' employment intentions, in order to solve the problems in the background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A precise guidance method for college students' employment intentions includes the following steps: Step S1: Obtain students' job-seeking intention information in real time, and establish employment intention tags, career planning timelines and job-seeking demand parameters based on the students' job-seeking intention information, thereby establishing a corresponding employment intention map; Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated.

[0008] Furthermore, the process of obtaining students' job intentions in real time includes: The student job intention information is formed by collecting and integrating students' online job search behavior data, and deploying backend log tracking, frontend behavior capture instructions, and periodic survey questionnaire interfaces within the employment guidance platform; Each student is assigned a unique virtual identity identifier, Stu, and all collected behavioral data is integrated in time series format to generate students' job intention information.

[0009] Furthermore, the process of establishing employment intention tags, career planning timelines, and job demand parameters based on students' job-seeking intention information includes: The system aggregates and assigns weights to students' job-seeking intentions based on time sequence, establishing a job intention tagging system. These tags include primary tags for industry preference, job type, region selection, and salary expectations. Each primary tag has fine-grained secondary tags and quantified weights. Simultaneously, a career planning timeline is established, which includes three scales: short-term goals, medium-term goals, and long-term goals. Job search demand parameters are also established, including the desired company type, internship requirements, and skills enhancement needs.

[0010] Furthermore, the process of establishing an employment intention map includes: Each student is treated as a central node of intent, and a graph timeline is set up in a two-dimensional or three-dimensional visualization space based on the career planning timeline. Multiple intention attribute nodes are radially connected around the intention center node. The intention attribute nodes are divided into three layers: the first layer is direct label nodes, the second layer is associated skill nodes, and the third layer is external constraint nodes. The connection edges between nodes are assigned weights, which are obtained by quantifying the intention intensity. Furthermore, a temporal attention mechanism is introduced to capture the vibrational signals of intention. A sliding time window of length L is set, and all events generated within the sliding time window are regarded as a sequence. For each intention attribute node, a multi-head attention mechanism is used to obtain its cumulative activation count A within the current sliding time window. i(t) A i(t)It is the weighted sum of the contributions of all events to the intention attribute node i within the window before time t; Set a dynamic threshold θ for each intention attribute node i(t) θ i(t) It is the sum of the exponential moving average of the cumulative activation count of intention attribute node i over its history and a baseline deviation. When the cumulative activation intensity A of a certain intention attribute node is... i(t) Continuously exceeding its dynamic threshold θ i(t) When the preset number of confirmations is reached, the core intent update event is triggered.

[0011] Furthermore, the process of extracting the employment intention map of each student to generate an employment intention vector includes: The employment intention map established for each student is vectorized and encoded to generate a fixed-dimensional employment intention vector. The dimensions of the employment intention vector must comprehensively cover the key information of the intention map, including industry, job position, region, salary, and trait preference dimensions. Finally, the employment intention vector for each student is obtained, denoted as V. Stu .

[0012] Furthermore, a career perception space is established, and the employment intention vectors of all students are input into the career perception space. The process of dividing the career intention clusters according to the distribution of the employment intention vectors in the career perception space includes: A career perception space is set up to accommodate the employment intention vectors of all students and perform relation calculations, including a generalized career dimension, a characteristic career dimension, and a corporate career dimension. The vector V of all students' employment intentions Stu Input into the career perception space to obtain the employment intention vector V Stu The employment intention vector V is adjusted based on the cosine similarity of the nodes in each dimension. Stu Spatial location within the professional perception space; Within the occupational perception space, several vectors V, each representing a different employment intention, are randomly divided. Stu The intention clusters are formed, and then, with each intention cluster as the central intention cluster, the employment intention vectors V outside the central intention clusters are obtained using the Euclidean distance formula. Stu The distance to the edge of the central intended cluster, and a threshold for the mean distance is set; Based on the principle of closest distance, each employment intention vector V is... Stu Assign to the intention cluster with the smallest distance, and then obtain all employment intention vectors V within the intention cluster. Stu Calculate the mean distance between them, and determine the employment intention vectors V within the current intention cluster. Stu Whether the mean distance between them is less than or equal to the mean distance threshold, several intention clusters are divided based on the judgment result.

[0013] Furthermore, the process of continuously tracking the real-time evolution of each intention cluster and simultaneously adjusting the distribution of intention clusters in the career perception space includes: Set a state update period t. After each state update period t, the employment intention vectors of all students may undergo a slight shift, which will cause changes in the shape, position and membership relationship of the intention clusters. By comparing the distribution of intention clusters in two consecutive state update periods t and t+1, we can track their evolution process. The evolution process is tracked by analyzing the changes in the inter-cluster distance matrix and the variance of the average displacement vector of intra-cluster members, and the occupational perception space is dynamically adjusted based on the tracking results.

[0014] Furthermore, a personalized guidance scheme library is set up. The process of generating personalized guidance schemes based on the real-time distribution of intention clusters includes: The personalized guidance solution library consists of three components: resume optimization templates, interview simulation scripts, and employment policy interpretation packages. Furthermore, after each state update cycle t, the personalized guidance solution library first retrieves the employment intention vector V at the center position of each intention cluster. Stu Generate corresponding common guidance plans, and then combine these common guidance plans with each student's employment intention vector V. Stu Generate personalized guidance plans.

[0015] Furthermore, a closed-loop tracking mechanism for the employment process is established. Whenever it is determined that a student has not completed an employment contract and their job-seeking intention information is updated, a new personalized guidance plan is generated based on the updated information. The process of generating or updating the class employment statistics report whenever it is determined that a student has completed an employment contract or their job-seeking intention information is updated includes: The closed-loop tracking mechanism for the employment process includes determining whether an employment contract has been signed, and if not, further determining whether there has been an update on the student's job search intention information. If the completion of the employment contract signing is determined through the three-party agreement upload interface of the academic affairs system or by a flag manually confirmed by the counselor, then the corresponding student will be moved to the "employed - final" state, and the generation of new personalized guidance plans will be stopped, and the corresponding employment intention vector V will be removed. Stu ; If it is determined that an employment contract has not been signed, the next step is to determine whether there has been an update to the student's job intention information. If any core intention tag in the job intention map changes, or the cumulative activation intensity exceeds a threshold leading to a core intention update, or if more than one hard cycle has passed since the last update, then the student's job intention information is considered to have been updated, and a new job intention vector V is generated. StuThe distribution of intention clusters is updated synchronously, and new personalized guidance plans are generated. The process of generating new personalized guidance plans is repeated until the corresponding student is judged to have completed the employment contract signing process. In addition, whenever it is determined that a student has completed an employment contract or updated their job search intention information, a class employment statistics report is generated or updated.

[0016] A precise guidance system for college students' employment intentions includes a student employment guidance module and a teacher tutoring module; The student employment guidance module is used to obtain students' job intention information in real time and receive personalized guidance plans from the teacher tutoring module. Based on students' job-seeking intention information, we establish employment intention tags, career planning timelines, and job-seeking demand parameters, and then build a corresponding employment intention map. Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated.

[0017] The teacher guidance module includes an intention analysis unit and an employment report unit; The intention analysis unit is used to establish employment intention tags, career planning timelines, and job demand parameters based on students' job-seeking intention information, thereby establishing a corresponding employment intention map, extracting employment intention vectors from each student's employment intention map, and setting up a career perception space. The employment intention vectors of all students are input into the career perception space, and intention clusters are divided in the career perception space according to the distribution of employment intention vectors. The real-time evolution process of each intention cluster is continuously tracked, and the distribution of intention clusters in the career perception space is adjusted synchronously. A personalized guidance scheme library is set up. The personalized guidance scheme library generates personalized guidance schemes based on the real-time distribution of intention clusters, and a closed-loop tracking mechanism for the employment process is established. Whenever it is determined that a student has not completed an employment contract and the student's job-seeking intention information is updated, a new personalized guidance scheme is generated based on the updated student's job-seeking intention information. The employment reporting unit is used to generate or update the class employment statistics report whenever it is determined that a student has completed an employment contract or that a student's job search intention information has been updated.

[0018] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs an employment intention map that includes a temporal attention mechanism, which can capture the vibration signals of students' intentions in real time and automatically update the core intention tags, realizing high-frequency mapping from fragmented behavioral data to a structured intention model, thereby improving the response speed and personalized accuracy of employment guidance.

[0019] 2. By setting up a multi-layered career perception space and continuously tracking the evolution of intention clusters (such as popular convergence directions, differentiation trends, etc.), this invention can identify the employment dynamics of groups in advance and generate class employment statistics reports and hierarchical classification lists of unemployed students, providing data-driven precise assistance decision support for universities. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0021] Figure 1 This is a flowchart of the method of the present invention.

[0022] Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 As shown, a precise guidance method for college students' employment intentions includes the following steps: Step S1: Obtain students' job-seeking intention information in real time, and establish employment intention tags, career planning timelines and job-seeking demand parameters based on the students' job-seeking intention information, thereby establishing a corresponding employment intention map; Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated.

[0025] Furthermore, step S1 is implemented through the following process: Step S101: Obtain students' job intention information in real time. The specific process includes: The students' job-seeking intention information is formed by integrating online job-seeking behavior data collected from all dimensions of the employment guidance platform's backend logs. The employment guidance platform deploys backend log tracking, frontend behavior capture commands, and periodic survey questionnaire interfaces. The backend log tracking is used to silently collect students' active search behavior on data sources such as official employment websites, academic affairs systems, and cooperative recruitment platforms, including but not limited to search keywords (such as "Java development" and "product manager"), search frequency, search timestamps, and search result click-through rates. Front-end behavior capture instructions are used to record students' behaviors such as browsing job announcements, watching employment guidance videos, and downloading job attachments on mobile or PC devices, and generate behavior sequences with time tags; The periodic survey questionnaire interface is used to proactively push standardized questionnaires to students after each preset collection period (e.g., 30 days) or after a specific event is triggered (e.g., completion of a professional course) to obtain their self-reported changes in intention status. Each student is assigned a unique virtual identity identifier Stu, and all collected behavioral data is integrated in time series form to generate student job intention information. Before being stored in the database, the behavioral data needs to undergo data cleaning and standardization: first, page dwell records with a duration of less than a preset threshold (such as 5 seconds) are removed to eliminate accidental touch behavior; then, natural language processing is performed on the search keywords to extract core entities. For example, "want to find a job with high pay, low workload and close to home" is standardized into three structured tags: "high salary", "low intensity" and "local employment".

[0026] Step S102: Based on students' job-seeking intention information, establish employment intention tags, career planning timelines, and job-seeking demand parameters. The specific process includes: A time-series aggregation and weight allocation of students' job-seeking intention information was conducted to establish an employment intention tagging system. The employment intention tags include, but are not limited to, primary tags such as industry preference, job type, region selection, and salary expectation. Each primary tag has fine-grained secondary tags and quantitative weights. Industry Preference: Based on the national economic industry classification standard, it is subdivided into secondary tags such as "information technology", "finance", "manufacturing" and "education". The preference intensity is accumulated based on the frequency and duration of students' searches or browsing of relevant industry content. For example, if students search for "semiconductor" related positions for three consecutive days, the intensity weight of the "manufacturing-semiconductor" tag will increase. Job type: Subdivided into secondary tags such as "Technology R&D", "Marketing", "Operations Management" and "Functional Support", which can be further linked to specific job titles (such as "Front-end Engineer" and "Marketing Planner"). Region selection: Based on the city or region searched by the student, combined with IP address location, establish a region preference vector, in the format of [province, city, city level (first-tier / new first-tier / second-tier, etc.), whether to accept different regions]; Salary expectation: Cluster the salary ranges in the job postings that students click or browse to form an expected salary range [lower limit, upper limit] and a median expected value; Simultaneously, a career planning timeline is established, which includes three scales: short-term goals (first job within 0-6 months before graduation), mid-term goals (career accumulation within 1-3 years after graduation), and long-term goals (career achievements within 3-5 years after graduation). Time-related patterns are extracted from students' behavioral sequences. For example, if a student frequently browses "management trainee" positions and watches "leadership training" courses, while paying less attention to "senior expert" positions, it can be inferred that their short-term goal is "management trainee," their mid-term goal is "team leader," and their long-term goal is "department manager." Establish job search requirement parameters, which include the desired company type (state-owned enterprises, private enterprises, foreign-funded enterprises, public institutions, etc., represented by discrete labels); internship requirements (status such as "internship required", "accept employment first, internship later", "internship not required"); and skills enhancement needs (extracted from searched course keywords, such as "Python data analysis" and "PMP certification", with intensity levels such as "urgent", "general", and "understand").

[0027] Step S103: Establish an employment intention map. The specific process includes: Each student is taken as the central node of intention. A graph timeline is set in a two-dimensional or three-dimensional visualization space based on the career planning timeline. The graph timeline is not a static straight line, but is stretched or compressed according to the student's intention activity at different times. The timeline of the stage with frequent intention changes is lengthened to accommodate more state nodes. Multiple intention attribute nodes are radially connected around the intention center node. The intention attribute nodes are divided into three layers: the first layer is direct label nodes (such as "industry preference - information technology"), the second layer is related skill nodes (such as "Java" and "Spring framework"), and the third layer is external constraint nodes (such as "family expectations - stay in the province" and "economic pressure - urgent need for employment"). The connection edges between nodes are assigned weights, which are obtained by quantifying the intention intensity. The intention strength is calculated as follows: for any edge E connecting the intention center node and the intention attribute node i... i Its intention intensity S(E) i ) =α·F(c i ) +β·R(t i ) +γ·M(v i ); Wherein, F(c) i R(t_i) is a function based on behavior frequency (such as number of searches, dwell time), R(t_i) is a behavior recency function (the closer the behavior is to the current time, the higher the weight), and M(v_i) is a function based on behavior frequency (such as number of searches, dwell time). i ) is a behavioral value function (e.g., the weight of actively submitting a resume is higher than that of passively browsing), α, β, and γ are preset weighting coefficients, and α+β+γ=1; Furthermore, a temporal attention mechanism is introduced to capture the vibrational signals of intention. The specific process includes: setting a sliding time window of length L (e.g., 30 days), and treating all events generated within the sliding time window (such as a retrieval or a delivery) as a sequence; For each intention attribute node, a multi-head attention mechanism is used to obtain its cumulative activation count A within the current sliding time window. i(t) A i(t) It is the weighted sum of the contributions of all events to the intention attribute node i within the window before time t, where the weight of the event is determined by the attention score, so that recent key behaviors (such as revising the desired city in the resume) receive more attention. Set a dynamic threshold θ for each intention attribute node i(t) θ i(t) It is the sum of the exponential moving average of the cumulative activation count of intention attribute node i over its history and a baseline deviation. When the cumulative activation intensity A of a certain intention attribute node is... i(t) Continuously exceeding its dynamic threshold θ i(t)When the preset number of confirmations is reached (e.g., 3 times), the core intent update event is triggered; The core intention update event will automatically update the student's employment intention graph, mark the node as a new core intention label, and reduce the weight of other intention attribute nodes that semantically conflict with this intention attribute node.

[0028] Furthermore, step S2 is implemented through the following process: Step S201: Extract the employment intention map of each student to generate an employment intention vector. The specific process includes: The employment intention map established for each student is vectorized and encoded to generate a fixed-dimensional employment intention vector. The dimensions of the employment intention vector must be designed to comprehensively cover the key information of the intention map. Specific dimensions include: Industry Dimension: Major categories related to student employment in the national economic industry classification (such as manufacturing, construction, and finance) are used as dimension components. The value of each component is the student's intention intensity S(E) for the corresponding industry. i ); Job Role Dimension: Several representative job clusters (such as R&D, testing, product, etc.) are selected, and the value of each dimension is also the normalized intention intensity S(E). i ); Geographic dimension: Based on historical employment big data, the country is divided into several popular regions (such as Beijing-Tianjin-Hebei, Yangtze River Delta, Pearl River Delta, Chengdu-Chongqing, and the middle reaches of the Yangtze River). The value of each dimension is the intensity of students' intention to work in that region, and is modulated by whether students accept working in other places (yes / no).

[0029] Salary dimension: These represent the lower limit, median, and upper limit of expected salary, the expected annual growth rate of salary increase (expressed as a percentage), and the degree of importance attached to non-salary benefits (such as stock options and housing allowances) (a scalar between 0 and 1). Trait Tendency Dimensions: Based on student behavior sequences, several quantifiable non-cognitive trait tendencies are extracted, including but not limited to: risk preference (inferred from the stability of the companies applied to), innovation tendency (inferred from the frequency of searching for novel positions), autonomy (inferred from the ratio of active searching to passive acceptance), etc. Finally, we obtain the employment intention vector for each student, denoted as V. Stu .

[0030] Step S202: Establish a career perception space by inputting the employment intention vectors of all students into the career perception space. Based on the distribution of these vectors, intention clusters are identified within the career perception space. The specific process includes: A career perception space is set up to accommodate the employment intention vectors of all students and perform relation calculations, including a generalized career dimension, a characteristic career dimension, and a corporate career dimension. Generalized Occupational Dimension: Located at the highest abstraction layer of the space, it contains several generalized occupational dimension nodes, such as "Technology R&D", "Marketing", "General Management", "Public Service", etc. Each generalized occupational dimension node is associated with a set of feature occupational dimension nodes with high semantic similarity. For example, the "Technology R&D" generalized node is associated with multiple feature nodes such as "Backend Development", "Frontend Development", "Algorithm Engineering", "Test Development" etc. Featured Occupation Dimension: Located in the middle layer of the space, it contains several generalized occupational dimension nodes, such as "Java Backend Engineer", "Fast Moving Consumer Goods Marketing Specialist", and "Construction Worker". Each featured occupational dimension node can be associated with one or more generalized occupational dimension nodes at the same time (for example, "Quantitative Researcher" can be associated with "Technical Research and Development" and "Financial Analysis" at the same time), and with multiple enterprise occupational dimension nodes at the same time.

[0031] Enterprise Career Dimension: Located at the bottom layer of the space, it represents a set of job openings in an enterprise. It contains several enterprise career dimension nodes. Each enterprise career dimension node corresponds to a specific job information that exists in the labor market and includes a detailed description of the job: enterprise name, enterprise type, job responsibilities, job requirements, work location, salary range, etc. An enterprise career dimension node must be associated with at least one characteristic career dimension node and can selectively be associated with generalized dimensions. The vector V of all students' employment intentions Stu Input into the occupational perception space, due to V Stu The employment intention vector V of each dimension node in the career perception space Stu This indicates that within the same semantic space, the employment intention vector V is obtained. Stu The employment intention vector V is adjusted based on the cosine similarity of the nodes in each dimension. Stu In the spatial location of the career perception space, that is, adjusting the employment intention vector V according to the magnitude of cosine similarity. Stu The distance between nodes in each dimension is calculated by the cosine similarity; the greater the cosine similarity, the closer the distance, and so on. Within the occupational perception space, several vectors V, each representing a different employment intention, are randomly divided. Stu The intention clusters are formed, and then, with each intention cluster as the central intention cluster, the employment intention vectors V outside the central intention clusters are obtained using the Euclidean distance formula. Stu The distance to the edge of the central intended cluster, and a threshold for the mean distance is set; Based on the principle of closest distance, each employment intention vector V is... StuAssign to the intention cluster with the smallest distance, and then obtain all employment intention vectors V within the intention cluster. Stu Calculate the mean distance between them, and determine the employment intention vectors V within the current intention cluster. Stu If the mean distance between them is less than or equal to the mean distance threshold, then repeat the above process to update the employment intention vector V within the intention cluster. Stu The operation continues until all employment intention vectors V within the intention cluster are processed. Stu The intention clusters are divided into several groups until the mean distance between them is greater than the mean distance threshold. Each cluster of intentions represents a group of students with similar employment goals. The shape of a cluster of intentions in space (spherical, elongated, or irregular) reflects the consistency and diversity of intentions within the group. The core area of ​​the cluster of intentions represents the most typical career intentions of the group, while the peripheral area of ​​the cluster represents students in the stage of intention transformation or ambiguity. In addition, each intention cluster automatically generates common short-term goals, common medium-term goals, and common long-term goals for the group based on the career planning timeline; For example, for a dense cluster of "information technology - backend development" aspirations, their common short-term goal can be identified as "mastering the basics of Java / Go and obtaining an internship offer", the medium-term goal is "becoming a core developer of a project or switching to an architecture direction", and the long-term goal is "a technical expert or technical manager".

[0032] Step S203: Continuously track the real-time evolution of each intention cluster and simultaneously adjust the distribution of intention clusters in the career perception space. The specific process includes: Set a state update period t. After each state update period t, the employment intention vectors of all students may undergo a slight shift, which will cause changes in the shape, position and membership relationship of the intention clusters. By comparing the distribution of intention clusters in two consecutive state update periods t and t+1, we can track their evolution process. Evolutionary processes are tracked by analyzing key indicators both between and within clusters: Inter-cluster distance matrix change: Calculate the Euclidean distance matrix between the centroids of all intention clusters. If the Euclidean distance D(t) between two intention clusters at time t is greater than D(t+1) and the difference exceeds the preset fusion threshold, it is marked as a merging trend. When the merging trend continues for more than τ cycles (e.g., 3 state update cycles) and the growth rate of the merged cluster size (member growth rate) exceeds the preset popular speed threshold (e.g., weekly growth rate of 30%), the merging direction is marked as a popular merging direction. The variance of the average displacement vector of members within a cluster: Calculate the variance Var of the average displacement of all individual vectors of members within an intention cluster relative to the centroid of the cluster. If the variance Var gradually increases over time and exceeds the preset differentiation threshold, it is marked as a differentiation trend. This indicates that although the overall intention of the group seems stable, the differences between individuals within the group are significantly increasing. At this time, the intention cluster is no longer a highly homogeneous group and needs to be further split into new intention clusters. The occupational perception space is dynamically adjusted based on the tracking results. For example, if the popular confluence direction points to an emerging occupation that is not covered by the existing characteristic occupational dimensions (such as "prompt word engineer"), then characteristic occupational dimension nodes are added to the characteristic occupational dimension layer, and their vector representations are initialized using a small number of labeled samples or from the split of generalized occupational dimension nodes. If a certain intention cluster shrinks or even disappears over a long period of time, the weight of its corresponding occupational dimension node will be decayed, but its historical trajectory is retained for backtracking analysis.

[0033] Furthermore, step S3 is implemented through the following process: Step S301: Set up a personalized guidance scheme library. The personalized guidance scheme library generates personalized guidance schemes based on the real-time distribution of intention clusters. The specific process includes: The personalized guidance solution library consists of three components: resume optimization templates, interview simulation scripts, and employment policy interpretation packages. Furthermore, after each state update cycle t, the personalized guidance solution library first retrieves the employment intention vector V at the center position of each intention cluster. Stu Generate corresponding common guidance plans, and then combine these common guidance plans with each student's employment intention vector V. Stu Generate personalized guidance plans; The resume optimization template outputs a keyword whitelist (e.g., keywords for backend development positions include "high concurrency," "distributed," and "JVM tuning") and a blacklist (e.g., invalid words such as "hardworking" and "responsible") based on the job position dimension and trait tendency dimension in the employment intention vector. Then, it calls natural language generation technology to rewrite the student's original resume in a structured way: highlighting and placing words matching the whitelist at the beginning, compressing or deleting irrelevant experiences, and adjusting the language style according to the nature of the intended company. The interview simulation script is used to pre-set multiple core questions for each specific position under each characteristic occupational dimension. Each core question is accompanied by a scoring standard to evaluate the student's expression ability, logic and professionalism. The employment policy interpretation package connects with local human resources and social security bureaus and the Ministry of Education's employment information platform through data interfaces to capture the latest regional talent introduction policies (such as settlement subsidies and settling-in allowances), grassroots projects ("Three Supports and One Assistance" and "Western Plan"), and industry-specific policies (such as tax incentives in the fields of integrated circuits and artificial intelligence). It then selects the most relevant policies based on geographical, salary, and industry dimensions, and generates a structured interpretation report. The report includes: the policy's scope of application, application conditions and procedures, the degree of matching with the student's intentions (expressed as a percentage), and comparative analysis. Then, the output results of each component are integrated to generate common guidance schemes and personalized guidance schemes.

[0034] Step S302: Establish a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed an employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed an employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated. The specific process includes: The closed-loop tracking mechanism for the employment process includes determining whether an employment contract has been signed, and if not, further determining whether there has been an update on the student's job search intention information. If the completion of the employment contract signing is determined through the three-party agreement upload interface of the academic affairs system or by a flag manually confirmed by the counselor, then the corresponding student will be moved to the "employed - final" state, and the generation of new personalized guidance plans will be stopped, and the corresponding employment intention vector V will be removed. Stu ; If it is determined that an employment contract has not been signed, the system further checks whether there has been an update to the student's job intention information. If any core intention tag in the job intention map changes, or if the cumulative activation intensity exceeds a threshold leading to a core intention update, or if more than a hard period (e.g., 14 days) has passed since the last update, then the student's job intention information is considered to have been updated, and a new job intention vector V is generated. Stu The distribution of intention clusters is updated synchronously, and new personalized guidance plans are generated. The process of generating new personalized guidance plans is repeated until the corresponding student is judged to have completed the employment contract signing process. At the same time, whenever it is determined that a student has completed an employment contract or updated a student's job intention information, a class employment statistics report is generated or updated. The class employment statistics report includes the class employment rate: (number of students who have signed contracts / total number of students in the class) * 100%, and includes the month-on-month change (compared to last week / last month). Intention Matching Degree: Calculate the semantic similarity between the actual job position of the signed student and the job dimension of the last core intention tag, and take the class average. Key bottlenecks for unemployed students: The main reasons for "unemployment" are categorized in pie chart or bar chart form, including but not limited to: "low resume pass rate" (application-interview conversion rate < 10%), "low mock interview score" (below the threshold of 60 points), "frequent changes in intentions" (core intentions changed more than 3 times in the past 30 days), "no action record" (no search or application behavior in the past 14 days), and "insufficient job supply" (the job demand in the area where the intention cluster is located is significantly less than the number of job seekers). Employment progress details for each cluster: For each cluster, the current number of students, the number of students who have signed contracts, the average number of applications, the average number of interviews, and the student intention-action consistency score (correlation coefficient between intention strength and application behavior) are listed. In particular, clusters corresponding to popular convergence directions are highlighted and the current competition intensity of that direction is estimated (number of students in the cluster / number of new positions in that direction in the market).

[0035] Please see Figure 2 As shown, a precision guidance system for college students' employment intentions includes a student employment guidance module and a teacher tutoring module. The student employment guidance module is used to obtain students' job intention information in real time and receive personalized guidance plans from the teacher tutoring module. Based on students' job-seeking intention information, we establish employment intention tags, career planning timelines, and job-seeking demand parameters, and then build a corresponding employment intention map. Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated.

[0036] The teacher guidance module includes an intention analysis unit and an employment report unit; The intention analysis unit is used to establish employment intention tags, career planning timelines, and job demand parameters based on students' job-seeking intention information, thereby establishing a corresponding employment intention map, extracting employment intention vectors from each student's employment intention map, and setting up a career perception space. The employment intention vectors of all students are input into the career perception space, and intention clusters are divided in the career perception space according to the distribution of employment intention vectors. The real-time evolution process of each intention cluster is continuously tracked, and the distribution of intention clusters in the career perception space is adjusted synchronously. A personalized guidance scheme library is set up. The personalized guidance scheme library generates personalized guidance schemes based on the real-time distribution of intention clusters, and a closed-loop tracking mechanism for the employment process is established. Whenever it is determined that a student has not completed an employment contract and the student's job-seeking intention information is updated, a new personalized guidance scheme is generated based on the updated student's job-seeking intention information. The employment reporting unit is used to generate or update the class employment statistics report whenever it is determined that a student has completed an employment contract or that a student's job search intention information has been updated.

[0037] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for precise guidance on college students' employment intentions, characterized in that, Includes the following steps: Step S1: Obtain students' job-seeking intention information in real time, and establish employment intention tags, career planning timelines and job-seeking demand parameters based on the students' job-seeking intention information, thereby establishing a corresponding employment intention map; Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated.

2. The method for precise guidance of college students' employment intentions according to claim 1, characterized in that, The process of obtaining students' job intentions in real time includes: The student job intention information is formed by collecting and integrating students' online job search behavior data, and deploying backend log tracking, frontend behavior capture instructions, and periodic survey questionnaire interfaces within the employment guidance platform; Each student is assigned a unique virtual identity identifier, Stu, and all collected behavioral data is integrated in time series format to generate students' job intention information.

3. The method for precise guidance on college students' employment intentions according to claim 2, characterized in that, The process of establishing employment intention tags, career planning timelines, and job demand parameters based on students' job-seeking intention information includes: The system aggregates and assigns weights to students' job-seeking intentions based on time sequence, establishing a job intention tagging system. These tags include primary tags for industry preference, job type, region selection, and salary expectations. Each primary tag has fine-grained secondary tags and quantified weights. Simultaneously, a career planning timeline is established, which includes three scales: short-term goals, medium-term goals, and long-term goals. Job search demand parameters are also established, including the desired company type, internship requirements, and skills enhancement needs.

4. The method for precise guidance on college students' employment intentions according to claim 3, characterized in that, The process of establishing an employment intention map includes: Each student is treated as a central node of intent, and a graph timeline is set up in a two-dimensional or three-dimensional visualization space based on the career planning timeline. Multiple intention attribute nodes are radially connected around the intention center node. The intention attribute nodes are divided into three layers: the first layer is direct label nodes, the second layer is associated skill nodes, and the third layer is external constraint nodes. The connection edges between nodes are assigned weights, which are obtained by quantifying the intention intensity. Furthermore, a temporal attention mechanism is introduced to capture the vibrational signals of intention. A sliding time window of length L is set, and all events generated within the sliding time window are regarded as a sequence. For each intention attribute node, a multi-head attention mechanism is used to obtain its cumulative activation count A within the current sliding time window. i(t) A i(t) It is the weighted sum of the contributions of all events to the intention attribute node i within the window before time t; Set a dynamic threshold θ for each intention attribute node i(t) θ i(t) It is the sum of the exponential moving average of the cumulative activation count of intention attribute node i over its history and a baseline deviation. When the cumulative activation intensity A of a certain intention attribute node is... i(t) Continuously exceeding its dynamic threshold θ i(t) When the preset number of confirmations is reached, the core intent update event is triggered.

5. A method for precise guidance on college students' employment intentions according to claim 4, characterized in that, The process of extracting the employment intention map of each student and generating the employment intention vector includes: The employment intention map established for each student is vectorized and encoded to generate a fixed-dimensional employment intention vector. The dimensions of the employment intention vector must comprehensively cover the key information of the intention map, including industry, job position, region, salary, and trait preference dimensions. Finally, the employment intention vector for each student is obtained, denoted as V. Stu .

6. The method for precise guidance of college students' employment intentions according to claim 5, characterized in that, A career perception space is set up, and the employment intention vectors of all students are input into the career perception space. The process of dividing the career intention clusters in the career perception space according to the distribution of employment intention vectors includes: A career perception space is set up to accommodate the employment intention vectors of all students and perform relation calculations, including a generalized career dimension, a characteristic career dimension, and a corporate career dimension. The vector V of all students' employment intentions Stu Input into the career perception space to obtain the employment intention vector V Stu The employment intention vector V is adjusted based on the cosine similarity of the nodes in each dimension. Stu Spatial location within the professional perception space; Within the occupational perception space, several vectors V, each representing a different employment intention, are randomly divided. Stu The intention clusters are formed, and then, with each intention cluster as the central intention cluster, the employment intention vectors V outside the central intention clusters are obtained using the Euclidean distance formula. Stu The distance to the edge of the central intended cluster, and a threshold for the mean distance is set; Based on the principle of closest distance, each employment intention vector V is... Stu Assign to the intention cluster with the smallest distance, and then obtain all employment intention vectors V within the intention cluster. Stu Calculate the mean distance between them, and determine the employment intention vectors V within the current intention cluster. Stu Whether the mean distance between them is less than or equal to the mean distance threshold, several intention clusters are divided based on the judgment result.

7. A method for precise guidance on college students' employment intentions according to claim 6, characterized in that, The process of continuously tracking the real-time evolution of each intention cluster and simultaneously adjusting the distribution of intention clusters in the career perception space includes: Set a state update period t. After each state update period t, the employment intention vectors of all students may undergo a slight shift, which will cause changes in the shape, position and membership relationship of the intention clusters. By comparing the distribution of intention clusters in two consecutive state update periods t and t+1, we can track their evolution process. The evolution process is tracked by analyzing the changes in the inter-cluster distance matrix and the variance of the average displacement vector of intra-cluster members, and the occupational perception space is dynamically adjusted based on the tracking results.

8. A method for precise guidance on college students' employment intentions according to claim 7, characterized in that, The process of setting up a personalized guidance scheme library, which generates personalized guidance schemes based on the real-time distribution of intention clusters, includes: The personalized guidance solution library consists of three components: resume optimization templates, interview simulation scripts, and employment policy interpretation packages. Furthermore, after each state update cycle t, the personalized guidance solution library first retrieves the employment intention vector V at the center position of each intention cluster. Stu Generate corresponding common guidance plans, and then combine these common guidance plans with each student's employment intention vector V. Stu Generate personalized guidance plans.

9. A method for precise guidance on college students' employment intentions according to claim 8, characterized in that, Establish a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed an employment contract and their job intention information is updated, a new personalized guidance plan is generated based on the updated information. The process of generating or updating the class employment statistics report whenever it is determined that a student has completed an employment contract or their job intention information is updated includes: The closed-loop tracking mechanism for the employment process includes determining whether an employment contract has been signed, and if not, further determining whether there has been an update on the student's job search intention information. If the completion of the employment contract signing is determined through the three-party agreement upload interface of the academic affairs system or by a flag manually confirmed by the counselor, then the corresponding student will be moved to the "employed - final" state, and the generation of new personalized guidance plans will be stopped, and the corresponding employment intention vector V will be removed. Stu ; If it is determined that an employment contract has not been signed, the next step is to determine whether there has been an update to the student's job intention information. If any core intention tag in the job intention map changes, or the cumulative activation intensity exceeds a threshold leading to a core intention update, or if more than one hard cycle has passed since the last update, then the student's job intention information is considered to have been updated, and a new job intention vector V is generated. Stu And simultaneously update the distribution of intention clusters, thereby generating new personalized guidance plans; In addition, whenever it is determined that a student has completed an employment contract or updated their job search intention information, a class employment statistics report is generated or updated.

10. A precise guidance system for college students' employment intentions, used to implement the precise guidance method for college students' employment intentions as described in any one of claims 1-9, characterized in that, Includes a student career guidance module and a teacher tutoring module; The student employment guidance module is used to obtain students' job intention information in real time and receive personalized guidance plans from the teacher tutoring module. Based on students' job-seeking intention information, we establish employment intention tags, career planning timelines, and job-seeking demand parameters, and then build a corresponding employment intention map. Step S2: Extract the employment intention map of each student to generate employment intention vector, and set up a career perception space. Input the employment intention vectors of all students into the career perception space, divide the intention clusters in the career perception space according to the distribution of employment intention vectors, continuously track the real-time evolution process of each intention cluster, and adjust the distribution of intention clusters in the career perception space in sync. Step S3: Set up a personalized guidance plan library. The personalized guidance plan library generates personalized guidance plans based on the real-time distribution of intention clusters and establishes a closed-loop tracking mechanism for the employment process. Whenever it is determined that a student has not completed the employment contract and the student's job intention information is updated, a new personalized guidance plan is generated based on the updated student's job intention information. Whenever it is determined that a student has completed the employment contract or the student's job intention information is updated, a class employment statistics report is generated or updated. The teacher guidance module includes an intention analysis unit and an employment report unit; The intention analysis unit is used to establish employment intention tags, career planning timelines, and job demand parameters based on students' job-seeking intention information, thereby establishing a corresponding employment intention map, extracting employment intention vectors from each student's employment intention map, and setting up a career perception space. The employment intention vectors of all students are input into the career perception space, and intention clusters are divided in the career perception space according to the distribution of employment intention vectors. The real-time evolution process of each intention cluster is continuously tracked, and the distribution of intention clusters in the career perception space is adjusted synchronously. A personalized guidance scheme library is set up. The personalized guidance scheme library generates personalized guidance schemes based on the real-time distribution of intention clusters, and a closed-loop tracking mechanism for the employment process is established. Whenever it is determined that a student has not completed an employment contract and the student's job-seeking intention information is updated, a new personalized guidance scheme is generated based on the updated student's job-seeking intention information. The employment reporting unit is used to generate or update the class employment statistics report whenever it is determined that a student has completed an employment contract or that a student's job search intention information has been updated.

Images