Search

CN-121981858-A - Educational intelligent evaluation method integrating knowledge patterns

CN121981858ACN 121981858 ACN121981858 ACN 121981858ACN-121981858-A

Abstract

The invention relates to the technical field of education and discloses an intelligent education evaluation method integrating knowledge graphs, which comprises the following steps of carrying out multi-source data acquisition on an education scene to construct a knowledge graph comprising concept entities, value association relations and education target attributes; the method comprises the steps of carrying out semantic enhancement processing on a knowledge graph, extracting multidimensional features of text emotion, behavior patterns and cognition level, establishing an evaluation index association model, analyzing educational effect dimensionality and time-related influence weights, identifying an evaluation action range according to target achievement dynamic characteristic data, coupling educational element patterns to generate educational action coupling data, and constructing a machine learning evaluation model to realize intelligent evaluation and real-time optimization of educational effects. According to the method, through dynamic fusion of the knowledge graph and the multidimensional features, the problems of strong subjectivity and single data of a traditional evaluation method are solved, and the accuracy and timeliness of evaluation are improved.

Inventors

  • WANG XIAOXUAN
  • YAO LI
  • LI YANGYANG

Assignees

  • 安徽审计职业学院

Dates

Publication Date
20260505
Application Date
20250710

Claims (10)

  1. 1. The intelligent education evaluating method integrating the knowledge graph is characterized by comprising the following steps of: The method comprises the steps of A1, carrying out multi-source data acquisition on an education scene to obtain original evaluation data comprising course texts, activity records and interaction logs, and constructing an education knowledge graph based on the original evaluation data, wherein the knowledge graph comprises concept entities, value association relations and education target attributes; A2, carrying out semantic enhancement processing on the knowledge graph, and carrying out multidimensional feature extraction based on text emotion, behavior pattern and cognition level on the original evaluation data to generate feature characterization data; A3, establishing an evaluation index association model according to the characteristic characterization data and the knowledge graph, carrying out educational effect dimension analysis based on the evaluation index association model, carrying out time-related influence weight evaluation, and generating effect evaluation dynamic data; A4, performing educational objective achievement degree association processing according to the effect evaluation dynamic data and the characteristic characterization data to generate objective achievement dynamic characteristic data, performing evaluation influence range recognition based on the objective achievement dynamic characteristic data to generate evaluation action range data, performing educational element mode coupling based on knowledge mastery, value acceptance and behavior trampling degree according to the evaluation action range data to generate educational action coupling data; And A5, constructing a machine learning evaluation model according to the educational action coupling data, intelligently evaluating the educational effect by using the machine learning evaluation model to generate evaluation result data, and optimizing the educational process in real time based on the evaluation result data to obtain educational improvement feedback data.
  2. 2. The educational intelligent evaluation method fusing knowledge patterns according to claim 1, wherein step A1 comprises the steps of: Step A11, carrying out data interface butt joint on an education course platform, a practice activity system and a teacher-student interaction terminal to obtain multi-source original data comprising course lecture texts, practice activity records and question-answer interaction logs; a12, carrying out data deduplication, format unification and missing value filling treatment on the multisource original data so as to obtain structured evaluation data; A13, entity extraction based on concept terms, value guide keywords and education target identifiers is carried out on the structured evaluation data, so that entity data are obtained; A14, carrying out relation mining based on semantic co-occurrence, logic deduction and target association on the entity data so as to obtain entity relation data, wherein the relation mining comprises mapping analysis of course content and value targets, corresponding analysis of practical activities and behavior requirements and association analysis of interactive feedback and cognition improvement; step A15, combining entity relation data, entity data and structural evaluation data into an education knowledge graph; And step A16, determining a key evaluation dimension according to the education knowledge graph, and establishing a multi-source data acquisition network comprising a text analysis module, a behavior tracking module and a cognitive evaluation module.
  3. 3. The educational intelligent evaluation method fusing knowledge patterns according to claim 2, wherein step a16 comprises the steps of: Step A161, carrying out hierarchical analysis on education content boundary characteristics according to education knowledge patterns to obtain content hierarchical point data, wherein the hierarchical analysis specifically identifies nodes with concept levels and significant change of value guide intensity; Step A162, performing sensitivity analysis on an educational objective implementation path according to an educational knowledge graph, and identifying links with large objective achievement effect change to obtain objective sensitive area data, wherein the sensitivity analysis comprises memory strength assessment based on knowledge mastery, emotion tendency assessment based on value acceptance and habit development assessment based on behavior trampling degree; Step A163, determining the coverage range of a key evaluation dimension according to preset education quality requirement data, content layering point data and target sensitive area data to obtain evaluation dimension arrangement data; And step A164, determining the type of the evaluation module according to the evaluation dimension arrangement data, and establishing a multi-source data acquisition network comprising a text emotion analysis module, a behavior pattern tracking module, a cognition level evaluation module and a target association verification module.
  4. 4. The educational intelligent evaluation method fusing knowledge patterns according to claim 3, wherein step A2 comprises the steps of: a21, carrying out enhancement processing based on semantic disambiguation and noise filtering on a multi-source data acquisition network, and extracting real-time features in the education process so as to obtain feature characterization data; A22, segmenting text emotion in the feature characterization data based on a time sequence, so as to establish an emotion-time relation curve; A23, calculating the frequency characteristic and the intensity distribution of emotion fluctuation according to the emotion-time relation curve, and analyzing the correlation between the emotion fluctuation and the education stage, so as to obtain emotion-stage evolution characteristic data; A24, establishing a relation curve between a behavior mode and an education stage according to the characteristic characterization data, and calculating the mode change rate of different stages so as to obtain mode space-time evolution characteristic data; A25, carrying out real-time change recording according to the cognitive level in the characteristic characterization data, and carrying out associated characteristic analysis of the cognitive level and the education content so as to obtain cognitive space-time evolution characteristic data; a26, carrying out comprehensive influence analysis on an evaluation result by the feature combination according to the emotion-stage evolution feature data, the mode space-time evolution feature data and the cognitive space-time evolution feature data, and identifying a critical value and an effective interval of the key feature combination so as to generate space-time evolution feature data of the feature combination; And A27, establishing a dynamic evolution model of the three-dimensional feature space according to the space-time evolution feature data, and extracting feature change features to obtain feature change feature data.
  5. 5. The educational intelligent evaluation method fusing knowledge patterns according to claim 4, wherein step a21 comprises the steps of: And carrying out enhancement processing based on semantic disambiguation and noise filtering on the multi-source data acquisition network, and extracting real-time characteristics in the education process to obtain characteristic characterization data, wherein the semantic disambiguation comprises context labeling of polysemous words, normalization processing of synonyms and explicit analysis of fuzzy expression, and the noise filtering comprises removing irrelevant chatter information, combining repeated feedback and correcting abnormal extreme data.
  6. 6. The educational intelligent evaluation method fusing knowledge patterns according to claim 5, wherein step a27 comprises the steps of: Step A271, constructing a three-dimensional feature space with text emotion, behavior pattern and cognition level as coordinate axes according to the space-time evolution feature data, so as to obtain feature space coordinate data; step A272, establishing a characteristic change track based on characteristic space coordinate data, and constructing a characteristic motion track curve through space mapping of time sequence sampling points so as to obtain characteristic track data; A273, carrying out layering processing based on education stages on the feature track data, and identifying feature change features comprising change trends, change rates and mutual correlations at different stages so as to obtain layering feature data; Step A274, constructing a dynamic model of the three-dimensional feature space according to the layered feature data, and establishing continuous expression of feature change through spatial interpolation and numerical fitting so as to obtain dynamic evolution model data; Step A275, extracting the characteristics of the gradient characteristics, the curvature characteristics and the speed characteristics based on the characteristic change from the dynamic evolution model data, thereby obtaining model characteristic data; Step A276, carrying out feature correlation analysis based on association relation and action mechanism between features based on the model feature data so as to obtain feature association data; And A277, extracting indexes representing the dynamic change rule of the feature according to the model feature data and the feature association data, thereby generating feature change feature data.
  7. 7. The educational intelligent evaluation method fusing knowledge patterns according to claim 6, wherein step A3 comprises the steps of: A31, establishing a characteristic-index correspondence table comprising mapping relations between characteristic changes and evaluation indexes in different education stages according to characteristic change characteristic data and education knowledge maps, so as to obtain index response data; a32, carrying out data standardization and dimension reduction processing on the index response data, and establishing a mathematical model of characteristic-index response so as to obtain response model data; A33, performing machine learning training based on the response model data, and constructing a nonlinear mapping relation between the characteristics and the evaluation indexes so as to obtain an evaluation index association model; Step A34, carrying out educational effect dimension analysis comprising knowledge transfer effect, value internalization degree and behavior conversion efficiency according to an evaluation index association model so as to obtain effect dimension data; A35, performing time-dependent influence weight calculation based on the effect dimension data so as to obtain influence weight data, wherein the influence weight calculation comprises contribution degree analysis of different education stages to a final effect and weight distribution of key time nodes; and step A36, identifying key characteristics and turning nodes in the educational effect evolution process of the effect evaluation dynamic data, thereby obtaining the effect evaluation dynamic data.
  8. 8. The educational intelligent evaluation method fusing knowledge patterns according to claim 7, wherein step A4 comprises the steps of: a41, analyzing the influence rule of different feature combinations on achievement degree of the educational objective according to the effect evaluation dynamic data and the feature characterization data, so as to obtain achievement influence data; Step A42, performing achievement performance evaluation based on knowledge memory persistence, value acceptance stability and behavior trampling persistence on achievement influence data so as to obtain performance evaluation data; A43, performing evolution characteristic analysis of educational objective achievement effect along with time on the performance evaluation data so as to obtain objective achievement dynamic characteristic data; a44, carrying out numerical simulation of an evaluation influence area according to target achievement dynamic characteristic data, and establishing a multi-field coupling analysis model comprising a knowledge transfer field, a value internalization field and a behavior conversion field so as to obtain influence area data; a45, carrying out boundary identification and space partitioning on the data of the influence area so as to obtain data of an evaluation action range; and step A46, performing educational element mode coupling based on knowledge mastery, value acceptance and behavior trampling degree according to the evaluation action range data, and identifying educational field evolution rules and key state characteristics to generate educational action coupling data.
  9. 9. The educational intelligent evaluation method fusing knowledge patterns according to claim 8, wherein step A5 comprises the steps of: a51, carrying out feature extraction on educational action coupling data, and constructing a training sample set of a machine learning model so as to obtain training sample data, wherein the training sample data comprises input features and evaluation labels; step A52, constructing a machine learning network structure based on a mixed architecture of an attention mechanism and a long-short-term memory network according to training sample data, and performing model training to obtain an effect evaluation model; Step A53, optimizing model prediction precision of the effect evaluation model based on a cross verification method, so as to obtain evaluation model data; a54, performing intelligent evaluation on the educational effect by using the evaluation model data so as to obtain evaluation result data, wherein the evaluation result data comprises knowledge mastering scores, value acceptance scores and behavior trampling scores; Step A55, carrying out key stage and effect turning feature analysis in the education effect formation process on the evaluation result data so as to obtain effect evolution data, wherein the key stage comprises a knowledge input stage, a value perception stage, an emotion recognition stage, a behavior try stage, a habit solidification stage and an effect stabilization stage; And step A56, carrying out real-time optimization on the educational process of each stage according to the effect evolution data, thereby obtaining educational improvement feedback data.
  10. 10. The educational intelligent evaluation method fusing knowledge patterns according to claim 9, wherein step a55 comprises the steps of: step A551, carrying out time sequence division of education phases on the evaluation result data to obtain phase division data; Step A552, based on the stage division data, carrying out in-stage mean value calculation and inter-stage difference analysis on knowledge mastering scores, value acceptance scores and behavior trampling scores so as to obtain score change characteristic data; Step A553, setting a threshold value of a key stage according to the grading change characteristic data, and identifying a stage with the grading increase rate exceeding a preset value as the key stage so as to obtain key stage data; Step A554, performing pattern matching of effect turning features on the key stage data, wherein the pattern matching comprises positioning of grading mutation points, directional judgment of grading trend and stability evaluation of grading fluctuation, so as to obtain effect turning feature data; And step A555, combining the key stage data and the effect turning feature data into effect evolution data.

Description

Educational intelligent evaluation method integrating knowledge patterns Technical Field The invention relates to the technical field of education, in particular to an education intelligent evaluation method integrating knowledge graphs. Background Currently, education is an important way to cultivate the value of students, and the effect evaluation of the education mainly depends on traditional means, such as questionnaire investigation, examination evaluation or subjective evaluation of teachers. The method has obvious limitations that firstly, the evaluation dimension is single, knowledge mastering, value recognition, behavior trampling and other educational objectives are difficult to fully cover, secondly, data acquisition fragmentation is carried out, systematic integration of multi-source data is lacking, evaluation results are on one side, and thirdly, the dynamic performance is insufficient, the variation rules of emotion, cognition and behavior of students in the education process cannot be captured in real time, and accurate educational intervention is difficult to support. With the development of information technology, the knowledge graph technology provides a new thought for education and evaluation. Knowledge maps can structurally represent concepts, relationships and targets in education, but their application in the field of evaluation is still in the exploration phase. In the prior art, the knowledge graph is mostly used for static knowledge representation and is not yet deeply fused with the dynamic evaluation requirement. For example, the conventional method cannot correlate educational content with student behavior data through a knowledge graph, and is difficult to quantify the value internalization process, and meanwhile, the evaluation model lacks consideration of time dimension weight, so that the analysis of the staged educational effect is insufficient. In addition, the existing evaluation system has limited capability of extracting features such as emotion fluctuation and behavior pattern evolution, and the like, so that the fine evaluation of educational effects is difficult to realize. Aiming at the problems, an intelligent evaluation method capable of integrating a knowledge graph and multi-source dynamic data is needed to solve the problems of single dimension, data rupture and insufficient dynamics in education evaluation. Through constructing the education knowledge graph and combining with feature extraction of text emotion, behavior pattern and cognition level, a dynamic association model is established, and finally, the precise evaluation and real-time optimization of the education effect are realized. Disclosure of Invention The invention aims to provide an educational intelligent evaluation method integrating knowledge maps, which aims to solve the problems in the background technology. In order to achieve the purpose, the invention provides the following technical scheme that the education intelligent evaluation method integrating the knowledge graph comprises the following steps: The method comprises the steps of A1, carrying out multi-source data acquisition on an education scene to obtain original evaluation data comprising course texts, activity records and interaction logs, and constructing an education knowledge graph based on the original evaluation data, wherein the knowledge graph comprises concept entities, value association relations and education target attributes; A2, carrying out semantic enhancement processing on the knowledge graph, and carrying out multidimensional feature extraction based on text emotion, behavior pattern and cognition level on the original evaluation data to generate feature characterization data; A3, establishing an evaluation index association model according to the characteristic characterization data and the knowledge graph, carrying out educational effect dimension analysis based on the evaluation index association model, carrying out time-related influence weight evaluation, and generating effect evaluation dynamic data; A4, performing educational objective achievement degree association processing according to the effect evaluation dynamic data and the characteristic characterization data to generate objective achievement dynamic characteristic data, performing evaluation influence range recognition based on the objective achievement dynamic characteristic data to generate evaluation action range data, performing educational element mode coupling based on knowledge mastery, value acceptance and behavior trampling degree according to the evaluation action range data to generate educational action coupling data; And A5, constructing a machine learning evaluation model according to the educational action coupling data, intelligently evaluating the educational effect by using the machine learning evaluation model to generate evaluation result data, and optimizing the educational process in real time based on the evaluation result