An evaluation method based on a student innovation and entrepreneurship capability model

Abstract

The application discloses a kind of evaluation methods based on student innovation and entrepreneurship ability model, it is related to educational evaluation technical field, the specific steps of this method are as follows: first, collect student innovation and entrepreneurship multi-source heterogeneous data and pre-process;Then build two-dimensional evaluation model, distribute weight;Again, each cycle data is quantified and scored, fusion generates time sequence characteristic library;Subsequently, cause-effect analysis is carried out, and the ability dynamic image is generated;Finally, the intervention priority is calculated to generate a scheme, and the above steps are repeated according to the cycle to form a closed-loop management to improve the student innovation and entrepreneurship ability;The application builds two-dimensional evaluation model to realize full-dimension coverage evaluation, breaks through the traditional limitation, and guarantees the result to be comprehensive and accurate;Also, the ability variable causal chain is excavated, the dynamic image is drawn, the short board is diagnosed, the trend is predicted, the exclusive guidance scheme is generated and closed-loop iteration is carried out, the change from passive evaluation to active intervention is realized, and the student innovation and entrepreneurship ability is systematically improved.

Description

Technical Field

[0001] This invention relates to the field of educational evaluation technology, specifically to an evaluation method based on a student innovation and entrepreneurship ability model. Background Technology

[0002] Innovation and entrepreneurship education has become an important part of the higher education system. Universities attach great importance to the cultivation and evaluation of students' innovation and entrepreneurship abilities. Students' participation in innovation and entrepreneurship practices is becoming increasingly diverse, generating a large amount of relevant data in areas such as competition participation, intellectual property application, and project operation. At the same time, information that can be mined can also be formed through various behaviors and texts regarding intrinsic qualities such as innovative thinking and teamwork. With the advancement of digital education, multi-source and heterogeneous innovation and entrepreneurship-related data continues to accumulate. The evaluation needs of the education field for students' innovation and entrepreneurship abilities have shifted from a single outcome assessment to a comprehensive evaluation that is multi-dimensional and process-oriented. It is not only necessary to quantitatively evaluate students' innovation and entrepreneurship abilities, but also to use the evaluation results to diagnose ability gaps, predict development trends, and provide personalized guidance. This requires evaluation methods that can adapt to the processing needs of multiple types of data, taking into account both the external performance and intrinsic qualities of abilities, and supporting the refined and scientific development of innovation and entrepreneurship education.

[0003] Traditional methods for evaluating students' innovation and entrepreneurship abilities have many obvious limitations. Their evaluation dimensions often focus on explicit achievements, neglecting the core supporting role of implicit traits such as innovative thinking, business acumen, and resilience. This makes it difficult to achieve a comprehensive evaluation of students' abilities. In terms of data processing, there is a lack of standardized collection and preprocessing procedures, and the multi-source, heterogeneous innovation and entrepreneurship data is not systematically integrated, resulting in insufficient data validity and standardization. Furthermore, the allocation of indicator weights often relies on subjective experience, lacking scientific quantitative allocation methods, leading to poor objectivity and accuracy in the evaluation results. In addition, traditional evaluations are mostly static, one-off assessments, failing to form a time-series ability tracking system. This makes it impossible to uncover the intrinsic connections between different ability dimensions, only identifying superficial weaknesses, failing to pinpoint the core root causes of problems, and unable to generate targeted guidance based on the evaluation results, thus failing to provide effective support for improving students' innovation and entrepreneurship abilities. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an evaluation method based on a student innovation and entrepreneurship ability model. This method forms a closed-loop system for the entire process of evaluating student innovation and entrepreneurship abilities. Based on periodic collection of multi-source heterogeneous data, after standardized preprocessing, a dual-dimensional evaluation model of explicit achievements and implicit traits is constructed, and indicator weights are scientifically allocated. A time-series feature library is generated through feature fusion, and then time-series causal analysis is used to mine ability correlations, completing dynamic profile creation, root cause diagnosis of weaknesses, and trend prediction. Finally, personalized guidance plans are generated based on intervention priorities and iteratively optimized at fixed intervals. This method breaks through the single-dimensional and static limitations of traditional evaluation, achieving full-dimensional, time-series, and precise ability evaluation. It provides systematic technical support for personalized guidance and scientific cultivation in innovation and entrepreneurship education, effectively promoting the systematic improvement of students' innovation and entrepreneurship abilities.

[0005] To solve the above-mentioned technical problems, this invention provides the following technical solution: an evaluation method based on a student innovation and entrepreneurship ability model, the specific steps of which are as follows: S1, Data Acquisition and Preprocessing: Collect multi-source heterogeneous data on student innovation and entrepreneurship according to a preset fixed period, and perform standardized preprocessing on the data of each period to obtain the standardized time series dataset corresponding to each period. S2, Model Construction and Weight Allocation: Based on the standardized time-series dataset, a two-dimensional evaluation model covering explicit outcome dimensions and implicit trait dimensions is constructed. The weights of each indicator are allocated through the analytic hierarchy process to form a standardized model system. S3, Feature Fusion and Temporal Feature Library Generation: Based on the standardized model system, the data of each period are quantitatively scored for each indicator. The quantitative results of the explicit achievement dimension and the implicit trait dimension of the same period are input into the dual-dimensional temporal capability fusion algorithm to calculate and generate a single-period standardized feature vector. All period feature vectors are summarized to form a temporal feature library. The time-series feature library arranges and stores all single-cycle standardized feature vectors in chronological order of collection cycles. Each single-cycle standardized feature vector is configured with a corresponding cycle time identifier, indicator scoring record, and data source identifier, forming a traceable, updatable, and callable time-series feature data set. S4, Causal Analysis and Profile Generation: Input the time-series feature library into the time-series causal effect strength algorithm to calculate the causal transmission path and effect strength between each ability variable, generate the ability evolution causal chain, and generate a dynamic profile of students' innovation and entrepreneurship ability, diagnosis of the root causes of ability shortcomings, and prediction of ability evolution trends based on the causal chain analysis results. S5, Intervention plan generation and closed-loop iteration: Input the effect intensity of each upstream node in the causal chain of capability evolution into the root cause node intervention priority algorithm, calculate the intervention priority value of each upstream node, generate a personalized entrepreneurship guidance service plan based on the priority value, and repeat steps S1 to S5 according to a preset cycle to form closed-loop management.

[0006] Furthermore, in step S1, the collection of multi-source heterogeneous data on student innovation and entrepreneurship according to a preset fixed period specifically means: the preset fixed period is a natural quarter, or it can be dynamically adjusted according to the student's entrepreneurial stage. The collection period is once a month in the initial stage and once a natural quarter in the mature stage. The collected multi-source heterogeneous data includes explicit achievement-type structured data and implicit characteristic-type unstructured data. Explicit achievement-type structured data includes innovation and entrepreneurship competition data, intellectual property data, project operation data, and innovation and entrepreneurship practice experience data. Implicit characteristic-type unstructured data includes text data, audio data, video data, and behavioral process data.

[0007] Furthermore, in step S1, the standardization preprocessing performed on the data for each period specifically includes: performing missing value filling, outlier removal, format unification, and categorized data encoding on the structured data of explicit results; performing garbled character cleaning, redundant content removal, format unification, and semantic normalization on the text data of implicit characteristics; performing environmental noise reduction, non-human voice segment clipping, and format unification on the audio data of implicit characteristics; and performing blurred frame removal, irrelevant segment clipping, audio-visual synchronization alignment, and format unification on the video data of implicit characteristics.

[0008] Furthermore, in step S2, the dual-dimensional evaluation model covering both explicit and implicit trait dimensions is specifically as follows: The explicit results dimension sets four primary indicators: competition results, intellectual property, project operation, and innovation and entrepreneurship practice experience. The competition results primary indicator corresponds to a secondary sub-indicator of competition level; the intellectual property primary indicator corresponds to a secondary sub-indicator of patent type; the project operation primary indicator corresponds to a secondary sub-indicator of revenue scale; and the innovation and entrepreneurship practice experience primary indicator corresponds to a secondary sub-indicator of training duration. The implicit trait dimension sets five primary indicators: innovative thinking, business acumen, team leadership, resilience, and communication skills. The innovative thinking primary indicator corresponds to a secondary sub-indicator of innovation degree; the business acumen primary indicator corresponds to a secondary sub-indicator of market analysis accuracy; the team leadership primary indicator corresponds to a secondary sub-indicator of coordination ability; the resilience primary indicator corresponds to a secondary sub-indicator of risk response ability; and the communication skills primary indicator corresponds to a secondary sub-indicator of logical expression ability. Each indicator has a clearly defined quantitative scoring range and judgment criteria. The specific steps of quantitatively scoring each indicator of data in each cycle based on the standardized model system are as follows: For the explicit achievement-type structured data in the standardized time series dataset of each cycle, the standardized scoring of each sub-indicator is completed item by item, based on the indicators and scoring rules of each level of the explicit achievement dimension in the standardized model system and the actual innovation and entrepreneurship achievements generated by students in that cycle; For the implicit trait-type unstructured data in the standardized time series dataset of each cycle, the standardized scoring of each type of unstructured data is completed item by item, based on the indicators and scoring rules of each level of the implicit trait dimension in the standardized model system and the corresponding sub-indicators, with all indicator scores using a quantitative standard of 0-100 points. After scoring, the quantitative results of the explicit achievement dimension and implicit trait dimension of each cycle are formed.

[0009] Furthermore, in step S2, the allocation of weights for each indicator using the analytic hierarchy process specifically involves: constructing judgment matrices for the first-level indicators and second-level sub-indicators in the dual-dimensional evaluation model, performing a consistency check on the judgment matrices, calculating the eigenvectors of each indicator after passing the consistency check, normalizing the eigenvectors to obtain the corresponding weight allocation for each indicator, and completing the weight allocation within and between the explicit outcome dimension and the implicit trait dimension based on the normalization results.

[0010] Furthermore, in step S3, the mathematical expression of the dual-dimensional temporal capability fusion algorithm is: in, For the first Periodic single-period standardized feature vector, This is a two-dimensional adaptation coefficient, with a value ranging from 0.3 to 0.5. For the explicit results dimension Itemized indicator weights, For the first The first dimension of periodic explicit results Standardized scores for each indicator This is the time-series decay coefficient. For the latent trait dimension Itemized indicator weights, For the first Periodic latent trait dimension Standardized scores for each indicator , These are the total number of sub-indicators for explicit and implicit dimensions, respectively; The quantitative results of explicit achievement dimension and implicit trait dimension in the same period are input into the dual-dimensional time-series capability fusion algorithm to calculate and generate a single-period standardized feature vector. Specifically, the quantitative scores of all sub-indicators of explicit achievement dimension in the same period are extracted to form an explicit quantitative feature sequence. Then, the quantitative scores of all sub-indicators of implicit trait dimension in the same period are extracted to form an implicit quantitative feature sequence. The two feature sequences are simultaneously input into the dual-dimensional time-series capability fusion algorithm. The algorithm combines the dual-dimensional adaptation coefficient, time-series decay coefficient and the weight of each indicator to perform unified spatial mapping and weighted integration of the two feature sequences. After calculation and processing, a single-period standardized feature vector with fixed dimensions and numerical standardization to the 0-1 interval is generated. This vector fully represents the full-dimensional state of students' innovation and entrepreneurship capabilities in the same period.

[0011] Furthermore, in step S4, the mathematical expression for the temporal causal effect strength algorithm is: in, For the capability dimension right The intensity of the temporal causal effect, ranging from 0 to 1. For the first Cyclical capability dimension Standardized scores For the capability dimension Average score over the entire period The periodic correlation coefficient, As a causal transmission attenuation factor, This represents the total number of data collection cycles. The time-series feature library is input into the time-series causal effect strength algorithm to calculate the causal transmission path and effect strength between each capability variable, generating a capability evolution causal chain. Specifically, the single-period standardized feature vectors of all periods in the time-series feature library are first arranged in chronological order, and the standardized scores corresponding to each capability variable in each vector are extracted to form the full-period time-series data of each capability variable. Then, the time-series data is input into the time-series causal effect strength algorithm. The algorithm first performs a stationarity test on the time-series data of each capability variable, removes data without stable change patterns, and then verifies the time-series influence relationship between each capability variable through calculation, clarifies the causal correspondence between each capability variable, selects capability variable pairs with significant causal effects, sorts out the complete causal transmission path from upstream node to downstream node, and quantifies the effect strength of each path. Finally, all causal transmission paths and effect strength data are integrated to generate a visualized capability evolution causal chain.

[0012] Furthermore, in step S4, the process of generating a dynamic profile of students' innovation and entrepreneurship capabilities, diagnosing the root causes of capability shortcomings, and predicting capability evolution trends based on the causal chain analysis results specifically involves: combining the capability evolution causal chain with the time-series change data of each capability dimension to draw a dynamic profile of capabilities that includes the time-series change curves of all indicators and the periodic quantitative score distribution; tracing back along the capability evolution causal chain to locate the upstream capability nodes corresponding to the shortcomings and sorting out the transmission path to form the content of capability shortcoming root cause diagnosis; and deducing the change trends of each capability dimension within a future set period based on historical time-series data and causal transmission relationships to determine potential change nodes in the capability development process.

[0013] Furthermore, in step S5, the mathematical expression of the root cause node intervention priority algorithm is: in, For the capability dimension The intervention priority value ranges from 0 to 1. For the capability dimension The strength of the total causal effect across all downstream dimensions For the capability dimension Current shortcomings and gaps For the capability dimension Average score over the entire period To determine the intervention cost-effectiveness coefficient, the effect intensity of each upstream node in the capability evolution causal chain is input into the root cause node intervention priority algorithm. Specifically, the intervention priority value of each upstream node is calculated by first extracting all upstream root cause nodes from the capability evolution causal chain, summarizing the causal effect intensity of each upstream root cause node on all its downstream nodes, calculating the total causal effect intensity of each upstream root cause node, and then obtaining the current shortcoming gap and the average score of the whole cycle for each upstream root cause node. The total causal effect intensity, the current shortcoming gap, the average score of the whole cycle, and the intervention cost-effectiveness coefficient are simultaneously input into the root cause node intervention priority algorithm. The intervention priority value of each upstream root cause node is calculated through the algorithm formula. All priority values ​​are in the range of 0-1, and the higher the value, the earlier the intervention order of the upstream root cause node.

[0014] Furthermore, in step S5, the personalized entrepreneurship guidance service plan generated based on the intervention priority values ​​of each upstream node specifically includes: a phased root cause capability enhancement plan divided into three preset phases: short-term, medium-term, and long-term; each phase clearly defines the enhancement goals, targeted training tasks, supporting innovation and entrepreneurship practice projects, and specific acceptance criteria for the highest priority upstream root cause node; a dedicated learning path that selects and matches basic introductory, advanced, and practical application learning content based on the capability gaps of the highest priority root cause node, clarifying the learning cycle, core learning focus, specific assignment requirements, and phased assessment nodes for each phase; and a one-on-one mentorship plan that matches mentors in corresponding industry fields based on the student's entrepreneurial track and the enhancement needs of the priority root cause node, clarifying the frequency of guidance, including online communication, offline tutoring, and project review, determining specific guidance themes in each phase, and clarifying the core content and process of each guidance session.

[0015] Compared with existing technologies, this evaluation method based on a student innovation and entrepreneurship ability model has the following beneficial effects: I. This invention constructs a dual-dimensional evaluation model encompassing both explicit achievements and implicit traits. By combining periodic collection and standardized preprocessing of multi-source heterogeneous data, it achieves a comprehensive evaluation of students' innovation and entrepreneurship capabilities across all dimensions. The analytic hierarchy process (AHP) is used to scientifically allocate weights to each indicator, forming a standardized model system. Corresponding indicators are matched to different types of data for unified quantitative scoring. A dual-dimensional time-series capability fusion algorithm then integrates the data, generating single-cycle standardized feature vectors and compiling them into a time-series feature library. This approach breaks away from the limitations of traditional evaluations that focus solely on explicit achievements. It considers both external achievements and internal traits in capability development, while comprehensively recording the changes in capabilities across different cycles in a time-series format. This makes capability evaluation more aligned with actual development patterns, ensuring the comprehensiveness and accuracy of evaluation results and truly representing the full-dimensional state of students' innovation and entrepreneurship capabilities.

[0016] Second, this invention utilizes a time-series causal effect strength algorithm to mine the causal transmission paths and effect strengths among various ability variables, generating a causal chain of ability evolution. Based on this, it completes the dynamic profiling of students' innovation and entrepreneurship abilities, diagnoses the root causes of shortcomings, and predicts development trends. It accurately locates the core sources of ability shortcomings from the perspective of time-series causal relationships, rather than merely identifying surface problems. Then, a root cause node intervention priority algorithm calculates the intervention priority of each upstream node, generating a phased, personalized entrepreneurship guidance service plan, forming a closed-loop iterative process from S1 to S5 with preset cycles as nodes. This approach provides scientific data and causal logic support for the formulation of innovation and entrepreneurship guidance services, realizing a shift from passive evaluation to proactive intervention and precise guidance. Simultaneously, through closed-loop iteration, it continuously tracks ability evolution and optimizes guidance plans, making ability cultivation and guidance services more targeted and timely, promoting a systematic improvement in students' innovation and entrepreneurship abilities.

[0017] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0019] Figure 1 A flowchart of an evaluation method based on a student innovation and entrepreneurship ability model; Figure 2 A flowchart for feature fusion and temporal feature library generation; Figure 3 This is a flowchart for causal analysis and profile generation. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0021] The evaluation method based on the student innovation and entrepreneurship ability model of the present invention, such as Figure 1 As shown, the process includes, in sequence, data collection and preprocessing steps, model building and weight allocation steps, feature fusion and temporal feature library generation steps, causal analysis and profile generation steps, and intervention plan generation and closed-loop iteration steps. Each step is executed sequentially. After the intervention plan generation and closed-loop iteration steps are completed, the next preset fixed cycle is used as a new starting point to repeat all the aforementioned steps. The steps work together to form a closed-loop dynamic evaluation and cultivation guidance system for students' innovation and entrepreneurship capabilities, realizing full-cycle tracking, multi-dimensional evaluation, root cause diagnosis, and personalized improvement of students' innovation and entrepreneurship capabilities.

[0022] S1, Data Acquisition and Preprocessing: This step is the foundational data acquisition stage of the entire evaluation method. It involves the timed collection and standardization of multi-source heterogeneous data on student innovation and entrepreneurship, providing a standardized and high-quality time-series dataset for subsequent model construction and analysis. The specific implementation consists of three parts: pre-set collection period, multi-source heterogeneous data collection, and standardization preprocessing, which are carried out sequentially. The data collection cycle of this invention adopts a combination of fixed and dynamic adjustment. The basic fixed cycle is the natural quarter. At the same time, it is dynamically adjusted according to the stage of students' entrepreneurship. For students in the early stage of entrepreneurship, whose innovation and entrepreneurship behaviors and abilities change rapidly and frequently, the data collection cycle is adjusted to once a month to achieve high-frequency tracking of ability status. For students in the mature stage, whose innovation and entrepreneurship development status tends to be stable, the basic data collection cycle of once a natural quarter is maintained to balance the timeliness of evaluation and data processing efficiency.

[0023] Multi-source heterogeneous data collection follows a pre-set collection cycle, gathering multi-source heterogeneous data related to student innovation and entrepreneurship. Data types are divided into two main categories: structured data representing explicit achievements and unstructured data representing implicit traits. These two types of data complement each other, achieving a comprehensive portrayal of students' innovation and entrepreneurship capabilities. Structured data of explicit achievements: Standardized data that directly reflects the actual achievements of students in innovation and entrepreneurship, including data on innovation and entrepreneurship competitions, intellectual property rights, project operation, and innovation and entrepreneurship practice experience. This type of data can be directly statistically and quantitatively analyzed. Implicit trait unstructured data: Non-standardized data that reflects students' intrinsic abilities and traits in innovation and entrepreneurship, specifically including text data, audio data, video data, and behavioral process data. This type of data needs to be processed in a targeted manner before it can be quantitatively analyzed.

[0024] Standardization preprocessing is performed on the multi-source heterogeneous data acquired in each acquisition period, according to data type, to eliminate noise, format differences, and invalid information in the data, resulting in standardized time-series datasets for each period. The preprocessing operations for different types of data are targeted, and the specific implementation is as follows: For structured data with explicit results: perform missing value imputation, outlier removal, format standardization, and categorical data coding operations in sequence to ensure the integrity, validity, and standardization of the data; For text data with implicit traits: perform garbled text cleaning, redundant content removal, format unification, and semantic normalization operations in sequence to eliminate invalid information in the text data and achieve semantic standardization; For audio data with implicit characteristics: perform environmental noise reduction, non-human voice segment trimming, and format unification operations in sequence to retain the effective information in the audio data and improve data usability; For video data with implicit characteristics: perform the following operations in sequence: blur frame removal, irrelevant segment cropping, audio-visual synchronization alignment, and format unification to ensure the image quality and information validity of the video data and achieve format standardization.

[0025] S2, Model Construction and Weight Allocation: This step, based on a standardized time-series dataset, completes the construction of a two-dimensional evaluation model and the scientific allocation of indicator weights, forming a standardized model system that can be directly used for quantitative scoring. The specific implementation consists of three parts: constructing the two-dimensional evaluation model, allocating indicator weights using the analytic hierarchy process (AHP), and quantitative scoring for each indicator, as detailed below: A two-dimensional evaluation model is constructed, covering both explicit achievements and implicit traits. Both dimensions have primary and secondary sub-indicators, and all indicators have clearly defined quantitative scoring ranges and judgment criteria, enabling a tiered and refined characterization of students' innovation and entrepreneurship abilities. The explicit results dimension includes four primary indicators: competition results, intellectual property rights, project operation, and innovation and entrepreneurship practice experience. Each primary indicator has corresponding secondary sub-indicators. Competition results correspond to the competition level, intellectual property rights correspond to the patent type, project operation corresponds to the revenue scale, and innovation and entrepreneurship practice experience corresponds to the training duration. Each secondary sub-indicator has specific judgment criteria based on the actual innovation and entrepreneurship scenario. Implicit Trait Dimension: Five primary indicators are set: innovative thinking, business acumen, team leadership, resilience, and communication skills. Each primary indicator has corresponding secondary sub-indicators. Innovative thinking corresponds to the degree of innovation, business acumen corresponds to the accuracy of market analysis, team leadership corresponds to coordination skills, resilience corresponds to risk management skills, and communication skills correspond to logical expression skills. Specific judgment criteria are set for each secondary sub-indicator based on the characteristics of unstructured data.

[0026] The Analytic Hierarchy Process (AHP) is used to assign weights to all indicators in the two-dimensional evaluation model, achieving a scientific weight allocation within and between the explicit outcome dimension and the implicit trait dimension. The specific implementation steps are as follows: Judgment matrices were constructed for the primary indicators and secondary sub-indicators in the dual-dimensional evaluation model. The construction of the judgment matrices combined the opinions of experts in the field of innovation and entrepreneurship with the actual needs of cultivating students' innovation and entrepreneurship abilities. Perform a consistency check on the constructed judgment matrix, remove matrices that fail the consistency check and reconstruct the matrix to ensure the rationality of the weight allocation; The eigenvectors of each indicator are calculated from the judgment matrix that passes the consistency test, and the eigenvectors are normalized to obtain the weights assigned to each indicator. Based on the normalization results, the weights within the explicit results dimension, the implicit traits dimension, and between the explicit results dimension and the implicit traits dimension are configured to form a complete indicator weight system.

[0027] The indicator-by-indicator quantitative scoring is based on the established standardized model system. It performs quantitative scoring on each indicator of the standardized time-series dataset for each period, using a scoring standard of 0-100 points. This yields quantitative results for both explicit outcome dimensions and implicit trait dimensions for each period. Specific scoring methods are differentiated according to data type. Score of structured data for explicit achievements: Based on the indicators and scoring rules of each level of explicit achievements in the standardized model system, and taking the actual innovation and entrepreneurship achievements generated by students in this period as the direct basis, we complete the standardized scoring of each secondary sub-indicator one by one, and then summarize the overall quantitative scores of the primary indicators and explicit achievements according to the weight system. Scoring of unstructured data with latent traits: By comparing the indicators and scoring rules of each level of the latent trait dimension in the standardized model system, different types of unstructured data are matched to the corresponding secondary sub-indicators. The standardized scoring of each secondary sub-indicator is completed through feature extraction and quantification transformation. Finally, the overall quantitative score of the primary indicator and the latent trait dimension is obtained by summarizing according to the weight system.

[0028] S3, Feature Fusion and Temporal Feature Library Generation: This step generates single-cycle feature vectors using a dual-dimensional temporal capability fusion algorithm and aggregates all periodic feature vectors to form a temporal feature library. This achieves a temporal and standardized representation of students' innovation and entrepreneurship capabilities. The specific implementation consists of three parts: dual-dimensional temporal capability fusion algorithm calculation, single-cycle standardized feature vector generation, and temporal feature library aggregation. (Details are as follows...) Figure 2 As shown: This invention employs a proprietary two-dimensional temporal capability fusion algorithm to fuse the quantization results of explicit and implicit dimensions. The mathematical expression of the two-dimensional temporal capability fusion algorithm is as follows: in, For the first Periodic single-period standardized feature vector, This is a two-dimensional adaptation coefficient, with a value ranging from 0.3 to 0.5. For the explicit results dimension Itemized indicator weights, For the first The first dimension of periodic explicit results Standardized scores for each indicator This is the time-series decay coefficient. For the latent trait dimension Itemized indicator weights, For the first Periodic latent trait dimension Standardized scores for each indicator , These represent the total number of sub-indicators for explicit and implicit dimensions, respectively.

[0029] The single-period standardized feature vector generation involves inputting the quantification results of the explicit achievement dimension and the implicit trait dimension of the same period into the aforementioned two-dimensional time-series capability fusion algorithm to calculate and generate a single-period standardized feature vector. The specific implementation steps are as follows: From the quantitative results of this period, extract the quantitative scores of all sub-indicators of the explicit results dimension, and arrange them in order of indicators to form an explicit quantitative feature sequence; Extract the quantitative scores of all sub-indicators of the latent trait dimension for this period, and arrange them in order of the indicators to form a latent quantitative feature sequence; Two feature sequences are simultaneously input into a two-dimensional temporal capability fusion algorithm. The algorithm combines preset two-dimensional adaptation coefficients, temporal decay coefficients, and assigned weights of each indicator to perform unified spatial mapping and weighted integration of the two feature sequences. After processing by the algorithm, a single-cycle standardized feature vector with fixed dimensions and numerical values ​​standardized to the 0-1 range is generated. This vector fully represents the full-dimensional state of students' innovation and entrepreneurship capabilities within this cycle.

[0030] The time-series feature library summarizes and organizes the standardized feature vectors of each collection period according to the order of collection time. All feature vectors are integrated and stored to form a time-series feature library of students' innovation and entrepreneurship capabilities. This feature library fully records the changes in students' innovation and entrepreneurship capabilities in each period, providing a time-series data analysis foundation for subsequent causal analysis and profile generation.

[0031] S4, Causal Analysis and Profile Generation: This step utilizes a time-series causal effect strength algorithm to mine the causal relationships between ability variables, generating causal chains for ability evolution. Based on this, it completes dynamic profile generation, root cause diagnosis of ability shortcomings, and prediction of ability evolution trends, achieving in-depth analysis of students' innovation and entrepreneurship abilities. The specific implementation consists of three parts: calculation using the time-series causal effect strength algorithm, generation of causal chains for ability evolution, and output of multi-dimensional analysis results. (Details follow...) Figure 3 As shown: This invention employs a proprietary time-series causal effect strength algorithm to calculate the causal transmission path and effect strength between various capability variables. The mathematical expression of the time-series causal effect strength algorithm is as follows: in, For the capability dimension right The intensity of the temporal causal effect, ranging from 0 to 1. For the first Cyclical capability dimension Standardized scores For the capability dimension Average score over the entire period The periodic correlation coefficient, As a causal transmission attenuation factor, This represents the total number of data collection cycles.

[0032] The generation of capability evolution causal chains involves inputting the time-series feature library into the aforementioned time-series causal effect strength algorithm to calculate the causal transmission paths and effect strengths between each capability variable, ultimately generating a visualized capability evolution causal chain. The specific implementation steps are as follows: The single-cycle standardized feature vectors of all periods in the time series feature library are arranged in chronological order, and the standardized scores of each capability variable are extracted from each vector to form the full-cycle time series data of each capability variable. The full-cycle time series data of each capability variable are input into the time series causal effect strength algorithm. The algorithm first performs a stationarity test on the time series data, removes invalid data with no stable change pattern, and retains the time series data with analytical value. Calculate the selected valid data to verify the time-series influence relationship between each capability variable and clarify the causal correspondence between each capability variable; From the validation results, pairs of capability variables with significant causal effects are selected. According to the direction of causal influence, a complete causal transmission path from upstream node to downstream node is formed, and the causal effect strength of each path is quantified by algorithm. By integrating all causal transmission paths with corresponding effect intensity data, a visualized causal chain of ability evolution is generated, which intuitively shows the causal relationship between various ability variables.

[0033] The multi-dimensional analysis results are based on the analysis of the causal chain of ability evolution. Combining the time-series change data of each ability dimension, the system sequentially generates a dynamic profile of students' innovation and entrepreneurship abilities, diagnoses the root causes of ability shortcomings, and predicts the trend of ability evolution. The specific implementation is as follows: Dynamic Ability Profile Generation: Combining the causal chain of ability evolution with the time-series change data of each ability dimension, a visual dynamic ability profile is drawn. The profile includes the time-series change curves of all indicators, the distribution of quantitative scores in each period, the score proportion of each ability dimension, etc., which intuitively shows the development and change process of students' innovation and entrepreneurship abilities. Diagnosis of the root causes of competency gaps: Starting from the indicators of students' shortcomings in innovation and entrepreneurship, we trace back along the causal chain of competency evolution to locate the upstream competency nodes corresponding to the shortcomings, and sort out the causal transmission path from the upstream nodes to the shortcomings, clarify the root causes of competency gaps, and form a complete diagnosis of the root causes of competency gaps. Capability Evolution Trend Prediction: Based on historical time-series data of each capability dimension and combined with the causal transmission relationship in the capability evolution causal chain, a time-series extrapolation method is used to predict the numerical change trend of each capability dimension within a set period in the future. At the same time, potential change nodes in the capability development process are identified to provide direction for subsequent intervention guidance.

[0034] S5, Intervention Plan Generation and Closed-Loop Iteration: This step determines the intervention order through a root cause node intervention priority algorithm, generates personalized entrepreneurship guidance service plans, and forms a closed-loop iterative evaluation and guidance system to achieve targeted improvement of students' innovation and entrepreneurship capabilities. The specific implementation consists of three parts: root cause node intervention priority algorithm calculation, personalized entrepreneurship guidance service plan generation, and closed-loop iterative execution, as detailed below: The root cause node intervention priority algorithm of this invention employs a proprietary algorithm to calculate the intervention priority value of each upstream node in the causal chain. The mathematical expression of the root cause node intervention priority algorithm is as follows: in, For the capability dimension The intervention priority value ranges from 0 to 1. For the capability dimension The strength of the total causal effect across all downstream dimensions For the capability dimension Current shortcomings and gaps For the capability dimension Average score over the entire period To determine the cost-effectiveness ratio of intervention, the algorithm inputs the effect intensity of each upstream node in the capability evolution causal chain into the algorithm to calculate the intervention priority value of each upstream node. The specific implementation steps are as follows: Extract all upstream root cause nodes from the causal chain of capability evolution as candidate nodes for this intervention; The causal effect strength of each upstream root cause node on all its downstream nodes is summarized, and the total causal effect strength of each upstream root cause node is obtained by summing. From the root cause diagnosis results of capability shortcomings and the time series feature library, obtain the current shortcoming gap and the average score of the whole cycle for each upstream root cause node; The total causal effect intensity, current shortcoming gap, average score over the whole cycle, and preset intervention cost-effectiveness coefficient of each upstream root cause node are simultaneously input into the root cause node intervention priority algorithm. The intervention priority value of each upstream root cause node is calculated by the algorithm formula. All values ​​are in the range of 0-1. The higher the value, the earlier the intervention order of the upstream root cause node.

[0035] The personalized entrepreneurship guidance service plan is generated based on the intervention priority values ​​of each upstream root cause node. The plans are arranged in descending order of value to create a targeted, personalized entrepreneurship guidance service plan for each student. The plan includes three parts: a phased root cause-based capability enhancement plan, a dedicated learning path, and a one-on-one mentorship plan. Each part is tailored to the student's specific capability weaknesses and the innovation and entrepreneurship track, as detailed below: The phased root cause capability enhancement plan is divided into three phases: short-term, medium-term, and long-term. Each phase has specific enhancement goals, targeted training tasks, and supporting innovation and entrepreneurship practice projects for the upstream root cause node with the highest priority. At the same time, specific acceptance criteria are set for each phase to ensure the effectiveness of capability enhancement. Personalized learning path: Based on the highest priority root cause of the ability gap, matching basic introductory, advanced and practical application learning content is selected from the innovation and entrepreneurship learning resource library to customize a personalized learning path for students, clarifying the learning cycle, core learning focus and specific assignment requirements of each stage, and setting stage assessment nodes to control the learning progress and effect. One-on-one mentorship program: Based on the student's entrepreneurial track and priority root cause improvement needs, the program matches the student with an innovation and entrepreneurship mentor in the corresponding industry field, clarifies the mentor's guidance frequency, and includes three guidance methods: online communication, offline tutoring, and project review. The specific guidance topics are determined in stages, and the core content and implementation process of each guidance session are clearly defined to achieve precise one-on-one guidance.

[0036] After the implementation of the closed-loop iterative execution personalized entrepreneurship guidance service plan, the next preset fixed cycle will be used as a new starting point to return to the data collection and preprocessing steps and repeat all the steps of this invention: In the new collection cycle, multi-source heterogeneous data on students' innovation and entrepreneurship after receiving guidance will be collected. Through subsequent steps, new ability evaluation, root cause diagnosis and trend prediction will be completed, and the personalized entrepreneurship guidance service plan will be iteratively optimized based on the new analysis results, forming a closed-loop system of "data collection - model evaluation - causal analysis - intervention guidance - data re-collection", realizing the full-cycle dynamic evaluation and continuous personalized improvement of students' innovation and entrepreneurship capabilities.

[0037] The evaluation method based on the student innovation and entrepreneurship ability model of this invention completes the full-cycle collection and standardized processing of multi-source heterogeneous data through a combination of fixed and dynamic methods, constructs a two-dimensional evaluation model and realizes scientific indicator weight allocation, uses a dedicated algorithm to complete feature fusion and causal relationship mining, generates a visualized ability evolution causal chain and realizes multi-dimensional ability analysis, and finally combines the root cause node intervention priority to generate personalized guidance plans and form a closed-loop iterative system. This achieves refined and dynamic evaluation and root cause-based and personalized improvement of students' innovation and entrepreneurship abilities, effectively improving the pertinence and effectiveness of students' innovation and entrepreneurship ability cultivation, and providing scientific methodological support and implementation path for the development of innovation and entrepreneurship education in universities.

[0038] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An evaluation method based on a student innovation and entrepreneurship ability model, characterized in that, The specific steps of this method are as follows: S1, Data Acquisition and Preprocessing: Collect multi-source heterogeneous data on student innovation and entrepreneurship according to a preset fixed period, and perform standardized preprocessing on the data of each period to obtain the standardized time series dataset corresponding to each period. S2, Model Construction and Weight Allocation: Based on the standardized time-series dataset, a two-dimensional evaluation model covering explicit outcome dimensions and implicit trait dimensions is constructed. The weights of each indicator are allocated through the analytic hierarchy process to form a standardized model system. S3, Feature Fusion and Temporal Feature Library Generation: Based on the standardized model system, the data of each period are quantitatively scored for each indicator. The quantitative results of the explicit achievement dimension and the implicit trait dimension of the same period are input into the dual-dimensional temporal capability fusion algorithm to calculate and generate a single-period standardized feature vector. All period feature vectors are summarized to form a temporal feature library. S4, Causal Analysis and Profile Generation: Input the time-series feature library into the time-series causal effect strength algorithm to calculate the causal transmission path and effect strength between each ability variable, generate the ability evolution causal chain, and generate a dynamic profile of students' innovation and entrepreneurship ability, diagnosis of the root causes of ability shortcomings, and prediction of ability evolution trends based on the causal chain analysis results. S5, Intervention plan generation and closed-loop iteration: Input the effect intensity of each upstream node in the causal chain of the capability evolution into the root cause node intervention priority algorithm, calculate the intervention priority value of each upstream node, generate a personalized entrepreneurship guidance service plan based on the priority value, and repeat steps S1 to S5 with the next preset fixed period as the new starting point.

2. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S1, the collection of multi-source heterogeneous data on student innovation and entrepreneurship according to a preset fixed period specifically means: the preset fixed period is a natural quarter, or it can be dynamically adjusted according to the student's entrepreneurial stage. The collection period is once a month in the initial stage and once a natural quarter in the mature stage. The collected multi-source heterogeneous data includes explicit achievement-type structured data and implicit characteristic-type unstructured data. Explicit achievement-type structured data includes innovation and entrepreneurship competition data, intellectual property data, project operation data, and innovation and entrepreneurship practice experience data. Implicit characteristic-type unstructured data includes text data, audio data, video data, and behavioral process data.

3. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S1, the standardization preprocessing performed on the data for each period specifically includes: for explicit result-type structured data, missing value imputation, outlier removal, format unification, and categorized data encoding; for implicit characteristic-type text data, garbled character cleaning, redundant content removal, format unification, and semantic normalization; for implicit characteristic-type audio data, environmental noise reduction, non-human voice segment clipping, and format unification; and for implicit characteristic-type video data, blurred frame removal, irrelevant segment clipping, audio-visual synchronization alignment, and format unification.

4. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S2, the dual-dimensional evaluation model covering both explicit and implicit trait dimensions is as follows: The explicit results dimension includes four primary indicators: competition results, intellectual property, project operation, and innovation and entrepreneurship practice experience. The competition results primary indicator corresponds to a secondary sub-indicator of competition level; the intellectual property primary indicator corresponds to a secondary sub-indicator of patent type; the project operation primary indicator corresponds to a secondary sub-indicator of revenue scale; and the innovation and entrepreneurship practice experience primary indicator corresponds to a secondary sub-indicator of training duration. The implicit trait dimension includes five primary indicators: innovative thinking, business acumen, team leadership, resilience, and communication skills. The innovative thinking primary indicator corresponds to a secondary sub-indicator of innovation degree; the business acumen primary indicator corresponds to a secondary sub-indicator of market analysis accuracy; the team leadership primary indicator corresponds to a secondary sub-indicator of coordination ability; the resilience primary indicator corresponds to a secondary sub-indicator of risk management ability; and the communication skills primary indicator corresponds to a secondary sub-indicator of logical expression ability. Each indicator has a clearly defined quantitative scoring range and judgment criteria.

5. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S2, the allocation of weights for each indicator using the analytic hierarchy process specifically involves: constructing judgment matrices for the first-level indicators and second-level sub-indicators in the dual-dimensional evaluation model, performing a consistency check on the judgment matrices, calculating the eigenvectors of each indicator after passing the consistency check, normalizing the eigenvectors to obtain the corresponding weights, and completing the weight allocation within and between the explicit outcome dimension and the implicit trait dimension based on the normalization results.

6. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S3, the mathematical expression of the two-dimensional temporal capability fusion algorithm is: in, For the first Periodic single-period standardized feature vector, This is a two-dimensional adaptation coefficient, with a value ranging from 0.3 to 0.

5. For the explicit results dimension Itemized indicator weights, For the first The first dimension of periodic explicit results Standardized scores for each indicator This is the time-series decay coefficient. For the latent trait dimension Itemized indicator weights, For the first Periodic latent trait dimension Standardized scores for each indicator , These represent the total number of sub-indicators for explicit and implicit dimensions, respectively.

7. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S4, the mathematical expression of the time-series causal effect strength algorithm is: in, For the capability dimension right The intensity of the temporal causal effect, ranging from 0 to 1. For the first Cyclical capability dimension Standardized scores For the capability dimension Average score over the entire period The periodic correlation coefficient, As a causal transmission attenuation factor, This represents the total number of data collection cycles.

8. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S4, the process of generating a dynamic profile of students' innovation and entrepreneurship capabilities, diagnosing the root causes of capability shortcomings, and predicting capability evolution trends based on the causal chain analysis results specifically involves: combining the capability evolution causal chain with the time-series change data of each capability dimension to draw a dynamic profile of capabilities that includes the time-series change curves of all indicators and the periodic quantitative score distribution; tracing back along the capability evolution causal chain to locate the upstream capability nodes corresponding to the shortcomings and sorting out the transmission path to form the content of capability shortcoming root cause diagnosis; and based on historical time-series data and causal transmission relationships, inferring the change trends of each capability dimension within a future set period to determine potential change nodes in the capability development process.

9. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S5, the mathematical expression of the root cause node intervention priority algorithm is: in, For the capability dimension The intervention priority value ranges from 0 to 1. For the capability dimension The overall causal effect strength across all downstream dimensions For the capability dimension Current shortcomings and gaps For the capability dimension Average score over the entire period To intervene in the cost-effectiveness ratio.

10. The evaluation method based on a student innovation and entrepreneurship ability model according to claim 1, characterized in that, In step S5, the personalized entrepreneurship guidance service plan generated based on the intervention priority values ​​of each upstream node specifically includes: a phased root cause capability enhancement plan divided into three preset phases: short-term, medium-term, and long-term; each phase clearly defines the enhancement goals, targeted training tasks, supporting innovation and entrepreneurship practice projects, and specific acceptance criteria for the highest priority upstream root cause node; a dedicated learning path that selects and matches basic introductory, advanced, and practical application learning content based on the capability gaps of the highest priority root cause node, specifying the learning cycle, core learning focus, specific assignment requirements, and phased assessment nodes for each phase; and a one-on-one mentorship plan that matches mentors in corresponding industry fields based on the student's entrepreneurial track and the enhancement needs of the priority root cause node, specifying guidance at a preset frequency, including online communication, offline tutoring, and project review, determining specific guidance themes in each phase, and clarifying the core content and process of each guidance session.

Images