Research On Automatic Evaluation Technology Of Teaching Cases Based On Bloom’s Taxonomy

Curriculum evaluation is a key issue in teaching reform.It is not easy to directly evaluate the course,but it is feasible to divide the course into teaching cases,evaluate each teaching case,and finally summarize and analyze it.This article uses the Bloom classification method recommended in the CC2022 report to evaluate courses,introduces machine learning,compares expert annotation with machine automatic classification,discusses and improves teaching cases,and provides a new approach for the evaluation of computational subject courses.The main tasks are as follows:(1)In response to the current lack of high-quality teaching case datasets for computing courses,this article extracted 405 teaching cases from 26 computing courses at Guilin University of Electronic Science and Technology,and constructed a high-quality small sample case classification dataset based on Bloom classification method.This dataset is used for training and testing machine automatic classification,comparing the results of manual annotation and machine classification,improving the teaching objectives of cases,and providing assistance for the high-quality development of courses.(2)A case set automatic evaluation model based on Wo BERT and Text CNN is designed for the automatic evaluation of teaching cases in the field of computing.The case is annotated and judged by the semantic similarity of word vector mapping,and the text is represented by an improved Wo BERT encoder.Then,Text CNN is used as a text classifier to extract semantic features of the text,and the features are integrated and filtered,The final implementation of the case is automatic classification using Bloom classification method.The experimental results show that the proposed Wo BERT-CNN model has increased the f1 value in the cognitive process dimension by 8 percentage points and 2percentage points,respectively,compared to the CNN benchmark model and the BERT benchmark model;The f1 value in the knowledge dimension increased by 11.4 percentage points and 2.4 percentage points respectively.Verified the effectiveness of the fusion method of pre trained language model and convolutional neural network in assisting case judgment.(3)Aiming at the automatic classification of computing discipline case data sets in the small sample scenario,an automatic classification model ERNIE-CNN based on transfer learning is designed.The model uses the ERNIE model as the learner and CNN as the feature extractor.After fine-tuning the transfer learning model,it adapts to the automatic classification task for Canterbury Question Bank data sets,Improved the accuracy of the model in automatically classifying cognitive process dimensions for Canterbury Question Bank small sample datasets.The accuracy of ERNIE-CNN is 2.4% higher than the BERT benchmark model and 0.7% higher than the ERNIE benchmark model,verifying the feasibility of ERNIE-CNN in determining small sample datasets.