A method and system for automatically generating a comprehensive performance of a college student based on multi-source data fusion and process evaluation

Abstract

The application discloses a kind of based on multi-source data fusion and process evaluation's college student comprehensive performance automatic generation method and system, including S1 establishes ideological and political evaluation model, academic performance evaluation model, practical ability evaluation model, comprehensive quality evaluation model, development potential evaluation model comment model. S2 by middleware establishes standardization interface and automatically reads the heterogeneous data source data of educational administration system, school system, campus card system, library system, scientific research and teaching research practical training system, second classroom activity platform etc.S3 to behavior sequence is carried out pattern recognition and feature extraction, establishes student panoramic data warehouse, and the data of step S2 is labeled with S1 evaluation model label and storage.S4 user selects comment purpose.S5 dynamic weight distribution.S6 to student data is screened.S7 comment automatic generation.S8 artificial check comment.S9 generation visual report.The application improves work efficiency, and can provide accurate student performance in school for parents and employing units.

Description

Technical Field

[0001] This invention relates to a method and system for automatically generating comprehensive performance data of college students based on multi-source data fusion and process evaluation, in the field of information technology. Background Technology

[0002] In university student management, achieving comprehensive, fair, and developmental student evaluation is a current challenge for universities. Existing evaluation systems heavily rely on manual assessments by counselors and instructors, resulting in inconsistent evaluation standards, significant subjective influence, and difficulty in ensuring objectivity and fairness. The evaluation process requires manual collection, organization, and analysis of various student performance data, which is time-consuming, labor-intensive, and inefficient, especially when dealing with large student groups, making the evaluation work a heavy management burden. Current evaluations are mostly end-of-term or year-end summaries, failing to track students' learning and growth in real-time and continuously, and failing to reflect students' dynamic development and stage-specific changes. Student data is scattered across multiple independent systems such as academic affairs, student affairs, student ID cards, and the library, lacking effective correlation and integration, making it difficult to form a comprehensive and multi-dimensional student profile. Even when some systems can share basic data, they can only achieve simple data aggregation and display, lacking the ability to automatically extract features from multi-source data, quantify behavior, and generate intelligent analysis and evaluation conclusions. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings and deficiencies of the existing technologies mentioned above by proposing an automatic method and system for generating comprehensive performance data of college students based on multi-source data fusion and process evaluation. This method can automatically generate evaluation results based on students' comprehensive performance at school, reducing the workload of counselors, improving work efficiency, and providing parents and employers with accurate information on students' performance at school.

[0004] The technical solution adopted by this invention to solve its technical problem is: a method and system for automatically generating the comprehensive performance of college students based on multi-source data fusion and process evaluation. Its core technical solution adopts a three-layer architecture, as follows:

[0005] First layer: Automated acquisition and fusion layer of multi-source data

[0006] Establish standardized interfaces with heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform to realize the automated and periodic collection of multi-dimensional data such as student course grades, attendance, rewards and punishments, consumption, borrowing, experiments, and social practice.

[0007] Design a unified student data model, clean, denoise, normalize, and correlate the collected raw data to form a panoramic data warehouse centered on individual students.

[0008] Let the total number of heterogeneous data sources be n, and the original dataset of the i-th data source be D. i ={ , , …, }, i=1,2, …,n; Define the normalized mapping function f std : -> This function satisfies the requirements of data source heterogeneity adaptation rules and data dimension uniformity, through the formula. ={f std ( ), f std ( ), …, f std ( This yields the standardized dataset from the i-th data source. The standardized dataset is then processed. Set its reasonable threshold range as [ ], , Determined by the characteristics of the corresponding data source; define the cleaning function f. clean : -> Through formula The cleaned data is obtained, and the cleaned data set of the i-th data source is: ={ Set the student's unique identifier as ID. k Let k = 1, 2, …, K, where K is the total number of students. Define a data association function f. rel :{ }×ID k ->DW k This function uses a student's unique identifier as the association key to achieve the fusion and association of multi-source cleaned data, through the formula DW. k =f rel ({ }, ID k We obtain the panoramic data warehouse of the kth student.

[0009] Second layer: Intelligent analysis and model calculation layer

[0010] Behavioral Data Quantification Model: This model constructs quantitative sub-models for multiple evaluation dimensions, including academic performance, behavioral norms, social practice, innovation ability, and physical and mental health. Unstructured behavioral records (such as participation in activities and volunteer service) are transformed into computable structured indicators through pre-defined rules and machine learning methods. The comprehensive evaluation dimension set is defined as M = {M1, M2, M3, M4, M5, ...}, corresponding to evaluation dimensions such as academic performance, behavioral norms, social practice, innovation ability, and physical and mental health, respectively; the unstructured data of the k-th student in the m-th dimension is represented by U. km m = 1, 2, 3, 4, 5, ...; Define the quantization function f. qua U km ->X km This function integrates preset rules and machine learning algorithms to transform unstructured data into structured indicators, through formula X. km =f qua (U km We obtain the quantitative index value of the kth student in the m-th dimension.

[0011] Dynamic weight allocation module: Introduces a dynamic weight adjustment algorithm based on time decay, development stage (e.g., freshman to senior year), professional characteristics, and college requirements. The contribution weights of data from different evaluation dimensions and time intervals to the final comprehensive evaluation can be dynamically adjusted, making the evaluation more personalized and targeted. The time decay factor α(t), development stage factor β(s), professional characteristics factor γ(p), and college requirements factor δ(q) are used. Let the initial base weight of the m-th evaluation dimension be... ,satisfy

[0012] In the formula, n represents the number of evaluation dimensions. This is achieved through the formula w. km (t,s,p,q)= ɑ(t) ×β(s) ×γ(p) ×δ(q)× Calculate the dynamic weight of the k-th student in the m-th dimension, where the time decay factor adopts an exponential decay model: α(t) = λ is the attenuation coefficient, t0 is the data collection start time, and t is the current calculation time. The development stage factor β(s) is assigned a value according to the student's grade level, the professional characteristic factor γ(p) is assigned a value according to the professional training objectives, and the college requirement factor δ(q) is assigned a value according to the college's evaluation rules. The value range of each factor is [0,1], and the dynamic weights satisfy the normalization constraint.

[0013]

[0014] In the formula, n is the number of evaluation dimensions.

[0015] Process-oriented assessment engine: Continuously calculates the time-series changes of students' various indicators, generating development trend curves and periodic snapshots to achieve continuous tracking and evaluation of students' growth process. The assessment time series is defined as T={t1,t2, …,t…} l The k-th student at time t j The value of the m-th dimension quantification indicator is X km (t j ), j=1,2, …,l; through the formula

[0016]

[0017] Calculate the time point t for the k-th student j The comprehensive evaluation value; based on the time-series comprehensive evaluation value sequence {S k (t1), S k (t2), …, S k (t l Define the trend fitting function f. trend Through the formula Fk(t) = f trend ( {S k (t1), S k (t2),…, S k (t l Generate a function to show the student's growth and development trend.

[0018] Third layer: Report generation and human-computer collaboration layer

[0019] Automatic comment generation module: Based on natural language processing (NLP) technology, it automatically organizes the quantitative indicators, trend analysis and comparison results (such as personal historical comparison and comparison with the same professional group) calculated by the model into coherent, natural and insightful text comments, and supports the generation of periodic reports and comprehensive reports.

[0020] Visualized reporting system: Presents students' overall performance data, trends, radar charts, comparative analysis, etc., in intuitive forms such as charts and dashboards to form easy-to-interpret visual growth reports.

[0021] Human-machine collaborative feedback optimization mechanism: A feedback interface is designed to allow teachers and counselors to review, fine-tune, and supplement automatically generated reports and comments. The system uses human feedback as optimization data to adjust the quantitative model parameters and comment generation rules, enabling continuous iteration and optimization of the system.

[0022] To achieve the above objectives, the technical solution adopted in this invention is an automatic generation method for the comprehensive performance of college students based on multi-source data fusion and process evaluation, comprising the following steps:

[0023] S1: Establish evaluation models that reflect the evaluation criteria, such as ideological and political evaluation model, academic performance evaluation model, practical ability evaluation model, comprehensive quality evaluation model, and development potential evaluation model;

[0024] S2: Automatically reads data from heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform through middleware to establish standardized interfaces;

[0025] S3: Perform pattern recognition and feature extraction on the behavioral sequences, establish a student panoramic data warehouse, and label the data from S2 with the evaluation model from S1 and store it.

[0026] S4: Users (counselors, employers, parents) select the purpose of the comments;

[0027] S5: Dynamic weight allocation (for scholarship evaluation, academic performance has a higher weight; for job recommendation evaluation, internships and practical experience have a higher weight. The system allows administrators to configure or train weight models for different scenarios using machine learning.)

[0028] S6: Filter student data and perform a collision analysis with the S1 evaluation model based on the weights (e.g., build a local large model knowledge base and call the local general large model).

[0029] S7: Automatic Comment Generation

[0030] S8: Manually verify comments. Comments that do not meet the requirements are returned to S5, and comments that meet the requirements are entered into S9.

[0031] S9: Generate Visual Reports

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] Multi-source data automated acquisition: Enables seamless connection and automatic aggregation of data across systems.

[0034] Behavioral data quantification model: transforms complex behaviors into calculable and comparable objective indicators.

[0035] Dynamic weight allocation: This enables the evaluation system to adapt to different scenarios and individual differences, making it more scientific and flexible.

[0036] Automatic comment generation: Utilizing NLP technology to achieve a qualitative leap from "data" to "insight" and produce high-quality text evaluations.

[0037] Human-machine collaborative feedback optimization: forming a closed loop of "automatic machine generation - artificial intelligence intervention - machine self-learning" to continuously improve the accuracy and practicality of system evaluation.

[0038] 3. The technical effects achieved by this technical solution (i.e., what effects can be obtained after solving the problem by adopting this method).

[0039] After implementing the technical solution described in this invention, the following significant technical effects can be achieved:

[0040] (1) Greatly improves evaluation efficiency and objectivity:

[0041] Automated data collection and processing replaces most of the manual collection and organization work, and the evaluation generation process is fast and efficient, with automatic output in real time or periodically.

[0042] Calculations based on a unified quantitative model and objective data sources minimize the bias of subjective judgments, making the evaluation results more fair and comparable.

[0043] (2) Achieve dynamic comprehensive evaluation of the entire process and multiple dimensions:

[0044] By continuously collecting and analyzing student data across the entire lifecycle and multiple scenarios, a fundamental shift from "summative assessment" to "formational assessment" has been achieved, clearly reflecting students' growth trajectory and effort process.

[0045] By integrating data from multiple dimensions, including academics, behavior, practice, and innovation, a comprehensive student profile is constructed, resulting in a more holistic evaluation that avoids "grades-only" evaluation.

[0046] (3) Produce visualized and interpretable intelligent reports to support students' personalized development:

[0047] The generated comprehensive report not only contains data, but also provides insightful conclusions and growth suggestions from intelligent analysis, enabling students to clearly understand their own strengths, weaknesses and development trends.

[0048] Intuitive and visual reports facilitate understanding and communication among students, parents, and teachers, providing clear data support and directional guidance for students' personalized planning and improvement.

[0049] (4) Provide data support for the school's refined management and scientific decision-making:

[0050] The system generates group analysis data (such as the overall development trend of the major, the evaluation of the effectiveness of activities, and the distribution of early warning indicators) which can provide real and in-depth data insights for departments and schools to make decisions on teaching management, academic atmosphere construction, and resource allocation, and promote the transformation of management towards refinement and scientification.

[0051] Beneficial effects:

[0052] 1. This invention features automated multi-source data collection, achieving seamless integration and automatic aggregation of cross-system data. Its behavioral data quantification model transforms complex behaviors into calculable and comparable objective indicators. Dynamic weight allocation enables the evaluation system to adapt to different scenarios and individual differences, making it more scientific and flexible. Automatic comment generation utilizes NLP technology to achieve a qualitative leap from "data" to "insight," producing high-quality textual evaluations. Human-machine collaborative feedback optimization forms a closed loop of "automatic machine generation - artificial intelligence intervention - machine self-learning," continuously improving the accuracy and practicality of the system's evaluations.

[0053] 2. This invention greatly improves evaluation efficiency and objectivity, automating data collection and processing to replace most manual data collection and organization work. The evaluation generation process is fast and efficient, and can be automatically output in real time or periodically. Calculations based on a unified quantitative model and objective data sources minimize the bias of subjective judgment, making the evaluation results more fair and comparable.

[0054] 3. It enables dynamic and comprehensive evaluation across the entire process and multiple dimensions. Through continuous collection and analysis of student data throughout the entire lifecycle and across multiple scenarios, it achieves a fundamental shift from "summative evaluation" to "formational evaluation," clearly reflecting students' growth trajectory and efforts. By integrating data from multiple dimensions such as academics, behavior, practice, and innovation, it constructs a three-dimensional student profile, making the evaluation more comprehensive and avoiding "grades-only" evaluation.

[0055] 4. This invention produces visualized and interpretable intelligent reports to support students' personalized development. The comprehensive reports generated not only contain data but also provide insightful conclusions and growth suggestions from intelligent analysis, enabling students to clearly understand their strengths, weaknesses, and development trends. The intuitive and visualized reports facilitate understanding and communication among students, parents, and teachers, providing clear data support and directional guidance for students' personalized planning and improvement.

[0056] 5. Provide data support for the school's refined management and scientific decision-making. The group analysis data generated by the system (such as the overall development trend of majors, evaluation of activity effects, distribution of early warning indicators, etc.) can provide real and in-depth data insights for the teaching management, academic atmosphere construction, resource allocation and other decisions of departments and schools, and promote the transformation of management towards refinement and scientification. Attached Figure Description

[0057] Figure 1 This is a flowchart of the method of the present invention.

[0058] Figure 2 This is a system architecture diagram of the present invention. Detailed Implementation

[0059] The invention will now be described in further detail with reference to the accompanying drawings.

[0060] like Figure 2 As shown, this invention provides an automatic generation method and system for the comprehensive performance of college students based on multi-source data fusion and process evaluation. Its core technical solution adopts a three-layer architecture, as detailed below:

[0061] First layer: Automated acquisition and fusion layer of multi-source data

[0062] Establish standardized interfaces with heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform to realize the automated and periodic collection of multi-dimensional data such as student course grades, attendance, rewards and punishments, consumption, borrowing, experiments, and social practice.

[0063] Design a unified student data model, clean, denoise, normalize, and correlate the collected raw data to form a panoramic data warehouse centered on individual students.

[0064] Let the total number of heterogeneous data sources be n, and the original dataset of the i-th data source be D. i ={ , , …, }, i=1,2, …,n; Define the normalized mapping function f std : -> This function satisfies the requirements of data source heterogeneity adaptation rules and data dimension uniformity, through the formula. ={f std ( ), f std ( ), …, f std ( This yields the standardized dataset from the i-th data source. The standardized dataset is then processed. Set its reasonable threshold range as [ ], , Determined by the characteristics of the corresponding data source; define the cleaning function f. clean : -> Through formula The cleaned data is obtained, and the cleaned data set of the i-th data source is: ={ Set the student's unique identifier as ID. k Let k = 1, 2, …, K, where K is the total number of students. Define a data association function f. rel :{ }×ID k ->DW kThis function uses a student's unique identifier as the association key to achieve the fusion and association of multi-source cleaned data, through the formula DW. k =f rel ({ }, ID k We obtain the panoramic data warehouse of the kth student.

[0065] Second layer: Intelligent analysis and model calculation layer

[0066] Behavioral Data Quantification Model: This model constructs quantitative sub-models for multiple evaluation dimensions, including academic performance, behavioral norms, social practice, innovation ability, and physical and mental health. Unstructured behavioral records (such as participation in activities and volunteer service) are transformed into computable structured indicators through pre-defined rules and machine learning methods. The comprehensive evaluation dimension set is defined as M = {M1, M2, M3, M4, M5, ...}, corresponding to evaluation dimensions such as academic performance, behavioral norms, social practice, innovation ability, and physical and mental health, respectively; the unstructured data of the k-th student in the m-th dimension is represented by U. km m = 1, 2, 3, 4, 5, ...; Define the quantization function f. qua U km ->X km This function integrates preset rules and machine learning algorithms to transform unstructured data into structured indicators, through formula X. km =f qua (U km We obtain the quantitative index value of the kth student in the m-th dimension.

[0067] Dynamic weight allocation module: Introduces a dynamic weight adjustment algorithm based on time decay, development stage (e.g., freshman to senior year), professional characteristics, and college requirements. The contribution weights of data from different evaluation dimensions and time intervals to the final comprehensive evaluation can be dynamically adjusted, making the evaluation more personalized and targeted. The time decay factor α(t), development stage factor β(s), professional characteristics factor γ(p), and college requirements factor δ(q) are used. Let the initial base weight of the m-th evaluation dimension be... ,satisfy

[0068] In the formula, n represents the number of evaluation dimensions. This is achieved through the formula w. km (t,s,p,q)= ɑ(t) ×β(s) ×γ(p) ×δ(q)× Calculate the dynamic weight of the k-th student in the m-th dimension, where the time decay factor adopts an exponential decay model: α(t) = λ is the attenuation coefficient, t0 is the data collection start time, and t is the current calculation time. The development stage factor β(s) is assigned a value according to the student's grade level, the professional characteristic factor γ(p) is assigned a value according to the professional training objectives, and the college requirement factor δ(q) is assigned a value according to the college's evaluation rules. The value range of each factor is [0,1], and the dynamic weights satisfy the normalization constraint.

[0069]

[0070] In the formula, n is the number of evaluation dimensions.

[0071] Process-oriented assessment engine: Continuously calculates the time-series changes of students' various indicators, generating development trend curves and periodic snapshots to achieve continuous tracking and evaluation of students' growth process. The assessment time series is defined as T={t1,t2, …,t…} l The k-th student at time t j The value of the m-th dimension quantification indicator is X km (t j ), j=1,2, …,l; through the formula

[0072]

[0073] Calculate the time point t for the k-th student j The comprehensive evaluation value; based on the time-series comprehensive evaluation value sequence {S k (t1), S k (t2), …, S k (t l Define the trend fitting function f. trend Through the formula Fk(t) = f trend ( {S k (t1), S k (t2),…, S k (t l Generate a function to show the student's growth and development trend.

[0074] Third layer: Report generation and human-computer collaboration layer

[0075] Automatic comment generation module: Based on natural language processing (NLP) technology, it automatically organizes the quantitative indicators, trend analysis and comparison results (such as personal historical comparison and comparison with the same professional group) calculated by the model into coherent, natural and insightful text comments, and supports the generation of periodic reports and comprehensive reports.

[0076] Visualized reporting system: Presents students' overall performance data, trends, radar charts, comparative analysis, etc., in intuitive forms such as charts and dashboards to form easy-to-interpret visual growth reports.

[0077] Human-machine collaborative feedback optimization mechanism: A feedback interface is designed to allow teachers and counselors to review, fine-tune, and supplement automatically generated reports and comments. The system uses human feedback as optimization data to adjust the quantitative model parameters and comment generation rules, enabling continuous iteration and optimization of the system.

[0078] like Figure 1 As shown, this invention provides an automatic method for generating comprehensive performance data of college students based on multi-source data fusion and process evaluation, comprising the following steps:

[0079] S1: Establish evaluation models that reflect the evaluation criteria, such as ideological and political evaluation model, academic performance evaluation model, practical ability evaluation model, comprehensive quality evaluation model, and development potential evaluation model;

[0080] S2: Automatically reads data from heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform through middleware to establish standardized interfaces;

[0081] S3: Perform pattern recognition and feature extraction on the behavioral sequences, establish a student panoramic data warehouse, and label the data from S2 with the evaluation model from S1 and store it.

[0082] S4: Users (counselors, employers, parents) select the purpose of the comments;

[0083] S5: Dynamic weight allocation (for scholarship evaluation, academic performance has a higher weight; for job recommendation evaluation, internships and practical experience have a higher weight. The system allows administrators to configure or train weight models for different scenarios using machine learning.)

[0084] S6: Filter student data and perform a collision analysis with the S1 evaluation model based on the weights (e.g., build a local large model knowledge base and call the local general large model).

[0085] S7: Automatic Comment Generation

[0086] S8: Manually verify comments. Comments that do not meet the requirements are returned to S5, and comments that meet the requirements are entered into S9.

[0087] S9: Generate Visual Reports

[0088] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An automatic system for generating comprehensive performance data of college students based on multi-source data fusion and process evaluation, characterized in that, The system includes: First layer: Automated acquisition and fusion of multi-source data; Establish standardized interfaces with heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform to realize the automated and periodic collection of multi-dimensional data such as student course grades, attendance, rewards and punishments, consumption, borrowing, experiments, and social practice. Design a unified student data model, clean, denoise, normalize and correlate the collected raw data to form a panoramic data warehouse centered on individual students; Let the total number of heterogeneous data sources be n, and the original dataset of the i-th data source be D. i ={ , , …, }, i=1,2, …,n; Define the normalized mapping function f std : -> This function satisfies the requirements of data source heterogeneity adaptation rules and data dimension uniformity, through the formula. ={f std ( ), f std ( ), …, f std ( )} to obtain the standardized dataset of the i-th data source, and then... Set its reasonable threshold range as [ ], , Determined by the characteristics of the corresponding data source; define the cleaning function f. clean : -> Through the formula: The cleaned data is obtained, and the cleaned data set of the i-th data source is: ={ Set the student's unique identifier as ID. k Let k = 1, 2, …, K, where K is the total number of students. Define a data association function f. rel :{ }×ID k ->DW k This function uses a student's unique identifier as the association key to achieve the fusion and association of multi-source cleaned data, through the formula DW. k =f rel ({ }, ID k ) Obtain the panoramic data warehouse of the k-th student; The second layer: Intelligent analysis and model calculation layer; Behavioral Data Quantification Model: This model constructs quantitative sub-models for multiple evaluation dimensions, including academic performance, behavioral norms, social practice, innovation ability, and physical and mental health. Unstructured behavioral records, such as participation in activities and volunteer service, are transformed into computable structured indicators through pre-defined rules and machine learning methods. The comprehensive evaluation dimension set is defined as M = {M1, M2, M3, M4, M5, ...}, corresponding to the evaluation dimensions of academic performance, behavioral norms, social practice, innovation ability, and physical and mental health, respectively. The unstructured data of the k-th student in the m-th dimension is U. km m = 1, 2, 3, 4, 5, ...; Define the quantization function f. qua U km ->X km This function integrates preset rules and machine learning algorithms to transform unstructured data into structured indicators, through formula X. km =f qua (U km This yields the quantitative index value of the k-th student in the m-th dimension. Dynamic weight allocation module: Introduces a dynamic weight adjustment algorithm based on time decay, development stage (freshman to senior year), professional characteristics, and college requirements. The contribution weight of data from different evaluation dimensions and time intervals to the final comprehensive evaluation can be dynamically adjusted, making the evaluation more personalized and directional. The time decay factor α(t), development stage factor β(s), professional characteristics factor γ(p), and college requirements factor δ(q) are used. Let the initial base weight of the m-th evaluation dimension be... ,satisfy: In the formula, n is the number of evaluation dimensions, which is expressed by the formula w. km (t,s,p,q)= ɑ(t) ×β(s) ×γ(p) ×δ(q) × Calculate the dynamic weight of the k-th student in the m-th dimension, where the time decay factor adopts an exponential decay model: α(t) = λ is the attenuation coefficient, t0 is the data collection start time, and t is the current calculation time. The development stage factor β(s) is assigned a value according to the student's grade level, the professional characteristic factor γ(p) is assigned a value according to the professional training objectives, and the college requirement factor δ(q) is assigned a value according to the college's evaluation rules. The value range of each factor is [0,1], and the dynamic weights satisfy the normalization constraint. In the formula, n is the number of evaluation dimensions; Process-oriented assessment engine: Continuously calculates the time-series changes of students' various indicators, generates development trend curves and periodic snapshots, enabling continuous tracking and evaluation of students' growth process. The evaluation time series is set as T={t1,t2, …,t l The k-th student at time t j The value of the m-th dimension quantification indicator is X km (t j ), j=1,2, …,l; through the formula: Calculate the time point t for the kth student j The comprehensive evaluation value; based on the time-series comprehensive evaluation value sequence {S k (t1), S k (t2),…, S k (t l Define the trend fitting function f. trend Through the formula Fk(t) = f trend ( {S k (t1), S k (t2), …, S k (t l Generate a function to show student growth and development trends; The third layer: Report generation and human-computer collaboration layer; Automatic comment generation module: Based on natural language processing (NLP) technology, it automatically organizes the quantitative indicators, trend analysis, and comparison results calculated by the model, such as personal historical comparison and comparison with the same professional group, into coherent, natural, and insightful text comments, and supports the generation of periodic reports and comprehensive reports. Visualized reporting system: Presents students' overall performance data, trends, radar charts, comparative analysis, etc. in an intuitive form using charts and dashboards, forming an easy-to-interpret visual growth report; Human-machine collaborative feedback optimization mechanism: A feedback interface is designed to allow teachers and counselors to review, fine-tune, and supplement the automatically generated reports and comments. The system uses human feedback as optimization data to adjust the quantitative model parameters and comment generation rules, thereby achieving continuous iteration and optimization of the system.

2. A method for automatically generating comprehensive performance data of college students based on multi-source data fusion and process evaluation, characterized in that, The method includes the following steps: Step S1: Establish evaluation models for ideological and political assessment, academic performance assessment, practical ability assessment, comprehensive quality assessment, and development potential assessment, among others; Step S2: Establish a standardized interface through middleware to automatically read data from heterogeneous data sources such as the academic affairs system, student affairs system, campus card system, library system, scientific research and teaching training experiment system, and extracurricular activity platform; Step S3: Perform pattern recognition and feature extraction on the behavioral sequences, establish a student panoramic data warehouse, and label the data from S2 with the evaluation model from S1 and store it. Step S4: The user (counselor, employer, parent) selects the purpose of the evaluation; Step S5: Dynamic weight allocation (for scholarship evaluation, academic performance has a higher weight; for job recommendation evaluation, internships and practical experience have a higher weight. The system allows administrators to configure or train weight models for different scenarios through machine learning). Step S6: Filter the student data and perform a collision test with the S1 evaluation model based on the weights (e.g., build a local large model knowledge base and call the local general large model). Step S7: Comments are automatically generated; Step S8: Manually verify the comments. Comments that do not meet the requirements are returned to S5, and comments that meet the requirements are entered into S9. Step S9: Generate a visualization report.

Images