Search

CN-121980419-A - Program course development effect AI real-time evaluation method based on psychology fluctuation data

CN121980419ACN 121980419 ACN121980419 ACN 121980419ACN-121980419-A

Abstract

The invention discloses a program course development effect AI real-time evaluation method based on psychology fluctuation data, which comprises the steps of acquiring and synchronizing psychology data and programming behavior data of a learner in real time, quantifying cognitive load time sequence characteristics from the psychology data, quantifying dynamic complexity time sequence characteristics of codes by introducing a tree editing distance algorithm weighted by cognitive cost based on abstract syntax trees generated by code snapshots, inputting aligned cognitive load and code complexity characteristic sequences into a circulating neural network model with a built-in attention mechanism, and outputting learning state classification results such as smooth learning, effective challenges or learning faults in real time by the model. When a "learning fault" is detected, the invention can also correlate specific course knowledge points and trigger personalized teaching intervention. The invention realizes real-time accurate diagnosis and self-adaptive intervention of learning difficulty and provides a data-driven solution for iterative optimization of program courses.

Inventors

  • SHU SHENG

Assignees

  • 安徽商贸职业技术学院

Dates

Publication Date
20260505
Application Date
20260126

Claims (10)

  1. 1. The program course development effect AI real-time evaluation method based on the psychological fluctuation data is characterized by comprising the following steps: Acquiring multi-modal data of a learner in a programming process in real time, wherein the multi-modal data comprises psychophysiological data and programming behavior data, and performing time sequence synchronization processing on the multi-modal data; Based on the psychological and physiological data after the time sequence synchronization processing, extracting and quantizing to obtain cognitive load time sequence characteristics, and meanwhile, based on code snapshots in the programming behavior data, generating an abstract syntax tree, extracting and quantizing to obtain code complexity time sequence characteristics; The cognitive load time sequence characteristics and the code complexity time sequence characteristics are aligned in time sequence and input into a preset artificial intelligent model; and processing the input time sequence characteristics by the artificial intelligent model, outputting a classification result representing the learning state of the current learner, and taking the classification result as a real-time evaluation result of the program course development effect.
  2. 2. The program course development effect AI real-time evaluation method based on psychologic fluctuation data according to claim 1, wherein the time series synchronization processing is performed on the psychologic data, comprising: and performing on-line artifact removal processing on the electroencephalogram data in the psychophysiological data by adopting an artifact subspace reconstruction algorithm or an on-line recursion independent component analysis algorithm.
  3. 3. The program course development effect AI real-time evaluation method based on the psychology fluctuation data as claimed in claim 1, wherein the step of extracting and quantifying the obtained cognitive load time series characteristics comprises: calculating the power spectral density of a specific brain region under a preset frequency band based on the brain electrical data in the psychophysiological data, and carrying out combined operation on the power of at least two frequency bands according to a preset algorithm to obtain brain electrical cognitive load indexes; extracting an eye movement cognitive load index comprising pupil diameter change or glance path entropy based on the eye movement data in the psychophysiological data; and carrying out multi-mode data fusion on the brain electricity cognitive load index and the eye movement cognitive load index to generate the cognitive load time sequence characteristic.
  4. 4. The method for real-time evaluation of program course development effect AI based on psychologic fluctuation data according to claim 1, wherein the extracted and quantized code complexity time sequence features comprise static complexity features and dynamic complexity features, and the dynamic complexity features are used for representing the structural variation between abstract syntax trees corresponding to two consecutive code snapshots.
  5. 5. The program course development effect AI real-time evaluation method based on the psychoacoustic fluctuation data according to claim 4, wherein the quantification of the dynamic complexity characteristic comprises: presetting cognitive cost weights for different node types and different editing operations in the abstract syntax tree; and calculating the minimum accumulated cognitive cost required by transforming the abstract syntax tree at the previous moment into the abstract syntax tree at the current moment by adopting a tree editing distance algorithm, and taking the cognitive cost as a quantized value of the dynamic complexity characteristic.
  6. 6. The program course development effect AI real-time evaluation method based on the psychowave data according to claim 1, further comprising, before inputting the time series characteristics to a preset artificial intelligence model: Calculating the time change rate of the cognitive load time sequence characteristic in a preset time window; Calculating the time change rate of the code complexity time sequence characteristic; And calculating the time change rate of the cognitive load time sequence characteristic and the time change rate of the code complexity time sequence characteristic to generate a cognitive overload sensitivity index, and taking the index as one of the input characteristics of the artificial intelligent model.
  7. 7. The program course development effect AI real-time evaluation method based on psychowave data according to claim 1 or 6, wherein the preset artificial intelligence model is a cyclic neural network model including a long-short time memory network or a variant thereof.
  8. 8. The method for real-time evaluation of program course development effect AI based on psychowave data as recited in claim 7, wherein the recurrent neural network model comprises an attention mechanism layer for assigning different attention weights to different time steps in a time series feature sequence input to the model to identify a historical programming event having a greatest influence on a current learning state classification result.
  9. 9. The program course development effect AI real-time evaluation method based on the psychographic fluctuation data as recited in claim 1, wherein the classification result is output including one of a fluent learning state, an effective challenge state, and a learning fault state.
  10. 10. The program course development effect AI real-time evaluation method based on the psychographic fluctuation data as recited in claim 9, further comprising: When the classification result judges that the learner is in a learning fault state, identifying a course knowledge point label associated with the state occurrence time point, and triggering a preset personalized teaching intervention instruction based on the knowledge point label.

Description

Program course development effect AI real-time evaluation method based on psychology fluctuation data Technical Field The invention relates to the technical field of education of artificial intelligence, in particular to a program course development effect AI real-time evaluation method based on psychology fluctuation data. Background With the deep fusion of information technology and education fields, learning analysis (LEARNING ANALYTICS) technology has become a key driving force for improving teaching quality and realizing personalized education. In the field of programming education, researchers have turned to mining and analyzing learning process data from traditional assessment of finalization of dependent task scoring, examination performance, and the like. Early procedural analysis was primarily dependent on the log of the Learning Management System (LMS) and behavioral events in the Integrated Development Environment (IDE), such as code submission frequency, number of compilation errors, debug duration, etc., by which the learning behavior patterns of students were initially characterized. In recent years, with the development of physiological computing and emotion computing technologies, researchers begin to attempt to introduce physiological signals such as electroencephalogram (EEG), eye-tracking, and skin electricity to objectively and real-time quantify cognitive load, attention state and emotion fluctuation of a learner in a programming task, and provide a new technical path for revealing an intrinsic cognitive mechanism of programming learning. At the same time, at the code analysis level, the structured analysis of Abstract Syntax Trees (AST) is also used to evaluate the intrinsic complexity of the code, beyond the shallow metrics of traditional code Lines (LOC) and the like. Despite the advances made in the art, existing methods still suffer from significant shortcomings in achieving accurate, real-time assessment of program course development efforts. Firstly, the existing evaluation paradigm commonly has "evaluation hysteresis" and "attribution ambiguity", even if physiological data is introduced, most researches remain in macroscopic state analysis of the whole programming task period, and the cognitive difficulty of a learner cannot be precisely positioned to a specific knowledge point of course content or a certain key code logic construction link, so that the guidance value of an evaluation result on course design is limited. Secondly, on the data analysis level, the correlation analysis of programming behaviors and physiological responses is rough and shallow, most of the correlation statistics are simple, and a dynamic causal correlation model between code structure evolution and cognitive load fluctuation cannot be established. In particular, the quantification of programming difficulty in the prior art is mainly remained in the static complexity evaluation of the code product, and the programming process is seriously neglected, namely, when the programming process evolves from a simple code structure to a complex structure, a learner needs to cross the dynamic cognition cost, which is exactly the core of the generation of the learning fault. Therefore, there is an urgent need in the art for a method that can align and model the dynamic structural evolution of codes with the real-time psychological fluctuation of learners in depth and fine granularity, so as to detect learning faults in real time and objectively, and thus provide accurate data support for iterative optimization of courses. Disclosure of Invention This section is intended to outline some aspects of embodiments of the application and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section as well as in the description of the application and in the title of the application, which may not be used to limit the scope of the application. The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a program course development effect AI real-time evaluation method based on psychology fluctuation data, which is used for solving the problems in the background technology. In order to solve the technical problems, the invention provides a program course development effect AI real-time evaluation method based on mental fluctuation data, which comprises the following steps: Acquiring multi-modal data of a learner in a programming process in real time, wherein the multi-modal data comprises psychophysiological data and programming behavior data, and performing time sequence synchronization processing on the multi-modal data; Based on the psychological and physiological data after the time sequence synchronization processing, extracting and quantizing to obtain cognitive load time sequence characteristics, and meanwhile, based on code snapshots in the programming behavior data, generating an abstract