Search

CN-121981618-A - Teaching quality assessment method and system based on big data

CN121981618ACN 121981618 ACN121981618 ACN 121981618ACN-121981618-A

Abstract

The invention relates to the technical field of teaching evaluation, in particular to a teaching quality evaluation method and a teaching quality evaluation system based on big data, comprising the following steps of obtaining a teaching task performance point sequence and a target performance interval, constructing a multidimensional performance reference vector, calculating difference and generating a target offset measurement parameter, constructing an embedded graph and generating behavior transfer sparsity distribution, extracting a behavior block to generate a density projection curve and intercepting rupture points, constructing an evaluation vector and judging the grade. According to the invention, the multi-dimensional reference vector is constructed by extracting continuous teaching period data and combining with the teaching unit target performance interval, the teaching stability breaking point is extracted, a plurality of key features are mapped to the high-dimensional space construction evaluation vector and are compared with the standard state in a deviation manner, so that the multi-dimensional dynamic modeling and stability analysis of the teaching quality are realized, and the systematicness, the dynamic property and the objectivity of the teaching quality evaluation under a complex teaching scene are effectively improved.

Inventors

  • LI JING
  • HUANG GUOLI
  • ZHANG KUN

Assignees

  • 山东工程职业技术大学

Dates

Publication Date
20260505
Application Date
20260408

Claims (10)

  1. 1. The teaching quality assessment method based on big data is characterized by comprising the following steps: S1, acquiring an original performance point value sequence of a teaching task and a target performance interval of a teaching unit, extracting continuous teaching period data from the original performance point value sequence of the teaching task, and constructing a same-period multidimensional performance reference vector by combining the target performance interval of the teaching unit; s2, differentiating the teaching task original performance point value sequence and the same-period multidimensional performance reference vector, calculating the discrete fluctuation range in the period by using a dynamic time warping algorithm, and generating a target offset measurement parameter by calculating the ratio of the discrete fluctuation range in the period to a teaching unit target performance interval; S3, constructing a teaching activity embedded graph structure based on an original performance point value sequence of the teaching task, calculating co-occurrence probability difference values of adjacent nodes, accumulating the co-occurrence probability difference values exceeding a behavior continuity rupture threshold value, and generating behavior transfer sparsity distribution information; S4, extracting a continuous behavior block sequence from the teaching activity embedded graph structure, calculating Euclidean distance, generating a behavior concentration projection curve, utilizing noise based on density to apply a spatial clustering algorithm, and intercepting stability breaking points according to the behavior concentration projection curve; And S5, constructing a teaching quality evaluation vector based on the target offset measurement parameter, the behavior transfer sparsity distribution information and the stability breaking points, and calculating a deviation value of the teaching quality evaluation vector and a standard steady state vector to determine a teaching quality grade.
  2. 2. The big data based teaching quality assessment method according to claim 1, wherein the co-periodic multidimensional performance reference vector comprises a performance benchmark lower bound component, a target achievement expectancy mean value and a periodic fluctuation tolerance threshold, the target offset metric parameter comprises a positive-negative direction deviation polarity identification, a relative fluctuation amplitude ratio and a cumulative drift trend index, the behavior transition sparsity distribution information comprises inter-node transition probability density, path fracture cumulative weight and sparse interval span distribution, the stability breaking point comprises a mutation occurrence timestamp index, a concentration gradient jump amplitude and a behavior pattern transition type label, and the teaching quality grade comprises a comprehensive stability score, a teaching target achievement category and an abnormal risk early warning level.
  3. 3. The teaching quality assessment method based on big data according to claim 2, wherein the step of obtaining the co-periodic multidimensional performance reference vector specifically comprises: S111, invoking a teaching task original performance point value sequence comprising a history scoring record and a preset teaching unit target performance interval through a teaching management database, analyzing the time stamp attribute carried by each discrete data item in the teaching task original performance point value sequence one by one, executing ascending arrangement of the time dimension on the teaching task original performance point value sequence according to the numerical value of the time stamp attribute, and establishing a corresponding data index mapping relation between the teaching task original performance point value sequence and the teaching unit target performance interval to generate a teaching task basic data combination with the time stamp attribute; S112, calculating the time interval value between two adjacent time stamps in the sequence based on the teaching task basic data combination with the time stamp attribute, comparing the time interval value with a preset teaching period continuity judging threshold value, judging the corresponding data node as a coherent node if the time interval value is smaller than or equal to the teaching period continuity judging threshold value, identifying and extracting a time sequence segment formed by the coherent node, and removing isolated time point data which does not meet the continuity requirement to obtain a continuous teaching period performance data subsequence; S113, calling a teaching unit target performance interval in the continuous teaching period performance data subsequence and the teaching task basic data combination with the timestamp attribute, setting a numerical lower bound and a numerical upper bound of the teaching unit target performance interval as a reference axis of a multi-dimensional reference coordinate system, projecting each performance point value in the continuous teaching period performance data subsequence into the multi-dimensional reference coordinate system, calculating Euclidean space position coordinates and vertical distance parameters of each performance point value relative to the reference axis, vectorizing and splicing the position coordinates and the vertical distance parameters according to the sequence of time dimension, and constructing the same period multi-dimensional performance reference vector.
  4. 4. The teaching quality assessment method based on big data according to claim 3, wherein the step of obtaining the target offset measurement parameter specifically comprises: S211, performing point-by-point difference operation based on a time index aiming at the teaching task original performance point value sequence and the same-period multidimensional performance reference vector, calculating the numerical deviation of the teaching task original performance point value sequence and the same-period multidimensional performance reference vector at the same time node, and generating an instantaneous performance deviation sequence; s212, constructing a minimum accumulated distance path between an original performance point value sequence of the teaching task and the same-period multidimensional performance reference vector based on the instantaneous performance deviation sequence, calculating the overall fluctuation intensity of the sequence by integrating the path regular cost and the local deviation characteristic, and calculating and obtaining the discrete fluctuation range in the period; S213, analyzing upper and lower boundary values of the teaching unit target performance interval, calculating interval span amplitude, and calculating the ratio of the discrete fluctuation amplitude in the period to the interval span amplitude to obtain the target offset measurement parameter.
  5. 5. The teaching quality evaluation method based on big data according to claim 4, wherein the formula of the discrete variation amplitude in the operation acquisition period is specifically: ; Wherein, the Representing the amplitude of the discrete variation within the cycle, The sequence length representing the instantaneous performance bias sequence, A normalized instantaneous performance bias value representing a kth time point in the instantaneous performance bias sequence, Representing a decreasing weight factor over time for the kth point in time, Representing the normalized dynamic programming minimum accumulated path cost between the teaching task original performance point value sequence and the same-period multidimensional performance reference vector, Normalized teaching task performance values representing a kth time point in the sequence of teaching task raw performance point values, A normalized co-periodic reference value representing a kth point in time in the co-periodic multi-dimensional performance reference vector, Representing the normalized adjustment coefficient for the path cost.
  6. 6. The teaching quality assessment method based on big data according to claim 5, wherein the step of obtaining the behavior transfer sparsity distribution information specifically comprises: S311, analyzing teaching activity type identifiers corresponding to each performance point value based on an original performance point value sequence of the teaching task, instantiating the teaching activity type identifiers as independent nodes in a topological network, establishing directed connection edges according to the front-back adjacent sequence of the teaching activity type identifiers in a time dimension, and constructing a teaching activity embedded graph structure by taking the inverse of the time interval between the nodes as an edge weight parameter; s312, traversing all precursor node and subsequent node combinations connected through directional connection edges aiming at the teaching activity embedded graph structure, counting the joint occurrence frequency of each group of node combinations in the whole teaching period, converting the joint occurrence frequency into a conditional transition probability value, calculating the absolute value of the difference value of the conditional transition probability value between the nodes at the two ends of each directional connection edge, and generating a neighboring node co-occurrence probability difference matrix; S313, calling the co-occurrence probability difference matrix of the adjacent nodes to perform numerical comparison with a preset behavior continuity rupture threshold, screening abnormal difference items with the numerical value exceeding the behavior continuity rupture threshold in the matrix, and performing weighted accumulation summation on the abnormal difference items according to the execution sequence of teaching activities to construct behavior transfer sparsity distribution information.
  7. 7. The big data-based teaching quality assessment method according to claim 6, wherein the setting mode of the behavior coherence rupture threshold is specifically that a batch sample set of a historical teaching activity sequence is obtained, co-occurrence probability differences among all adjacent teaching activity node pairs in the sample set are calculated, a statistical distribution histogram of the co-occurrence probability differences is constructed, a mean value and a standard deviation of the difference distribution are calculated based on the histogram, and the standard deviation of the mean value plus a preset multiple is taken as the behavior coherence rupture threshold; The process of weighting, accumulating and summing the abnormal difference items according to the execution sequence of the teaching activity is specifically to assign a descending weight factor according to the reverse sequence arrangement of time position indexes by identifying the relative position indexes of the abnormal difference items in the execution sequence of the teaching activity, calculate the position weight coefficient of each abnormal difference item, multiply the numerical value of each abnormal difference item with the corresponding position weight coefficient, and then execute accumulated and summed on the product result.
  8. 8. The teaching quality assessment method based on big data according to claim 7, wherein the step of obtaining the stability breaking point specifically comprises: S411, extracting a continuous behavior block sequence formed by connecting nodes according to time sequence based on strong communication components in the structure traversal graph of the teaching activity embedded graph, extracting high-order attribute vectors representing topological structure features for each pair of adjacent behavior blocks in the sequence, calculating Euclidean distance values of the two high-order attribute vectors in a feature space, connecting the distance values according to time sequence, and performing smoothing treatment to generate a behavior concentration projection curve; s412, mapping the behavior concentration projection curve to a two-dimensional density clustering space, defining a neighborhood scanning radius and minimum inclusion points for each projection point in the space, dividing the projection points into core objects, boundary points or noise points according to neighborhood density attributes of the points, calculating density reduction rate when the projection points are transited from a high-density core area to a low-density noise area, and establishing an area concentration mutation gradient; S413, invoking the region concentration mutation gradient to carry out numerical comparison with a preset stability judging threshold, screening an abnormal fluctuation index with the gradient amplitude exceeding the stability judging threshold, mapping the abnormal fluctuation index back to an original time axis coordinate, and locking the specific moment of occurrence of quality change of a behavior mode in the teaching process to obtain a stability breaking point.
  9. 9. The method for evaluating teaching quality based on big data according to claim 8, wherein the step of obtaining the teaching quality level specifically comprises: S511, calling the target offset measurement parameters, the behavior transfer sparsity distribution information and the stability breaking points, performing standardization processing of feature dimensions on heterogeneous data to eliminate dimension differences, mapping the heterogeneous data to a preset high-dimensional feature space coordinate system as independent component coordinates, and performing vectorization splicing on the component coordinates according to the arrangement sequence of feature weights to construct a teaching quality assessment vector; s512, acquiring a preset standard steady state vector as a reference standard, calculating Euclidean distance between the teaching quality evaluation vector and the standard steady state vector in a high-dimensional feature space, and representing the deviation degree of the current teaching state relative to an ideal steady state by carrying out quantitative calculation on the modular length of a differential vector to generate a state vector space deviation value; S513, comparing the state vector space deviation value with a preset multi-level quality assessment threshold step by step, identifying a specific numerical value interval in which the state vector space deviation value falls, and retrieving a corresponding assessment label according to a mapping relation between a preset interval index and a level definition table to obtain the teaching quality level.
  10. 10. A big data based teaching quality assessment system for implementing the big data based teaching quality assessment method of any of claims 1-9, the system comprising: The multi-dimensional performance arrangement module acquires an original performance point value sequence of the teaching task and a teaching unit target performance interval, extracts continuous teaching period data from the original performance point value sequence of the teaching task and combines the teaching unit target performance interval to construct a same-period multi-dimensional performance reference vector; The target offset analysis module is used for differentiating the teaching task original performance point value sequence and the same-period multidimensional performance reference vector, calculating the discrete fluctuation range in the period by using a dynamic time warping algorithm, and generating a target offset measurement parameter by calculating the ratio of the discrete fluctuation range in the period to the teaching unit target performance interval; The behavior transfer recognition module is used for constructing a teaching activity embedded graph structure based on an original performance point value sequence of the teaching task, calculating co-occurrence probability difference values of adjacent nodes, accumulating the co-occurrence probability difference values exceeding a behavior continuity rupture threshold value, and generating behavior transfer sparsity distribution information; The stability analysis module is used for extracting a continuous behavior block sequence from the teaching activity embedded graph structure, calculating Euclidean distance, generating a behavior concentration projection curve, utilizing noise based on density to apply a spatial clustering algorithm, and intercepting stability breaking points according to the behavior concentration projection curve; And the teaching quality dividing module is used for constructing a teaching quality assessment vector based on the target offset measurement parameter, the behavior transfer sparsity distribution information and the stability breaking points, calculating the deviation value of the teaching quality assessment vector and the standard steady state vector and determining the teaching quality grade.

Description

Teaching quality assessment method and system based on big data Technical Field The invention relates to the technical field of teaching evaluation, in particular to a teaching quality evaluation method and system based on big data. Background The technical field of teaching evaluation relates to evaluation of core matters such as teaching activity effect, teaching process quality, teacher teaching behavior, student learning performance and the like, and aims to realize objective analysis and scientific judgment of teaching quality through quantitative and qualitative means. The technical field generally covers the aspects of evaluation index system construction, teaching data acquisition, statistical analysis, teaching result feedback mechanism and the like. The traditional teaching evaluation mode is based on manual scoring, questionnaire investigation, classroom observation and the like, and has the problems of single data source, limited evaluation dimension, strong subjectivity and the like, so that the actual teaching effect is difficult to comprehensively reflect. Along with the development of informatization education, big data technology is gradually introduced into a teaching evaluation process, and data support is provided for an evaluation model by carrying out centralized processing on multi-source heterogeneous teaching data, so that quantification, fineness and intellectualization of an evaluation means are realized. The traditional teaching quality assessment method is a method for judging teaching quality by adopting modes of statistics summarization, average value comparison and the like based on indexes such as student examination results, teacher teaching conditions, course attendance records and class participation. Such methods typically include manually collecting paper or electronic scores, sorting curriculum schedules, observing student classroom performance, issuing and retrieving paper or online questionnaires, and weighting each index according to a preset formula to form an evaluation conclusion. The method depends on limited sample data, the evaluation flow is mostly linear and static processing, and the method lacks analysis capability for large-scale, multidimensional and dynamic data and is not suitable for systematic evaluation of teaching quality in complex teaching scenes. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a teaching quality assessment method and system based on big data. In order to achieve the above purpose, the invention adopts the following technical scheme that the teaching quality assessment method based on big data comprises the following steps: S1, acquiring an original performance point value sequence of a teaching task and a target performance interval of a teaching unit, extracting continuous teaching period data from the original performance point value sequence of the teaching task, and constructing a same-period multidimensional performance reference vector by combining the target performance interval of the teaching unit; s2, differentiating the teaching task original performance point value sequence and the same-period multidimensional performance reference vector, calculating the discrete fluctuation range in the period by using a dynamic time warping algorithm, and generating a target offset measurement parameter by calculating the ratio of the discrete fluctuation range in the period to a teaching unit target performance interval; S3, constructing a teaching activity embedded graph structure based on an original performance point value sequence of the teaching task, calculating co-occurrence probability difference values of adjacent nodes, accumulating the co-occurrence probability difference values exceeding a behavior continuity rupture threshold value, and generating behavior transfer sparsity distribution information; S4, extracting a continuous behavior block sequence from the teaching activity embedded graph structure, calculating Euclidean distance, generating a behavior concentration projection curve, utilizing noise based on density to apply a spatial clustering algorithm, and intercepting stability breaking points according to the behavior concentration projection curve; And S5, constructing a teaching quality evaluation vector based on the target offset measurement parameter, the behavior transfer sparsity distribution information and the stability breaking points, and calculating a deviation value of the teaching quality evaluation vector and a standard steady state vector to determine a teaching quality grade. The invention improves that the same-period multidimensional performance reference vector comprises a performance benchmark lower bound component, a target achievement expected mean value and a period fluctuation tolerance threshold, the target deviation measurement parameter comprises a positive and negative deviation polarity mark, a relative fluctuation amplitude ratio an